WEBVTT
00:00:02.982 --> 00:00:03.815
Good morning everyone.
00:00:03.815 --> 00:00:06.770
And thank you for joining
today's public meeting.
00:00:06.770 --> 00:00:08.440
I'm Tony Noah, the program manager
00:00:08.440 --> 00:00:10.630
for the Wildfire Safety Enforcement Branch
00:00:10.630 --> 00:00:12.520
within the Safety Enforcement Division.
00:00:12.520 --> 00:00:14.860
And I'll be moderating today's meeting.
00:00:14.860 --> 00:00:16.660
A quick few administrative notes
00:00:16.660 --> 00:00:17.493
before we get started
00:00:17.493 --> 00:00:20.159
this meeting is livestream
on the CPC website.
00:00:20.159 --> 00:00:25.159
You can view the meeting a
www.adminmonitored.com/CA/CPUC
00:00:29.980 --> 00:00:31.290
closed captioning is available
00:00:31.290 --> 00:00:33.620
in English and Spanish through the webcast.
00:00:33.620 --> 00:00:35.590
You can click on the green button
00:00:35.590 --> 00:00:37.800
to select your language of choice.
00:00:37.800 --> 00:00:39.660
If you wish to make public comment,
00:00:39.660 --> 00:00:43.340
please dial into +1 800-857-1917
00:00:47.840 --> 00:00:50.870
and enter pass code 5180519 and press *1
00:00:57.210 --> 00:01:01.313
again that is, (800) 857-1917.
00:01:03.950 --> 00:01:08.950
with a passcode of 5180519 and press *1
00:01:11.270 --> 00:01:12.607
you'll be placed into a queue
00:01:12.607 --> 00:01:14.627
and the operator will take
your name and information
00:01:14.627 --> 00:01:16.410
and you will be called upon to speak.
00:01:16.410 --> 00:01:17.780
When we get to the public comment
00:01:17.780 --> 00:01:19.053
period in today's agenda.
00:01:19.980 --> 00:01:22.143
The purpose of today's meeting is
00:01:22.143 --> 00:01:25.104
present the results of the study of 2019
00:01:25.104 --> 00:01:28.573
PSPS events using new
wildfire modeling capabilities.
00:01:29.410 --> 00:01:31.163
Next slide, please for background.
00:01:38.580 --> 00:01:43.580
In late 2019 the CPUC became
aware of new fire modeling
00:01:43.970 --> 00:01:48.080
software by the company,
Technosilva, the CPUC desired
00:01:48.080 --> 00:01:50.720
to explore the new
modeling capabilities available
00:01:50.720 --> 00:01:52.203
in late 2019.
00:01:53.420 --> 00:01:55.000
Due to the timing of when we use the
00:01:55.000 --> 00:01:57.270
we determined the software realized
00:01:57.270 --> 00:01:58.470
that the software was out there.
00:01:58.470 --> 00:02:02.210
We used the October, 2019 PSPS events
00:02:02.210 --> 00:02:05.560
as the subject of the
modeling capability study.
00:02:05.560 --> 00:02:07.780
The objective was to allow the CPUC to better
00:02:07.780 --> 00:02:09.810
understand new modeling capabilities
00:02:09.810 --> 00:02:14.810
that portrays scenarios of
PSPS events and wildfire risk.
00:02:15.010 --> 00:02:17.270
The primary portion of today's meeting
00:02:17.270 --> 00:02:20.150
will be Technosilva and Mr. David Buckley
00:02:20.150 --> 00:02:24.920
presenting the results of
the modeling capabilities tests
00:02:24.920 --> 00:02:27.340
that they performed for the CPUC.
00:02:27.340 --> 00:02:29.870
So at this time I will
hand it over to Mr. Buckley
00:02:29.870 --> 00:02:32.810
to introduce his team
and do his presentation.
00:02:32.810 --> 00:02:33.810
Thank you very much.
00:02:39.098 --> 00:02:39.931
Good morning, everybody.
00:02:39.931 --> 00:02:40.863
David Buckley here.
00:02:42.020 --> 00:02:43.970
I'll just go ahead and share my screen.
00:02:48.120 --> 00:02:50.663
Tony, can you confirm the title slide?
00:02:52.460 --> 00:02:53.293
I can see it.
00:02:54.420 --> 00:02:55.253
How's my audio.
00:02:55.253 --> 00:02:56.086
Is that good?
00:02:58.320 --> 00:02:59.420
It's good on my end.
00:03:00.800 --> 00:03:01.633
All right.
00:03:01.633 --> 00:03:02.780
Excellent.
00:03:02.780 --> 00:03:04.450
Well welcome everybody.
00:03:04.450 --> 00:03:06.900
Thanks for the opportunity to present today.
00:03:06.900 --> 00:03:08.511
We're excited to show you the results
00:03:08.511 --> 00:03:13.511
of the 2019 PSPS event,
Wildfire Risk analysis
00:03:16.640 --> 00:03:17.840
Role introduction.
00:03:17.840 --> 00:03:20.165
The principal investigators were myself
00:03:20.165 --> 00:03:23.890
and my partner, Dr. Joaquin
Ramirez, who is our president
00:03:23.890 --> 00:03:25.673
and CTO and the visionary of
00:03:25.673 --> 00:03:27.883
of the fire modeling software
00:03:27.883 --> 00:03:30.370
that was used in this analysis.
00:03:30.370 --> 00:03:32.860
I'd like to also acknowledge
other participants
00:03:32.860 --> 00:03:35.770
that were involved first
and foremost, the CPUC
00:03:35.770 --> 00:03:39.210
for having the insight
to go out and have a look
00:03:39.210 --> 00:03:42.260
at integrating scientific modeling to really
00:03:42.260 --> 00:03:45.280
help evaluate some of
the possible consequences
00:03:45.280 --> 00:03:48.290
that could occur during these PSPS events.
00:03:48.290 --> 00:03:51.270
This is a novel and new approach that has not
00:03:51.270 --> 00:03:55.000
been done before by the
CPUC or government agencies.
00:03:55.000 --> 00:03:56.830
And we're excited to do that.
00:03:56.830 --> 00:03:58.480
I'd also like to acknowledge
00:04:00.721 --> 00:04:02.500
the California Department of Forestry
00:04:02.500 --> 00:04:04.760
and Fire Protection, CAL Fire.
00:04:04.760 --> 00:04:06.230
They have been significant and working
00:04:06.230 --> 00:04:08.050
with us over the last two years
00:04:08.050 --> 00:04:10.140
and really continuing that collaboration
00:04:11.160 --> 00:04:14.560
and integration with us to
really help refine the model
00:04:14.560 --> 00:04:17.320
and to really prove it in the real world
00:04:17.320 --> 00:04:22.320
during particularly the
horrendous 2020 fire season.
00:04:22.580 --> 00:04:24.310
Of course, I'd like to also acknowledge some
00:04:24.310 --> 00:04:25.760
of the researchers we work with
00:04:25.760 --> 00:04:27.610
some of the chief scientists on our team.
00:04:27.610 --> 00:04:29.320
There's a team of about seven people
00:04:29.320 --> 00:04:31.450
who've worked on this within Technosilva,
00:04:31.450 --> 00:04:34.070
but we also reached out
to San Jose state university
00:04:34.070 --> 00:04:36.019
and their fire weather program,
00:04:36.019 --> 00:04:37.663
particularly with Dr. Craig (indistinct)
00:04:37.663 --> 00:04:40.450
and Scott (indistinct) to support us
00:04:40.450 --> 00:04:41.970
with advanced weather analysis,
00:04:41.970 --> 00:04:44.096
to make sure that we got
all the weather data right
00:04:44.096 --> 00:04:46.040
on the front end of this.
00:04:46.040 --> 00:04:48.270
And also to some of those other researchers
00:04:48.270 --> 00:04:51.050
at the Missoula fire lab
in particular, Matt Jolly
00:04:51.050 --> 00:04:53.120
has continued to support us in the analysis,
00:04:53.120 --> 00:04:54.770
particularly a few moisture data.
00:04:58.930 --> 00:05:03.020
The purpose of today's
presentation is pretty simple.
00:05:03.020 --> 00:05:06.123
First and foremost is to
review the scope of the analysis.
00:05:09.224 --> 00:05:13.480
I just saw a notice, Tony,
that somebody they're not
00:05:13.480 --> 00:05:14.403
seeing our slides.
00:05:18.571 --> 00:05:20.685
(muffled speaking)
00:05:20.685 --> 00:05:23.860
the background is (indistinct)
00:05:28.970 --> 00:05:31.723
I'll just wait here while we
look at this technical issue.
00:05:32.620 --> 00:05:33.693
Yeah be standby.
00:05:47.890 --> 00:05:48.723
It's good now.
00:05:53.060 --> 00:05:55.030
Great, I think we're past that.
00:05:55.030 --> 00:05:56.690
There's always a demo demon,
00:05:56.690 --> 00:05:58.270
always raises its head anytime
00:05:58.270 --> 00:06:00.800
you try to do a presentation,
particularly with technology.
00:06:00.800 --> 00:06:02.830
So I think we just ran into it.
00:06:02.830 --> 00:06:05.943
Hopefully it goes asleep
for the rest of the session.
00:06:07.000 --> 00:06:08.920
All right, I'll jump back in.
00:06:08.920 --> 00:06:10.740
Purpose of today's presentation
00:06:10.740 --> 00:06:12.180
is really pretty straightforward.
00:06:12.180 --> 00:06:13.760
Our job is first and foremost
00:06:13.760 --> 00:06:16.200
to review the scope of the analysis
00:06:16.200 --> 00:06:19.370
identify the project tasks
that were undertaking.
00:06:19.370 --> 00:06:21.000
And then get into the dirty details
00:06:21.000 --> 00:06:22.690
of what was the data that used.
00:06:22.690 --> 00:06:25.237
It's very critical, the data
in these types of analysis.
00:06:25.237 --> 00:06:27.460
And we did our best efforts to ensure
00:06:27.460 --> 00:06:31.200
that the use the most
advanced and detailed data
00:06:31.200 --> 00:06:33.440
and some of this was obtained
00:06:33.440 --> 00:06:36.680
from the investor owned
utilities to their investments
00:06:36.680 --> 00:06:39.580
particularly in the advanced weather systems.
00:06:39.580 --> 00:06:43.063
And I'll talk a bit about that in detail.
00:06:45.710 --> 00:06:46.940
Mostly we're gonna focus
00:06:46.940 --> 00:06:49.234
on the analysis methods that were applied
00:06:49.234 --> 00:06:52.700
and the results of the analysis.
00:06:52.700 --> 00:06:56.710
Focusing on those factual
results and helping you
00:06:56.710 --> 00:06:59.990
understand how all the
components came together.
00:06:59.990 --> 00:07:03.133
And then lastly, we'll support
any questions and answers.
00:07:08.550 --> 00:07:12.080
As a preface, we wanted to set the scene here
00:07:12.080 --> 00:07:14.003
for what this analysis is all about.
00:07:15.010 --> 00:07:17.110
We all know that we see
an increased frequency
00:07:17.110 --> 00:07:19.051
of extreme weather events
00:07:19.051 --> 00:07:22.771
over the past several
years and this is leading
00:07:22.771 --> 00:07:26.300
to more rapid fire spread
and related destruction.
00:07:26.300 --> 00:07:30.630
And of course, we saw this in 2018
00:07:30.630 --> 00:07:34.000
and then also again in
2020 where we had our most
00:07:34.000 --> 00:07:35.913
unprecedented fire season ever,
00:07:36.860 --> 00:07:40.060
we've seen a practice of
de energizing power lines
00:07:40.060 --> 00:07:42.080
and response to these weather events.
00:07:42.080 --> 00:07:44.360
It's grown in its use and prevalence.
00:07:44.360 --> 00:07:47.497
And the term PSPS is
Public Safety Power Shutoff
00:07:47.497 --> 00:07:51.820
has become well known to
the public throughout California.
00:07:51.820 --> 00:07:53.833
And in fact, across the U.S nowadays.
00:07:55.120 --> 00:07:57.050
We also understand there's a better need
00:07:57.050 --> 00:07:59.560
to really understand and quantify
00:07:59.560 --> 00:08:02.970
what is the risk situation that
occurs during these events.
00:08:02.970 --> 00:08:06.183
And prior to this study
that hadn't really been done.
00:08:08.090 --> 00:08:10.730
We also appreciate that
there's a need to evolve
00:08:10.730 --> 00:08:15.410
and refine the use of
PSPS through collaboration
00:08:15.410 --> 00:08:18.574
with the public, the CPUC regulatory agencies
00:08:18.574 --> 00:08:22.120
authoritative fire agencies, such as CAL Fire
00:08:22.120 --> 00:08:23.633
and investor on utilities.
00:08:25.270 --> 00:08:26.607
As Tony identified early on
00:08:26.607 --> 00:08:30.950
the CPUC engaged Technosilva
from La Jolla California
00:08:30.950 --> 00:08:35.950
to conduct this advanced
analysis for these 2019 events.
00:08:38.960 --> 00:08:40.280
Let's briefly talk about
00:08:40.280 --> 00:08:42.573
the project objectives for this analysis.
00:08:49.320 --> 00:08:51.610
The scope is really the conducting analysis
00:08:51.610 --> 00:08:56.610
for each of the individual 2019 PSPS events
00:08:56.690 --> 00:09:00.300
with a focus on allowing the
CPUC to better understand
00:09:00.300 --> 00:09:02.580
one the severity of the weather conditions
00:09:02.580 --> 00:09:05.820
and two the potential risks that were averted
00:09:05.820 --> 00:09:08.620
from wildfires that could
have been been ignited
00:09:08.620 --> 00:09:12.580
from the electric utility
infrastructure ignition sources
00:09:12.580 --> 00:09:14.980
based on survey of damages sustained
00:09:14.980 --> 00:09:16.383
following the power shutoff.
00:09:20.400 --> 00:09:22.300
So the project really focuses
00:09:22.300 --> 00:09:26.030
on applying advanced fire
modeling to quantify the risk
00:09:26.030 --> 00:09:28.890
associated with these potential ignitions
00:09:28.890 --> 00:09:31.300
from the investor owned
assets during the event
00:09:32.200 --> 00:09:34.350
the key component of it, of course
00:09:34.350 --> 00:09:36.860
is utilizing this advanced
fire spread modeling
00:09:36.860 --> 00:09:38.087
is never done before.
00:09:38.087 --> 00:09:41.570
And we do this to really analyze the impact
00:09:41.570 --> 00:09:44.190
trying to quantify the consequence
00:09:44.190 --> 00:09:46.330
that could occur from damages
00:09:46.330 --> 00:09:48.033
that could lead to fire ignition.
00:09:51.870 --> 00:09:54.763
Quite simply, the analysis
workflow is three steps,
00:09:56.120 --> 00:09:58.837
phase one, investor own utilities
00:09:58.837 --> 00:10:01.420
based on regulatory requirements,
00:10:01.420 --> 00:10:03.890
after PSPS events go out
00:10:03.890 --> 00:10:07.985
and survey potential damages
along their infrastructure
00:10:07.985 --> 00:10:11.754
they recorded detail
the information about that
00:10:11.754 --> 00:10:15.020
and provide that as possible,
00:10:15.020 --> 00:10:18.650
we call damage incidents,
possible ignition sources.
00:10:18.650 --> 00:10:20.860
We use those ignition sources
00:10:20.860 --> 00:10:23.381
with the estimated time of damage
00:10:23.381 --> 00:10:26.340
to simulate potential fires
00:10:27.670 --> 00:10:32.670
within the assumption that
those ignitions would occur.
00:10:33.050 --> 00:10:34.973
If that line was not de-energized,
00:10:36.550 --> 00:10:38.840
once we get the results of all these,
00:10:38.840 --> 00:10:41.300
then we can start to analyze the results
00:10:41.300 --> 00:10:43.270
and rank those damage impacts
00:10:43.270 --> 00:10:44.510
to identify what we refer
00:10:44.510 --> 00:10:47.780
to as the most significant potential fires.
00:10:47.780 --> 00:10:50.960
Again, important to note
here that we're simulating fires.
00:10:50.960 --> 00:10:53.470
These are not fires that actually occurred.
00:10:53.470 --> 00:10:55.543
We're simulating what could have been.
00:10:57.960 --> 00:10:58.960
A few minutes minutes
00:10:58.960 --> 00:11:01.660
I'll show you a more detailed workflow
00:11:01.660 --> 00:11:03.330
it gets into some of the dirty details
00:11:03.330 --> 00:11:05.363
about the data and the analysis methods.
00:11:06.670 --> 00:11:09.620
Everything is driven by data
and the quality of that data
00:11:10.582 --> 00:11:12.930
bad data you're gonna get bad results,
00:11:12.930 --> 00:11:15.448
good data you'll get
good results that reflect
00:11:15.448 --> 00:11:18.300
observed and predicted fire behavior.
00:11:18.300 --> 00:11:20.640
The objective here for us as modelers
00:11:20.640 --> 00:11:23.010
is to match that as best we can.
00:11:23.010 --> 00:11:26.203
So let's go through some
of those key input datasets,
00:11:27.430 --> 00:11:28.550
of course, as we mentioned
00:11:28.550 --> 00:11:30.720
we have those damaged asset locations.
00:11:30.720 --> 00:11:31.960
I'll give you a little more details
00:11:31.960 --> 00:11:33.277
about those in the minutes.
00:11:33.277 --> 00:11:35.900
And they become the ignition sources
00:11:35.900 --> 00:11:37.383
for these potential fires.
00:11:38.310 --> 00:11:40.500
Key things are the location of those.
00:11:40.500 --> 00:11:43.730
And then the estimated
time of actually ignition
00:11:43.730 --> 00:11:45.253
or when damage occurred.
00:11:47.870 --> 00:11:49.390
That information came directly
00:11:49.390 --> 00:11:51.023
from the investor own utilities.
00:11:53.490 --> 00:11:56.853
A real fundamental input is weather data.
00:11:59.214 --> 00:12:03.530
In this regard, we're looking
at modeled prediction data.
00:12:03.530 --> 00:12:06.090
Now the IOUs have invested heavily
00:12:06.090 --> 00:12:10.010
over the last several years to generate these
00:12:10.010 --> 00:12:11.890
and build these advanced weather systems
00:12:11.890 --> 00:12:14.600
allow them to have proactive insights
00:12:14.600 --> 00:12:17.450
into the conditions coming down
00:12:17.450 --> 00:12:19.600
in the next two, three, four, or five days.
00:12:24.720 --> 00:12:28.360
I'm just seeing that there's
a bit of a delay on the slides.
00:12:28.360 --> 00:12:29.880
That's a significant
00:12:29.880 --> 00:12:32.933
just wanna confirm that that's happening.
00:12:37.770 --> 00:12:38.880
Yes, there's a slight delay
00:12:38.880 --> 00:12:41.540
between what you're seeing at your end
00:12:41.540 --> 00:12:43.623
and what (indistinct) relays.
00:12:46.560 --> 00:12:47.800
All right.
00:12:47.800 --> 00:12:51.717
I'll try to looks like I'm
on the weather slide
00:12:51.717 --> 00:12:54.260
and the ignition slide is the one showing
00:12:57.440 --> 00:12:58.920
I'll wait till you see the weather slide
00:12:58.920 --> 00:13:00.760
and we'll get going from there.
00:13:00.760 --> 00:13:02.610
Please let me know when that appears.
00:13:16.740 --> 00:13:17.790
Weather slide's up.
00:13:18.720 --> 00:13:19.940
Thank you.
00:13:19.940 --> 00:13:23.940
So a key element is this
predicted weather data
00:13:23.940 --> 00:13:25.940
which we received directly from the IOUs
00:13:26.970 --> 00:13:29.990
San Diego gas and electric
Southern California Edison
00:13:29.990 --> 00:13:32.140
and PG&E or collectively referred to
00:13:32.140 --> 00:13:34.560
as the IOUs for the rest of the presentation.
00:13:34.560 --> 00:13:36.970
They've invested heavily in the development
00:13:36.970 --> 00:13:40.180
of these systems to
support their decision-making
00:13:40.180 --> 00:13:41.390
for a variety of purposes.
00:13:41.390 --> 00:13:43.397
One of those is now PSPS.
00:13:43.397 --> 00:13:45.140
The slight variations
00:13:45.140 --> 00:13:49.780
and the format and content of that data.
00:13:49.780 --> 00:13:51.398
They're all one hour predictions
00:13:51.398 --> 00:13:53.360
through a series of variables.
00:13:53.360 --> 00:13:55.050
You'll see those on the left-hand side
00:13:55.050 --> 00:13:57.000
the example there is showing wind speed
00:13:58.040 --> 00:14:01.430
and they vary between 72 and 84 hours.
00:14:01.430 --> 00:14:04.360
Since this analysis that
weather data is now increased
00:14:04.360 --> 00:14:07.394
to a hundred and 124 plus hours.
00:14:07.394 --> 00:14:11.830
But for this, it was
between analysis 72 to 84.
00:14:11.830 --> 00:14:13.260
Spatially, the resolution
00:14:13.260 --> 00:14:15.307
is between two and three kilometers
00:14:15.307 --> 00:14:18.333
which compared to publicly available sources
00:14:18.333 --> 00:14:23.170
from NWS and others is
considered high resolution.
00:14:23.170 --> 00:14:26.023
So it varies in both spatial
and temporal resolution.
00:14:33.110 --> 00:14:37.930
In adjacent, in combination perhaps
00:14:37.930 --> 00:14:40.360
with the weather prediction day, of course
00:14:40.360 --> 00:14:43.410
is the weather station of
weather observation data.
00:14:43.410 --> 00:14:44.940
We're fortunate in California
00:14:44.940 --> 00:14:47.250
to have the densest network in the nation.
00:14:47.250 --> 00:14:50.880
That's a combination
of public, raw, temporary
00:14:50.880 --> 00:14:52.640
and then weather sites provide
00:14:52.640 --> 00:14:55.000
by the investor-owned utilities.
00:14:55.000 --> 00:14:57.540
And this, the bottom line is so
00:14:57.540 --> 00:15:00.110
as you can see from the density of the map
00:15:00.110 --> 00:15:01.630
is a significant amount of weather
00:15:01.630 --> 00:15:04.290
observation data with recordings coming
00:15:04.290 --> 00:15:06.880
in at least every 10 minutes.
00:15:06.880 --> 00:15:09.360
In combination with the weather forecast data
00:15:09.360 --> 00:15:11.870
this provides a fantastic basis
00:15:11.870 --> 00:15:14.570
for us to really evaluate what were actually
00:15:14.570 --> 00:15:16.823
the conditions during these PSPS events.
00:15:21.080 --> 00:15:22.870
Key input into fire modeling
00:15:22.870 --> 00:15:25.530
is the evaluation of fuel moistures.
00:15:25.530 --> 00:15:28.607
These are referred to as
dead and live fuel moistures.
00:15:29.980 --> 00:15:31.763
Again, the investor owned utilities
00:15:31.763 --> 00:15:36.110
have invested heavily in
developing these enhanced models
00:15:36.110 --> 00:15:37.660
for the state of California.
00:15:37.660 --> 00:15:39.290
This goes well above and beyond.
00:15:39.290 --> 00:15:42.240
What's available from the academic community
00:15:42.240 --> 00:15:46.493
and or the agency, whether
agencies across the U.S.
00:15:47.370 --> 00:15:50.010
We get the benefits from
this as it is a key input
00:15:50.010 --> 00:15:51.060
in the fire modeling.
00:15:52.270 --> 00:15:54.500
Each of the IOUs at the time of this
00:15:54.500 --> 00:15:57.277
use the consistent
provider to provide this data.
00:15:57.277 --> 00:15:59.880
And we can see from
this example on the screen
00:15:59.880 --> 00:16:01.880
the one hour dead fuel moisture
00:16:01.880 --> 00:16:03.763
and the resolution of this data.
00:16:08.120 --> 00:16:09.520
Are we tracking slides okay?
00:16:12.960 --> 00:16:15.570
Yes there's about a 30 to 42nd delay
00:16:15.570 --> 00:16:17.450
between when you switch the slide
00:16:17.450 --> 00:16:19.680
and when the admin monitor switches a slide
00:16:20.560 --> 00:16:22.723
it just switched to fuel moistures now.
00:16:23.775 --> 00:16:25.320
And that's unfortunate.
00:16:25.320 --> 00:16:27.723
Yeah, it's just the nature of the system.
00:16:30.450 --> 00:16:32.993
Let me know when you
can see the fuels data slide.
00:17:00.170 --> 00:17:01.003
Okay.
00:17:02.540 --> 00:17:05.110
So in addition to fuel
moistures, another key input
00:17:05.110 --> 00:17:08.310
for fire modeling is surfacing canopy fuels
00:17:08.310 --> 00:17:10.620
typically referred to as fuels data.
00:17:10.620 --> 00:17:13.070
Often the public and others get confused
00:17:13.070 --> 00:17:15.830
because of the term fuels in both of those
00:17:15.830 --> 00:17:18.420
one is fuel moisture information derived
00:17:18.420 --> 00:17:20.940
and typically is bundled with weather data.
00:17:20.940 --> 00:17:22.750
And this is a characterization
00:17:22.750 --> 00:17:25.460
of the landscape and
particularly the vegetation
00:17:25.460 --> 00:17:30.213
on the landscape about how
we behave in fire situations.
00:17:34.690 --> 00:17:38.710
The fuels data being
utilized here was by default
00:17:38.710 --> 00:17:43.117
the 2016 data source
Technosilva provided updates
00:17:45.235 --> 00:17:48.690
to that data for this analysis
to reflect disturbances
00:17:48.690 --> 00:17:51.000
such as fires, urban growth,
00:17:51.000 --> 00:17:54.310
that occurred from 2016 to 2019.
00:17:54.310 --> 00:17:56.910
This is one done to bring that fuels dataset
00:17:56.910 --> 00:18:00.060
up-to-date so that we had
the latest representation
00:18:00.060 --> 00:18:01.710
of those fields on the landscape.
00:18:02.580 --> 00:18:06.050
It uses the 2005 fuel model data set
00:18:06.050 --> 00:18:08.550
it's called classification methodology
00:18:08.550 --> 00:18:10.900
which is standard across
fire behavior modeling
00:18:11.760 --> 00:18:13.363
at a 30 meter resolution.
00:18:14.790 --> 00:18:17.510
However, it's a better address encroachment
00:18:17.510 --> 00:18:20.170
and potential consequences
and impact analysis.
00:18:20.170 --> 00:18:22.720
Technosilva has enhanced this
00:18:22.720 --> 00:18:26.660
with custom fuel models
that describe WUI areas
00:18:26.660 --> 00:18:28.520
wild land, urban interface areas
00:18:28.520 --> 00:18:32.130
where we get a mix of
vegetation with housing.
00:18:32.130 --> 00:18:34.090
We use a combination
00:18:34.090 --> 00:18:38.550
of building density using building footprints
00:18:38.550 --> 00:18:42.100
with fuel load to better
characterize the fuels
00:18:42.100 --> 00:18:46.533
around where consequences
occur, people and buildings.
00:18:48.570 --> 00:18:50.960
This allows us to utilize
it in the fire modeling
00:18:50.960 --> 00:18:55.090
to get much more accurate impact analysis.
00:18:55.090 --> 00:18:57.670
Important to note that
this fuel data has also
00:18:57.670 --> 00:19:01.257
been calibrated with
significantly with past fire behavior.
00:19:01.257 --> 00:19:03.550
And it is a specialty of Technosilva's
00:19:03.550 --> 00:19:06.420
and something we continue
to provide to our customers
00:19:06.420 --> 00:19:09.770
that goes well beyond
any publicly available data.
00:19:09.770 --> 00:19:12.473
So fuel moistures field data really critical.
00:19:15.603 --> 00:19:17.890
This slide gives you a little bit of a feel
00:19:17.890 --> 00:19:21.120
of the detail of that data with a legend.
00:19:21.120 --> 00:19:24.000
Appreciate you can't read the
legend on the left-hand side.
00:19:24.000 --> 00:19:26.470
You start to see the detailed resolution.
00:19:26.470 --> 00:19:28.490
The black areas represent urban areas
00:19:28.490 --> 00:19:30.200
the blue, obviously water.
00:19:30.200 --> 00:19:31.658
And then you'll see a color of
00:19:31.658 --> 00:19:34.440
of yellows representing grasses.
00:19:34.440 --> 00:19:35.330
We move into orange
00:19:35.330 --> 00:19:38.390
and Browns for shrub and grass shrub mixes.
00:19:38.390 --> 00:19:40.410
We move into greens and light blues.
00:19:40.410 --> 00:19:41.623
As we get into timber.
00:19:43.660 --> 00:19:45.270
There's a series of enhanced fuels
00:19:45.270 --> 00:19:48.324
at the bottom end, particularly
around those urban areas
00:19:48.324 --> 00:19:50.340
where we've done this enhancement
00:19:51.753 --> 00:19:54.393
to better accommodate
the consequences analysis.
00:19:57.390 --> 00:19:58.700
Now let's talk about
00:19:58.700 --> 00:20:01.413
the actual impact analysis and consequence.
00:20:03.310 --> 00:20:05.563
Let me know if you're
following with this slide.
00:20:07.590 --> 00:20:09.200
It just switched to field data.
00:20:09.200 --> 00:20:12.093
So give it a few seconds to catch up.
00:20:14.030 --> 00:20:14.863
Will do.
00:20:52.830 --> 00:20:56.350
Of course a critical
input is the consequences
00:20:56.350 --> 00:20:57.490
or impacts we care about.
00:20:57.490 --> 00:21:00.623
This is generally referred
to as values at risk.
00:21:02.318 --> 00:21:03.970
The two primary values at risk used
00:21:03.970 --> 00:21:06.850
in this analysis was
the location of buildings
00:21:06.850 --> 00:21:08.920
and particularly building footprints
00:21:08.920 --> 00:21:12.410
Microsoft 2018 data set was used,
00:21:12.410 --> 00:21:14.540
although it's been enhanced by Technosilva
00:21:15.540 --> 00:21:18.380
to add about 340,000 buildings at the time
00:21:18.380 --> 00:21:22.820
of this analysis, this is the
cleanup minor mistakes found
00:21:22.820 --> 00:21:25.190
new urban growth areas like that.
00:21:25.190 --> 00:21:28.981
This map shows original buildings in red
00:21:28.981 --> 00:21:31.910
buildings that dots in black are
00:21:31.910 --> 00:21:33.410
the ones added by Technosilva.
00:21:35.690 --> 00:21:37.956
In addition, we use population information
00:21:37.956 --> 00:21:41.104
the de facto standard for population data
00:21:41.104 --> 00:21:45.678
and risk analysis is you Oak
Ridge national laboratories
00:21:45.678 --> 00:21:47.760
land scan data.
00:21:47.760 --> 00:21:52.543
The best available data at
the time was the 2018 data.
00:21:53.510 --> 00:21:54.730
This data is really good
00:21:54.730 --> 00:21:57.840
because it characterizes
the location of populations
00:21:57.840 --> 00:22:01.840
and wild land areas where
the census is not so good.
00:22:01.840 --> 00:22:03.630
And it's become that defacto
00:22:03.630 --> 00:22:05.603
for risk assessments across the nation.
00:22:07.730 --> 00:22:09.170
Our focus was really
00:22:09.170 --> 00:22:13.752
on utilizing this data to
capture the information
00:22:13.752 --> 00:22:17.083
about potential impacts and consequence.
00:22:18.700 --> 00:22:21.200
It's important to note that
depending on the purpose
00:22:21.200 --> 00:22:23.740
of a risk analysis like this other values
00:22:23.740 --> 00:22:25.260
at risk could be applied
00:22:27.500 --> 00:22:32.300
such as sensitive areas, habitat areas,
00:22:32.300 --> 00:22:34.810
landslide potential, things of that nature.
00:22:34.810 --> 00:22:38.940
The focus of this was on
buildings and population impacts
00:22:38.940 --> 00:22:42.483
which tend to be the priority
when evaluating wildfire risk.
00:22:44.580 --> 00:22:46.503
Let's move on and talk about methods.
00:22:48.580 --> 00:22:51.290
We've got this baseline
of data used in the analysis
00:22:52.140 --> 00:22:54.050
but it's all about our methods.
00:22:54.050 --> 00:22:57.010
How do we apply data processing methods
00:22:57.010 --> 00:22:58.550
and into fire behavior modeling
00:22:58.550 --> 00:23:01.063
with weather analysis
to generate these results.
00:23:06.630 --> 00:23:09.870
This slide breaks down the simple flow chart
00:23:09.870 --> 00:23:12.017
you saw previously that identified
00:23:12.017 --> 00:23:15.290
the three core analysis tasks to do a series
00:23:15.290 --> 00:23:19.043
of more granular tasks
that define the activities.
00:23:23.100 --> 00:23:26.140
When this slide appears,
you'll see designation
00:23:26.140 --> 00:23:28.260
between red and blue.
00:23:28.260 --> 00:23:32.470
Blue boxes represent the process and data
00:23:32.470 --> 00:23:35.350
identified and provided
by the investor own utilities.
00:23:35.350 --> 00:23:38.933
The red ones represent the
ones undertaken by Technosilva.
00:23:40.450 --> 00:23:43.650
The square boxes represent general processes.
00:23:43.650 --> 00:23:46.840
Although there is many
sub or tasks involved in this
00:23:48.310 --> 00:23:50.150
the cylinder is at the bottom represent
00:23:50.150 --> 00:23:52.453
the actual core data sets that were used.
00:23:53.790 --> 00:23:55.070
Of course, this all starts
00:23:55.070 --> 00:23:58.140
with the IOU de-energizing their assets
00:23:58.140 --> 00:24:01.110
the term we're using to
represent the power lines
00:24:01.110 --> 00:24:04.170
and electric utility infrastructure.
00:24:04.170 --> 00:24:06.120
That of course identifies the boundaries
00:24:06.120 --> 00:24:10.423
of the PSPS event in which
power lines were de-energized.
00:24:11.420 --> 00:24:13.670
Once the event is over the IOUs
00:24:13.670 --> 00:24:16.890
go out and survey for
possible damage on those lines
00:24:16.890 --> 00:24:19.340
and record that information
with a lot of detail.
00:24:21.270 --> 00:24:24.093
That information was
requested as obtained by us.
00:24:24.960 --> 00:24:28.140
And this is where our part
of the analysis kicked in
00:24:29.310 --> 00:24:32.500
to do a series amounts
to identify the explicit
00:24:32.500 --> 00:24:36.720
process used by the IOUs
and to define in particular
00:24:36.720 --> 00:24:39.343
the time estimated time of those damages.
00:24:40.530 --> 00:24:42.950
We combine that with a
range of other datasets
00:24:42.950 --> 00:24:44.670
particularly the weather prediction
00:24:44.670 --> 00:24:46.767
and fuel moistures obtained by the IOUs
00:24:47.834 --> 00:24:49.360
with the other data compiled
00:24:49.360 --> 00:24:51.000
and developed by Technosilva,
00:24:51.000 --> 00:24:54.050
weather observation
data fields and landscape.
00:24:54.050 --> 00:24:57.210
And of course the values at
risk buildings and population.
00:24:57.210 --> 00:25:00.366
We conducted hundreds
of millions of fire simulations
00:25:00.366 --> 00:25:03.120
in an iterative process
with quality assurance
00:25:03.120 --> 00:25:06.240
and review to really
identify the potential impacts
00:25:06.240 --> 00:25:09.240
for each of those damage incidents.
00:25:09.240 --> 00:25:11.730
Each damage identified on an asset
00:25:11.730 --> 00:25:13.380
is referred to a damage incident.
00:25:14.460 --> 00:25:17.300
Then we can start to
analyze those by ranking them
00:25:17.300 --> 00:25:20.108
really identifying which
ones the most significant.
00:25:20.108 --> 00:25:23.330
And I'll talk about this
in a little more detail
00:25:23.330 --> 00:25:24.790
as we go through the methods
00:25:24.790 --> 00:25:26.812
once we've identified the most significant,
00:25:26.812 --> 00:25:29.366
then we can conduct even
more now detailed analysis
00:25:29.366 --> 00:25:31.663
with probabilistic methods.
00:25:36.170 --> 00:25:38.827
Let's talk about the
details of damage incidents.
00:25:41.579 --> 00:25:43.129
How we doing tracking on slide.
00:25:46.570 --> 00:25:47.500
I would just keep going.
00:25:47.500 --> 00:25:50.330
It seems to be hit or miss on some people are
00:25:50.330 --> 00:25:53.670
in totally in sync with you
David, and some have a delay.
00:25:53.670 --> 00:25:55.124
So I would continue on
00:25:55.124 --> 00:25:58.340
we'll post the slides after the presentation.
00:25:58.340 --> 00:25:59.190
It's a huge deck.
00:25:59.190 --> 00:26:01.650
We'll have to figure out
how to manage it and get it
00:26:01.650 --> 00:26:02.530
but we'll get it uploaded
00:26:02.530 --> 00:26:03.940
to the website after we're done with it.
00:26:03.940 --> 00:26:05.513
All right.
00:26:05.513 --> 00:26:06.346
Fair enough.
00:26:06.346 --> 00:26:08.316
I'll try to put the
slides up a little earlier
00:26:08.316 --> 00:26:10.940
so I can lead up to those.
00:26:10.940 --> 00:26:14.040
Let's talk a bit of detail about
the damage incident data
00:26:14.040 --> 00:26:15.660
these damaged asset locations.
00:26:15.660 --> 00:26:18.470
We recall damage
incidents were captured again
00:26:18.470 --> 00:26:21.683
by the IOUs to field surveys
using GPS techniques.
00:26:22.850 --> 00:26:24.120
So we have a very detailed
00:26:24.120 --> 00:26:26.740
and accurate location of that damage.
00:26:26.740 --> 00:26:29.380
They also collected a
variety of other information
00:26:29.380 --> 00:26:31.350
about the particular unique identifier
00:26:31.350 --> 00:26:33.890
for the asset characteristics of the assets
00:26:33.890 --> 00:26:36.893
including photographs that
helps to really document that.
00:26:38.240 --> 00:26:39.790
A key element of the process
00:26:39.790 --> 00:26:42.740
is the formal process undertaken by the IOUs
00:26:42.740 --> 00:26:46.027
was for their engineering
staff to evaluate that damage
00:26:46.027 --> 00:26:49.040
and determine if the likelihood of arcing
00:26:49.040 --> 00:26:51.380
if the system had remained energized
00:26:51.380 --> 00:26:54.630
would occur resulting
in a probable fire ignition.
00:26:54.630 --> 00:26:56.570
Ultimately we care about whether
00:26:56.570 --> 00:26:58.483
that ignition have occurred or not.
00:26:59.350 --> 00:27:01.742
Each IOUs slightly different methods
00:27:01.742 --> 00:27:04.490
although all of them use seasoned
00:27:04.490 --> 00:27:07.620
experienced engineers in
making that determination
00:27:07.620 --> 00:27:11.363
and also documented that with photographs.
00:27:12.770 --> 00:27:15.350
From this we got an estimated time of damage
00:27:15.350 --> 00:27:17.609
which would be our ignition time.
00:27:17.609 --> 00:27:22.609
We reviewed that curated
against the outage time
00:27:22.710 --> 00:27:25.500
starting and end times to
make sure it was within there.
00:27:25.500 --> 00:27:27.920
This is where we also pulled
in the advanced weather
00:27:27.920 --> 00:27:30.910
analysis to determine what
were the worst conditions
00:27:30.910 --> 00:27:33.840
during that event at that specific location.
00:27:33.840 --> 00:27:37.500
And this was aligned to the
damage time that was reported.
00:27:37.500 --> 00:27:39.270
Most cases that fit very nicely.
00:27:39.270 --> 00:27:41.920
In some cases, there's minor adjustments.
00:27:41.920 --> 00:27:44.882
We use that damage location and that time
00:27:44.882 --> 00:27:48.933
as the input into the fire spread it.
00:27:53.330 --> 00:27:55.497
This slide actually gives you an example
00:27:55.497 --> 00:27:57.653
from a PG and a bet on October 9th,
00:27:57.653 --> 00:27:59.680
that I'll go through in a few minutes
00:28:01.250 --> 00:28:04.800
that shows the PSPS
event boundaries on the left,
00:28:04.800 --> 00:28:06.870
and then the damage locations.
00:28:06.870 --> 00:28:08.543
It was about 116 of them,
00:28:09.776 --> 00:28:12.180
throughout their network.
00:28:12.180 --> 00:28:14.780
The electric utility network
is not shown on the map
00:28:16.300 --> 00:28:19.960
only the PSPS events boundaries,
and the damage locations.
00:28:19.960 --> 00:28:21.660
And they're numbered sequentially.
00:28:22.750 --> 00:28:25.820
the map on the right just
zooms in and shows you
00:28:25.820 --> 00:28:28.090
where there's a cluster of those locations.
00:28:28.090 --> 00:28:31.550
And we can see many
of those cluster together.
00:28:31.550 --> 00:28:33.560
So you can see the
occurred in close proximity
00:28:33.560 --> 00:28:36.653
of press along a segment or a power line.
00:28:41.660 --> 00:28:43.350
Let's talk about the fire spread modeling.
00:28:43.350 --> 00:28:45.590
This is the heart and soul of the analysis
00:28:45.590 --> 00:28:48.450
and is the new element that CPUC added
00:28:48.450 --> 00:28:50.850
into its arsenal tools.
00:28:50.850 --> 00:28:54.180
Simulations were performed
using Technosilva's wildfire
00:28:54.180 --> 00:28:57.930
and less enterprise software
that has a whole suite
00:28:57.930 --> 00:29:00.260
of key capabilities to distinguish it
00:29:00.260 --> 00:29:03.280
from any other modeling software out there
00:29:03.280 --> 00:29:05.453
on the marketplace or in academia.
00:29:06.910 --> 00:29:10.310
It is not one model it's
about 20 plus models.
00:29:10.310 --> 00:29:11.693
22, maybe at this point.
00:29:13.252 --> 00:29:15.910
And the important part is it incorporates
00:29:15.910 --> 00:29:18.220
custom fuels for those WUI areas
00:29:18.220 --> 00:29:21.210
a lot where people live that allows us to
00:29:21.210 --> 00:29:24.040
apply advanced urban encroachment algorithms
00:29:24.040 --> 00:29:27.283
to really calculate those potential impacts.
00:29:29.070 --> 00:29:30.963
This is a key component of the analysis
00:29:30.963 --> 00:29:34.920
that requires enhanced data
but also enhanced modeling.
00:29:34.920 --> 00:29:37.820
It of course it has all the
other standard components
00:29:37.820 --> 00:29:41.252
of fire behavior modeling
surface and crown fire
00:29:41.252 --> 00:29:45.820
potential analysis, incorporate
that detailed fuel moistures
00:29:45.820 --> 00:29:48.630
some of it through advanced
machine learning techniques
00:29:48.630 --> 00:29:51.610
and includes spotting and
it supports deterministic
00:29:51.610 --> 00:29:53.463
and probabilistic modes as well.
00:29:56.800 --> 00:30:00.644
So why does CPUC select
Technosilva's wildfire analyst
00:30:00.644 --> 00:30:03.150
over other options that are out there
00:30:03.150 --> 00:30:05.450
particularly those from university?
00:30:05.450 --> 00:30:07.920
Let me give you a bit of background on that.
00:30:07.920 --> 00:30:10.530
In 2019, April, 2019,
00:30:10.530 --> 00:30:12.660
the state of California published a request
00:30:12.660 --> 00:30:15.443
for innovative ideas driven by CAL Fire.
00:30:18.660 --> 00:30:20.440
That initiative was to focus
00:30:20.440 --> 00:30:24.067
on developing concise problem statements
00:30:24.067 --> 00:30:28.780
and having the industry
government and academia,
00:30:28.780 --> 00:30:30.550
the research environments respond
00:30:30.550 --> 00:30:32.490
to those with innovative ideas
00:30:32.490 --> 00:30:37.020
about how they might be solved
with science and technology.
00:30:37.020 --> 00:30:39.113
There were two key problems statements,
00:30:40.230 --> 00:30:41.992
early wildfire detection,
00:30:41.992 --> 00:30:46.200
although the detection network
in California is quite good
00:30:46.200 --> 00:30:49.150
and fires are identified pretty quickly.
00:30:49.150 --> 00:30:50.710
And then most importantly,
00:30:50.710 --> 00:30:52.587
how to predict the path of a fire
00:30:52.587 --> 00:30:55.660
and what the potential consequences would be.
00:30:55.660 --> 00:30:58.330
That's often was very difficult for them.
00:30:58.330 --> 00:31:02.420
And this process really focused
00:31:02.420 --> 00:31:04.643
on the requests and ideas on this area.
00:31:07.960 --> 00:31:09.470
CAL Fire and the state of California
00:31:09.470 --> 00:31:11.719
received over 130 submittals
00:31:11.719 --> 00:31:14.940
and ultimately selected
Technosilva is the vendor
00:31:14.940 --> 00:31:18.403
of choice to enter into a
proof of concept analysis.
00:31:20.141 --> 00:31:23.738
This analysis was
undertaken in 2019 fire season
00:31:23.738 --> 00:31:26.510
and based on a successful review
00:31:26.510 --> 00:31:28.040
of that proof of concept
00:31:28.040 --> 00:31:33.040
was implemented into full
production in 2019 and 2020.
00:31:33.161 --> 00:31:37.390
The product became the
authoritative fire modeling solution
00:31:37.390 --> 00:31:38.790
for the state of California.
00:31:46.230 --> 00:31:48.600
So what does Wildfire Analyst look like?
00:31:48.600 --> 00:31:50.080
It's a commercial product based
00:31:50.080 --> 00:31:53.620
on 30 years of science and 15 years
00:31:53.620 --> 00:31:56.900
of technology implementation that continually
00:31:56.900 --> 00:31:59.933
gets enhanced with new science as we go.
00:32:01.410 --> 00:32:04.850
Let me show you a little
bit about what it looks like.
00:32:04.850 --> 00:32:05.683
I pulled off
00:32:05.683 --> 00:32:07.817
the Wildfire Analyst
Enterprise environment here
00:32:07.817 --> 00:32:10.380
and I'm actually showing
you one of the simulations
00:32:10.380 --> 00:32:13.763
for our damage incident
it's 24 hour simulation.
00:32:16.368 --> 00:32:19.297
The red dots on the map
represent building locations.
00:32:21.540 --> 00:32:23.710
Now, of course, to run
these types of simulations
00:32:23.710 --> 00:32:27.160
they can be done interactively
like I'm showing you here
00:32:27.160 --> 00:32:29.680
like is done for operational response
00:32:29.680 --> 00:32:31.523
by CAL Fire during the fire season.
00:32:33.330 --> 00:32:34.320
In addition, they can be
00:32:34.320 --> 00:32:37.200
run in automated mode on supercomputers
00:32:38.310 --> 00:32:41.280
as was done a CPUC analysis that allowed us
00:32:41.280 --> 00:32:44.193
to run hundreds of
millions of these simulations.
00:32:46.120 --> 00:32:47.230
They're all retrievable
00:32:47.230 --> 00:32:50.610
and therefore can be analyzed and reviewed.
00:32:50.610 --> 00:32:53.440
And this is very important
for quality assurance.
00:32:53.440 --> 00:32:55.850
Another key element of a modeling environment
00:32:55.850 --> 00:32:59.680
like this is it, as you saw
requires significant input data.
00:32:59.680 --> 00:33:02.477
And so we need the
ability to bundle that data
00:33:02.477 --> 00:33:04.960
and to make it readily available
00:33:04.960 --> 00:33:07.120
and then perform the modeling
00:33:07.120 --> 00:33:09.110
in virtual real-time.
00:33:09.110 --> 00:33:11.923
The simulations are
undertaken in less than a minute.
00:33:13.600 --> 00:33:15.671
This technology is used by CAL Fire
00:33:15.671 --> 00:33:19.950
and in the 2020 fire season
modeled over 10,000 fires
00:33:22.221 --> 00:33:23.970
with simulations in less than a minute.
00:33:23.970 --> 00:33:26.190
This is important because previously
00:33:26.190 --> 00:33:28.450
we never had this capability to understand
00:33:28.450 --> 00:33:30.760
what is the probable path of the fire,
00:33:30.760 --> 00:33:32.453
what other potential impacts?
00:33:33.472 --> 00:33:36.303
In this case we're applying
it to these damage incidents.
00:33:40.240 --> 00:33:43.643
Now, fire spread modeling
has a lot of little details to it.
00:33:45.000 --> 00:33:47.200
We applied both what we call deterministic
00:33:47.200 --> 00:33:49.713
and probabilistic modes for modeling.
00:33:50.650 --> 00:33:53.510
Deterministic is most of
the fire simulations, you see
00:33:53.510 --> 00:33:55.260
like the example I just showed you,
00:33:56.820 --> 00:33:59.463
and they were run in this case 24 hours,
00:34:01.070 --> 00:34:04.080
the weather prediction
was used as key input data
00:34:05.670 --> 00:34:08.020
throughout those important to note
00:34:08.020 --> 00:34:09.420
that that weather data changed
00:34:09.420 --> 00:34:13.280
both spatially across
the landscape, every hour.
00:34:13.280 --> 00:34:16.150
So we're not using a lot of constant values,
00:34:16.150 --> 00:34:19.918
like a lot of the fire models
use when you see dynamically
00:34:19.918 --> 00:34:22.773
changing spatially and
temporally across the landscape.
00:34:23.960 --> 00:34:26.910
As I mentioned that live fuel moisture was in
00:34:26.910 --> 00:34:28.940
and surfacing canopy fuels were used
00:34:28.940 --> 00:34:32.060
including this enhancement
to better understand impacts
00:34:32.060 --> 00:34:32.893
into the air.
00:34:35.840 --> 00:34:40.840
With deterministic, the data
it defines a weather scenario
00:34:41.290 --> 00:34:42.930
fuel moistures and fuel scenario.
00:34:42.930 --> 00:34:45.013
And those inputs are used explicitly.
00:34:47.090 --> 00:34:49.617
As you just saw in the animated example
00:34:49.617 --> 00:34:54.010
this is what deterministic
simulation looks like.
00:34:54.010 --> 00:34:56.340
And this is one of the ones
we'll review in a minute.
00:34:56.340 --> 00:34:58.270
In this case buildings are in black.
00:34:58.270 --> 00:35:00.800
We see hourly perimeters in light gray
00:35:00.800 --> 00:35:03.900
and the shading of green
for the first few hours
00:35:03.900 --> 00:35:05.540
as it goes out to red.
00:35:05.540 --> 00:35:08.770
This particular example was 24 hours duration
00:35:08.770 --> 00:35:12.772
which we use for the
project about 24,000 acres.
00:35:12.772 --> 00:35:16.160
We'll talk about initial attack
assessment in a minute.
00:35:16.160 --> 00:35:18.090
We can identify the number of buildings
00:35:18.090 --> 00:35:22.763
and total population
potentially impacted in addition,
00:35:24.090 --> 00:35:26.700
to a whole suite of fire behavior metrics
00:35:26.700 --> 00:35:28.910
that we can analyze with respect to rates
00:35:28.910 --> 00:35:32.820
of spread lime length,
byline intensities, et cetera.
00:35:32.820 --> 00:35:34.850
It's important to note in the simulation
00:35:34.850 --> 00:35:37.120
that there was no suppression applied.
00:35:37.120 --> 00:35:39.530
These are without resource suppression.
00:35:39.530 --> 00:35:41.560
It's very difficult to analyze
00:35:41.560 --> 00:35:45.740
in weather scenarios that are extreme.
00:35:45.740 --> 00:35:49.300
We may have many simultaneous incidents
00:35:49.300 --> 00:35:50.600
going on at the same time.
00:35:54.580 --> 00:35:56.640
The other type of fire spread modeling that's
00:35:56.640 --> 00:35:58.523
opt-in uses called probabilistic.
00:35:59.440 --> 00:36:01.980
In our scenario we modeled
every damage incident
00:36:01.980 --> 00:36:03.830
using the deterministic.
00:36:03.830 --> 00:36:06.330
And then we identified, if
you remember the spreadsheet
00:36:06.330 --> 00:36:09.570
those most significant
fires with respect to impacts.
00:36:09.570 --> 00:36:11.960
For those we conducted more advanced analysis
00:36:11.960 --> 00:36:15.350
and included in this was probabilistic.
00:36:15.350 --> 00:36:17.730
Now probabilistic varies from deterministic
00:36:17.730 --> 00:36:20.359
because it adds a level of variation
00:36:20.359 --> 00:36:24.337
or uncertainty in the input data
00:36:24.337 --> 00:36:26.620
and particularly wind speed, wind direction.
00:36:26.620 --> 00:36:29.530
the dead fuel moistures
are the varied inputs.
00:36:29.530 --> 00:36:31.670
We use a variety of variants, typically
00:36:31.670 --> 00:36:34.770
in the 10 to 20% on
either side of those values.
00:36:34.770 --> 00:36:38.973
And we run multiple simulations
using this various inputs.
00:36:39.990 --> 00:36:43.017
It really addresses the
possible uncertainties in the data.
00:36:43.017 --> 00:36:44.330
And this could be important
00:36:44.330 --> 00:36:46.123
when you're using prediction data.
00:36:47.180 --> 00:36:48.800
For each probabilistic run
00:36:48.800 --> 00:36:50.970
we ran a hundred different simulations
00:36:50.970 --> 00:36:53.980
with these varied inputs
to get a similar result
00:36:53.980 --> 00:36:55.573
for that same damage incident.
00:36:56.990 --> 00:37:00.120
This shows an example of a probabilistic run
00:37:00.120 --> 00:37:03.030
it's color coded, and percent of probability
00:37:03.030 --> 00:37:05.880
from zero is less than five through pinks
00:37:05.880 --> 00:37:10.420
at five and 10 blue is 20 to
40, right up to 80 to a hundred
00:37:10.420 --> 00:37:12.330
at that red area.
00:37:12.330 --> 00:37:15.893
Same ignition source, different location.
00:37:17.010 --> 00:37:18.913
I'm sorry, different methodology.
00:37:19.980 --> 00:37:23.100
As referenced, the
boundary of the deterministic
00:37:23.100 --> 00:37:25.600
is shown on that black line work.
00:37:25.600 --> 00:37:27.110
So we get a feeling of comparing
00:37:27.110 --> 00:37:28.993
the deterministic and probabilistic.
00:37:30.680 --> 00:37:34.189
You can see the consequence
values are very similar
00:37:34.189 --> 00:37:39.003
in this analysis, although
we use variance in the inputs.
00:37:40.150 --> 00:37:42.020
So we applied both methods
00:37:42.020 --> 00:37:44.053
as we're moving through our methodology.
00:37:48.680 --> 00:37:50.080
How am I doing for tracking?
00:37:55.080 --> 00:37:57.890
Just continue on David I think you're okay.
00:37:57.890 --> 00:37:59.210
Excellent.
00:37:59.210 --> 00:38:01.800
Now not all fires and in our regards
00:38:01.800 --> 00:38:03.186
not all fire simulations
00:38:03.186 --> 00:38:07.170
in this project are created equally.
00:38:07.170 --> 00:38:10.780
So we need a metric to
be able to evaluate those
00:38:10.780 --> 00:38:13.180
to see what is the
likelihood they're all modeled
00:38:13.180 --> 00:38:14.570
for 24 hours.
00:38:14.570 --> 00:38:17.550
Can we distinguish which ones are more likely
00:38:17.550 --> 00:38:21.430
to perhaps get passing
in what we call initial attack
00:38:21.430 --> 00:38:23.640
to becoming an extended attack fire
00:38:23.640 --> 00:38:25.540
beyond that two to four hour timeframe
00:38:26.490 --> 00:38:30.850
and actually start to really
have a better probability
00:38:30.850 --> 00:38:34.453
of reaching those potential consequences.
00:38:36.300 --> 00:38:37.850
To do this, we developed a metric
00:38:37.850 --> 00:38:39.670
we call the initial attack assessment.
00:38:39.670 --> 00:38:41.530
You can see it's a five scale.
00:38:41.530 --> 00:38:44.900
It really uses behavior
and growth in the first couple
00:38:44.900 --> 00:38:49.690
of hours to really
determine is that fire based
00:38:49.690 --> 00:38:51.690
on its behavior and growth
00:38:51.690 --> 00:38:54.423
likely to escape that initial attack period.
00:38:55.650 --> 00:38:59.761
Remember, agencies like
CAL Fire have a strategic target
00:38:59.761 --> 00:39:04.761
to suppress and contain over 90% of fires
00:39:05.910 --> 00:39:09.200
less than 10 acres during
that initial attack period.
00:39:09.200 --> 00:39:11.300
And they're very successful at doing that.
00:39:11.300 --> 00:39:13.930
You don't hear about that
because of their success.
00:39:13.930 --> 00:39:16.070
Unfortunately, you hear
about the ones that are
00:39:16.070 --> 00:39:19.030
very difficult and escape, initial attack.
00:39:19.030 --> 00:39:22.640
We needed some measure of
evaluating these fire simulations
00:39:22.640 --> 00:39:26.900
to see which ones are
possibly escaping initial attack.
00:39:26.900 --> 00:39:29.120
We developed this metric in combination
00:39:29.120 --> 00:39:31.410
with seasoned fire behavior specialists
00:39:31.410 --> 00:39:33.960
and fire operations folks, many with more
00:39:33.960 --> 00:39:37.640
than 30 or 40 years experience in California.
00:39:37.640 --> 00:39:41.420
In addition, this was calibrated with 2019
00:39:41.420 --> 00:39:44.230
and also in the most
recent fire season, working
00:39:44.230 --> 00:39:48.250
with CAL Fire to help guide
them when we get multiple fires
00:39:48.250 --> 00:39:50.420
which ones are the ones
we need to pay attention to,
00:39:50.420 --> 00:39:53.360
in our regard for this analysis
00:39:53.360 --> 00:39:56.550
we're interested in identifying
for all those simulations
00:39:56.550 --> 00:39:59.520
which ones are more likely to be realized
00:39:59.520 --> 00:40:01.180
for a longer duration.
00:40:01.180 --> 00:40:03.453
And then also the impacts or consequence.
00:40:06.220 --> 00:40:09.510
Again, impact analysis was
conducted with the fire modeling
00:40:09.510 --> 00:40:12.477
on the back end, focusing
on building footprints
00:40:12.477 --> 00:40:15.328
and population count important to note
00:40:15.328 --> 00:40:17.683
we use the term impacted.
00:40:19.310 --> 00:40:21.550
We are unable to show the level of damage
00:40:21.550 --> 00:40:24.550
or loss at this time, whether
it was partially impacted
00:40:24.550 --> 00:40:27.380
or totally destroyed that can be done
00:40:27.380 --> 00:40:29.090
and has been done in the past
00:40:29.090 --> 00:40:31.240
but it requires very detailed data
00:40:31.240 --> 00:40:35.767
often assessor data that
defines the building time
00:40:36.730 --> 00:40:39.290
that level of resiliency of the structures
00:40:39.290 --> 00:40:42.360
and defensible space
on a California Statewide
00:40:42.360 --> 00:40:44.410
that data is not readily available
00:40:44.410 --> 00:40:46.000
or requires significant effort
00:40:46.000 --> 00:40:48.996
and cost to develop that
if it was to be compiled
00:40:48.996 --> 00:40:51.720
we did evaluate a commercial source
00:40:51.720 --> 00:40:54.490
that provides that level of detailed data,
00:40:54.490 --> 00:40:57.920
but in analysis it was not
consistent across the state.
00:40:57.920 --> 00:41:02.830
We feel with led to results
that could not be comparative
00:41:02.830 --> 00:41:04.730
can be applied together
00:41:04.730 --> 00:41:06.660
because of their lack of consistency.
00:41:06.660 --> 00:41:09.160
We decided to not use it.
00:41:09.160 --> 00:41:10.370
The other metric or impact
00:41:10.370 --> 00:41:12.640
of (indistinct) firesides potential of course
00:41:13.589 --> 00:41:15.960
is the potential acres burn from that fire.
00:41:15.960 --> 00:41:17.810
And that was also calculated.
00:41:17.810 --> 00:41:21.683
In addition with all the other impacts.
00:41:22.560 --> 00:41:26.173
Quick example to show
you the type of analysis
00:41:26.173 --> 00:41:30.667
this fire prediction is showing
you the population impacted.
00:41:30.667 --> 00:41:34.190
And we know the distance
from that ignition point
00:41:34.190 --> 00:41:35.990
we know the time of impact
00:41:35.990 --> 00:41:39.350
and the time to impact all
that information is captured
00:41:39.350 --> 00:41:41.150
with every one of these simulations.
00:41:44.300 --> 00:41:46.240
So we conducted this analysis
00:41:46.240 --> 00:41:50.610
for all the damage incidents
for each PSPS event.
00:41:50.610 --> 00:41:52.800
Of course, not only do we spatially see them
00:41:52.800 --> 00:41:54.267
like I showed you in the examples
00:41:54.267 --> 00:41:57.380
but we obviously have a
database that identifies each
00:41:57.380 --> 00:41:59.570
of these simulations,
what their start time was.
00:41:59.570 --> 00:42:03.000
And then all of the metrics
that were calculated about that
00:42:04.080 --> 00:42:07.142
the impact metrics such as acres bear burned
00:42:07.142 --> 00:42:10.530
but also metrics in the
first hour, second hour.
00:42:10.530 --> 00:42:13.340
So we can look at the
initial attack assessment
00:42:13.340 --> 00:42:15.030
buildings population that you notice
00:42:15.030 --> 00:42:18.830
these different columns
are color coded based on
00:42:18.830 --> 00:42:21.540
their classification of where they fit.
00:42:21.540 --> 00:42:24.800
The IAA and a little bit to the right,
00:42:24.800 --> 00:42:27.700
you'll see that combination of one to five,
00:42:27.700 --> 00:42:32.030
again one being fires, if
they occurred that were very
00:42:32.030 --> 00:42:36.480
likely to be contained and
suppressed during initial attack.
00:42:36.480 --> 00:42:38.750
So they're not gonna get very far.
00:42:38.750 --> 00:42:40.770
And five are ones that independent
00:42:40.770 --> 00:42:43.210
of the number of resources applied to it.
00:42:43.210 --> 00:42:45.410
It's growth and spread is
going to be so significant.
00:42:45.410 --> 00:42:47.640
It's gonna escape initial attack.
00:42:47.640 --> 00:42:49.440
This has been significantly validated
00:42:49.440 --> 00:42:51.550
during the 2020 fire season
00:42:51.550 --> 00:42:53.660
that helps substantiate
the use of this metric
00:42:53.660 --> 00:42:56.653
and the CPUC project was really appropriate.
00:42:58.790 --> 00:43:01.983
So now we have this list of simulations
00:43:01.983 --> 00:43:04.562
and the results quantifying impacts
00:43:04.562 --> 00:43:07.733
and consequence for all the damage incidents.
00:43:08.590 --> 00:43:11.143
Now we need a way of analyzing or evaluating.
00:43:11.990 --> 00:43:14.370
We looked at the different critique criteria
00:43:14.370 --> 00:43:16.220
population, building size
00:43:16.220 --> 00:43:18.455
in the initial attack
assessment to identify a list
00:43:18.455 --> 00:43:21.350
of what we call the most significant incident
00:43:23.275 --> 00:43:26.180
the IAA again, the initial attack assessment
00:43:26.180 --> 00:43:28.098
is a key indicator to help us again
00:43:28.098 --> 00:43:30.600
identify whether that's likely to go
00:43:30.600 --> 00:43:31.810
beyond initial attack.
00:43:31.810 --> 00:43:34.378
For each of the events, we identify
00:43:34.378 --> 00:43:37.470
a ranking of about the top five or 10.
00:43:37.470 --> 00:43:39.397
It's a number vary depending on that.
00:43:39.397 --> 00:43:40.930
And we use that looked
00:43:40.930 --> 00:43:42.680
at natural breaks in the data
00:43:43.820 --> 00:43:46.170
for those significant impacts,
00:43:46.170 --> 00:43:47.940
we took that next step to conduct those
00:43:47.940 --> 00:43:51.610
probabilistic fire simulations
and evaluate those results
00:43:51.610 --> 00:43:56.090
and then develop summaries
for each of those events.
00:43:56.090 --> 00:43:57.090
This is the same example
00:43:57.090 --> 00:43:59.650
you'll see many in a
minute that shows the type
00:43:59.650 --> 00:44:02.083
of results for the most significant fires.
00:44:03.130 --> 00:44:05.470
These incident, damage incidents.
00:44:05.470 --> 00:44:07.590
In this case, there was 10 extracted
00:44:07.590 --> 00:44:10.000
for this event, you know,
where are they account?
00:44:10.000 --> 00:44:11.890
There's the general summary by county
00:44:11.890 --> 00:44:13.530
and then what those metrics are.
00:44:13.530 --> 00:44:16.970
And you can see
consistently the ones that have
00:44:16.970 --> 00:44:18.774
the most significant impacts
00:44:18.774 --> 00:44:21.943
of an initial attack
assessment of five or four.
00:44:21.943 --> 00:44:23.670
But in this regard
00:44:23.670 --> 00:44:26.220
we see a good example of number nine there
00:44:26.220 --> 00:44:29.330
where there wasn't
significant significant spread,
00:44:29.330 --> 00:44:31.344
but due to the location of where it was
00:44:31.344 --> 00:44:33.413
with the terrain conditions
00:44:33.413 --> 00:44:36.251
and particularly relative to the population.
00:44:36.251 --> 00:44:39.520
Okay, and the buildings that people live in
00:44:39.520 --> 00:44:41.740
could still have pretty significant damages
00:44:41.740 --> 00:44:45.657
almost a thousand
population, 400 buildings, right?
00:44:45.657 --> 00:44:49.530
Although it's a small acreage
and didn't grow very fast.
00:44:49.530 --> 00:44:50.970
It's important to note that yes
00:44:50.970 --> 00:44:55.560
while large fires can cause
significant impacts, right?
00:44:55.560 --> 00:44:58.190
Small fires can also
cause significant impacts.
00:44:58.190 --> 00:45:00.470
And that's well understood in California
00:45:00.470 --> 00:45:01.820
because of the people living
00:45:01.820 --> 00:45:04.200
in the wild land, urban interface.
00:45:04.200 --> 00:45:05.450
Important to incorporate.
00:45:07.580 --> 00:45:10.980
In that, of course we can
also then summarize up
00:45:10.980 --> 00:45:15.370
this chart shows the
population in gold buildings
00:45:15.370 --> 00:45:18.430
in gray with the legend on the left.
00:45:18.430 --> 00:45:21.460
The total impacts for the
top 10 significant impacts
00:45:21.460 --> 00:45:22.820
for this particular event
00:45:22.820 --> 00:45:25.223
which was a PG&E event from October 9th.
00:45:26.100 --> 00:45:27.990
On the right-hand side, we see the legend
00:45:27.990 --> 00:45:31.680
for acres burned and that's represented
00:45:31.680 --> 00:45:34.623
by the green light and
the green line in the chart.
00:45:36.570 --> 00:45:39.000
So we see obviously a consistency,
00:45:39.000 --> 00:45:41.580
but if you look at number or a trend,
00:45:41.580 --> 00:45:44.623
but if you look at damage
incidents five and six,
00:45:46.010 --> 00:45:51.010
over a thousand people impacted well over 500
00:45:51.088 --> 00:45:54.190
yet the fires were very large.
00:45:54.190 --> 00:45:56.190
And so we see a lot of it
00:45:56.190 --> 00:45:59.420
at area didn't impact
people we see other ones
00:45:59.420 --> 00:46:02.970
like number two, where
there's a close association
00:46:02.970 --> 00:46:06.763
between large acreage and impacts.
00:46:08.910 --> 00:46:12.440
And again, as I mentioned,
map on the left hand side
00:46:12.440 --> 00:46:17.440
in blue shows the PSPS event boundaries
00:46:18.069 --> 00:46:21.510
i.e where lines were de energized.
00:46:21.510 --> 00:46:23.430
We see all the damage incident points.
00:46:23.430 --> 00:46:25.970
This one is about 116 on the left.
00:46:25.970 --> 00:46:27.663
And in this regard, we see 10
00:46:27.663 --> 00:46:31.150
that the most significant and those 160
00:46:31.150 --> 00:46:33.950
the other ones are showing a small gray dots.
00:46:33.950 --> 00:46:36.400
And they're numbered by their ranking.
00:46:36.400 --> 00:46:40.420
They're color coded by the IAA.
00:46:40.420 --> 00:46:45.070
So we can see which ones
are the extreme versus the low.
00:46:45.070 --> 00:46:47.200
And you see down in Sonoma number nine was
00:46:47.200 --> 00:46:50.420
that example I talked to,
which could be a small fire,
00:46:50.420 --> 00:46:51.950
but because of this location
00:46:51.950 --> 00:46:53.883
could cause significant impacts.
00:46:56.320 --> 00:46:58.220
These were the methods that we applied
00:46:58.220 --> 00:47:02.080
throughout the analysis,
lots of little details, of course
00:47:02.080 --> 00:47:04.910
difficult to go through all of those details
00:47:04.910 --> 00:47:07.450
but interested if you're interested,
00:47:07.450 --> 00:47:10.127
we can talk about that during Q&A.
00:47:10.127 --> 00:47:11.320
I'd like to move on now
00:47:11.320 --> 00:47:14.883
and really focus on results
for the rest of our time.
00:47:23.640 --> 00:47:28.520
This analysis was undertaken with each IOU
00:47:28.520 --> 00:47:30.963
for each of the individual PSPS events.
00:47:35.010 --> 00:47:36.800
PSPS events with a higher number
00:47:36.800 --> 00:47:39.360
of damage incidents were analyzed separately
00:47:39.360 --> 00:47:42.280
because of the volume
of data in other scenarios
00:47:42.280 --> 00:47:45.910
we had PSPS events with a
low number of damage incidents.
00:47:45.910 --> 00:47:47.840
They were analyzed individually
00:47:47.840 --> 00:47:49.270
each of those damaged incidents
00:47:49.270 --> 00:47:53.230
but grouped together for
summarizing and reporting results.
00:47:53.230 --> 00:47:56.310
And you'll see that as we
go through the examples
00:47:56.310 --> 00:47:59.130
let's have a look at PG and E first, right?
00:47:59.130 --> 00:48:00.580
Pacific gas and electric
00:48:00.580 --> 00:48:03.487
and the PSP events and the damage incidents.
00:48:05.680 --> 00:48:08.160
The simple table represents
the different events
00:48:08.160 --> 00:48:10.923
of the (indistinct) in 2019.
00:48:13.220 --> 00:48:15.100
The total number of damages reported
00:48:15.100 --> 00:48:17.630
from the field surveys is identified.
00:48:17.630 --> 00:48:20.410
In addition to the damages
00:48:20.410 --> 00:48:23.020
that were expected to ignite a fire
00:48:23.020 --> 00:48:26.000
not all damages were
expected to ignite a fire.
00:48:26.000 --> 00:48:27.020
We've highlighted here
00:48:27.020 --> 00:48:29.400
in the light yellow two different events
00:48:29.400 --> 00:48:32.600
the October 9th event
and the October 26th event.
00:48:32.600 --> 00:48:35.463
Those events of course
run often for multiple days.
00:48:38.320 --> 00:48:43.320
We can see in the 2019, we
had 193 damages reported
00:48:44.150 --> 00:48:47.120
but they estimated only 116 of those
00:48:48.020 --> 00:48:51.653
would cause arcing and
then probable to ignite a fire,
00:48:53.030 --> 00:48:57.680
October 26th we have 441,
a significant event, obviously
00:48:57.680 --> 00:49:01.890
and 422 of those would've ignited.
00:49:01.890 --> 00:49:04.430
So let's walk through those two big events
00:49:04.430 --> 00:49:06.920
'cause we analyze those separately.
00:49:06.920 --> 00:49:11.430
There's the October 9th
event, 193 damage assets
00:49:11.430 --> 00:49:14.070
116 were identified as potential.
00:49:14.070 --> 00:49:16.170
When we analyze the input data,
00:49:16.170 --> 00:49:18.580
we found that only 114 of those
00:49:18.580 --> 00:49:21.000
were located near a burnable fuel.
00:49:21.000 --> 00:49:23.960
And therefore we use to conduct the analysis.
00:49:23.960 --> 00:49:25.993
So we analyzed 114 of them.
00:49:29.640 --> 00:49:31.470
These maps to maps one on the left.
00:49:31.470 --> 00:49:32.420
One on the right, the one
00:49:32.420 --> 00:49:37.180
on the left shows the
location of the PSPS phases.
00:49:37.180 --> 00:49:40.120
During this particular
event on the ninth to the 12th
00:49:40.120 --> 00:49:42.680
blue represents phases one to three
00:49:42.680 --> 00:49:46.136
different phases of
implementation or de-energization .
00:49:46.136 --> 00:49:48.410
The phase four is the pink or purple.
00:49:48.410 --> 00:49:50.810
You'll see at the bottom of the map
00:49:50.810 --> 00:49:52.660
all the damage incidents were discovered
00:49:52.660 --> 00:49:54.630
as shown on the map on the right
00:49:54.630 --> 00:49:58.277
in areas of the phase, one
to three de-energization.
00:50:01.520 --> 00:50:06.520
Again 116 damages, 193, total,
00:50:06.580 --> 00:50:09.370
114 we analyzed a couple maps to just zoom
00:50:09.370 --> 00:50:12.160
in and show you the location of those.
00:50:12.160 --> 00:50:15.693
So you get a better
idea of a specific detail.
00:50:16.970 --> 00:50:19.870
Again, another map series
of maps that shows that
00:50:19.870 --> 00:50:21.350
and you start to see some pattern
00:50:21.350 --> 00:50:26.040
along what appears to be the same lines.
00:50:26.040 --> 00:50:28.880
If you look at the map
on the left, you see points
00:50:28.880 --> 00:50:31.327
in the bottom left-hand
corner between number 60
00:50:31.327 --> 00:50:34.091
and 58 numbering representing the ranking
00:50:34.091 --> 00:50:36.713
that came out and the damage incidents.
00:50:37.980 --> 00:50:40.130
Now on the right-hand side, you see a similar
00:50:40.130 --> 00:50:42.747
pattern where you get
this clustering occurring.
00:50:44.520 --> 00:50:47.230
We go away and we
run all our fire simulations
00:50:47.230 --> 00:50:50.970
counting up all our values
to our quality assurance.
00:50:50.970 --> 00:50:53.210
Sometimes that went back to adjust the fuels
00:50:53.210 --> 00:50:55.480
a little bit based on results
00:50:55.480 --> 00:50:58.210
as we analyzed the results
of all these individual ones.
00:50:58.210 --> 00:51:00.660
Ultimately, in this regard, we have a list
00:51:00.660 --> 00:51:03.148
of 114 different simulations
00:51:03.148 --> 00:51:06.573
with population impact, the building's impact
00:51:06.573 --> 00:51:09.710
what we call the firesides
potential acres burn
00:51:09.710 --> 00:51:13.280
and then a ranking, and then initial attack
00:51:13.280 --> 00:51:18.280
from this listing, we selected a top 10
00:51:18.720 --> 00:51:21.920
to identify that and there
was some subjective criteria
00:51:21.920 --> 00:51:23.920
by a variety of specialists
00:51:23.920 --> 00:51:26.230
on identifying how many of these
00:51:26.230 --> 00:51:28.643
will we then go for further analysis.
00:51:32.205 --> 00:51:33.210
This event
00:51:33.210 --> 00:51:35.910
these happened to be the
examples I showed you earlier
00:51:36.790 --> 00:51:39.140
the summary of the 10 most significant
00:51:39.140 --> 00:51:42.574
and then a chart
representing that distribution
00:51:42.574 --> 00:51:45.403
for population buildings and acres burn.
00:51:47.630 --> 00:51:51.490
If you've seen this let's
move on and identify
00:51:51.490 --> 00:51:53.293
where are they actually located?
00:51:55.670 --> 00:51:58.550
And this map shows you
all of the damage incidents
00:51:59.760 --> 00:52:02.633
with the potential population
impacts they would occur.
00:52:03.610 --> 00:52:06.380
Once you have this data,
you can slice and dice it
00:52:06.380 --> 00:52:09.150
to summarize it in many different ways.
00:52:09.150 --> 00:52:10.540
The reason I show you this chart
00:52:10.540 --> 00:52:12.343
is to simply demonstrate that.
00:52:15.030 --> 00:52:16.040
Let's jump in and look
00:52:16.040 --> 00:52:17.780
at some of the most significant ones
00:52:17.780 --> 00:52:20.860
from this October 9th to 12th PG&E event
00:52:22.170 --> 00:52:26.490
the number one was in a
North end up near Redding.
00:52:26.490 --> 00:52:28.980
We provided a review of that.
00:52:28.980 --> 00:52:31.470
You know, the initial attack was five
00:52:31.470 --> 00:52:33.420
over 2000 buildings population
00:52:33.420 --> 00:52:38.420
about 3,300 similar
simulation I showed you before,
00:52:40.430 --> 00:52:43.350
again buildings as reference or showed black.
00:52:43.350 --> 00:52:45.040
There's some terrain on there.
00:52:45.040 --> 00:52:48.930
And then we see the hourly
parameters for the spread.
00:52:48.930 --> 00:52:52.230
This in fact was the animated
example I showed you earlier
00:52:53.760 --> 00:52:56.593
when we talked about
wildfire analyst as the product.
00:52:58.560 --> 00:53:01.814
And here we see the probabilistic simulation
00:53:01.814 --> 00:53:04.190
with the deterministic.
00:53:04.190 --> 00:53:07.020
Again, set values for analysis
00:53:07.020 --> 00:53:11.550
opposed to variability
in input overlaid in black
00:53:11.550 --> 00:53:14.350
you can see the
deterministic aligns pretty good
00:53:15.583 --> 00:53:20.583
with the top 40 to a hundred percent.
00:53:23.280 --> 00:53:26.260
Again, representing over
a hundred simulations
00:53:26.260 --> 00:53:28.560
conducted just to
determine the probabilistic.
00:53:30.540 --> 00:53:32.920
We also calculate of course, a whole suite
00:53:32.920 --> 00:53:36.860
of metrics related to the
fire behavior charts on the left
00:53:36.860 --> 00:53:40.210
show rate of spread by different categories
00:53:40.210 --> 00:53:43.880
throughout those 24
hours and then flame life.
00:53:43.880 --> 00:53:45.880
So if you look at the rate of spread,
00:53:45.880 --> 00:53:48.410
we're seeing significant very high rates
00:53:48.410 --> 00:53:52.230
of spread 50 to 150 change per hour.
00:53:52.230 --> 00:53:55.610
That's the nomenclature used the Castro rate
00:53:55.610 --> 00:53:57.603
of spread a chain of 66 feet.
00:53:58.540 --> 00:54:02.103
It comes back to a
traditional forestry methods.
00:54:02.103 --> 00:54:05.610
So we're seeing that
orange in the chart represent.
00:54:05.610 --> 00:54:09.030
And those first six hours,
we got really significant spread
00:54:09.030 --> 00:54:10.790
and they pretty well continues on.
00:54:10.790 --> 00:54:14.993
And then we see another real
burst between hours 15 and 18.
00:54:17.170 --> 00:54:18.580
We don't see much blow spread
00:54:18.580 --> 00:54:22.070
until we get into the
back end of that 24 hours.
00:54:22.070 --> 00:54:25.930
Flame lengths represents
a level of intensity of the fire
00:54:25.930 --> 00:54:27.210
not just how fast of a move
00:54:27.210 --> 00:54:29.040
but how intense it will be.
00:54:29.040 --> 00:54:31.594
Then here we see pretty moderate
00:54:31.594 --> 00:54:35.610
pretty decent flame lengths
throughout the top end
00:54:35.610 --> 00:54:39.380
of this consistently again,
a lot in the beginning
00:54:39.380 --> 00:54:41.810
the first nine hours dies off a little bit
00:54:41.810 --> 00:54:45.523
in intensity and then kicks
back up an hour 15 through 21.
00:54:46.760 --> 00:54:48.020
We also did two charts
00:54:48.020 --> 00:54:50.150
on the right represent the combination
00:54:50.150 --> 00:54:53.697
of comparing the closest weather station
00:54:55.750 --> 00:54:58.140
to that particular damage ignition location
00:54:58.140 --> 00:55:00.274
and the weather prediction data.
00:55:00.274 --> 00:55:04.430
We did a lot of analysis and
seeing how close they were.
00:55:04.430 --> 00:55:07.690
This is where San Jose state
university was Craig climax
00:55:07.690 --> 00:55:10.460
and Scott pretty kicked
in to help us understand
00:55:10.460 --> 00:55:12.800
and adjust some of that prediction data.
00:55:12.800 --> 00:55:14.590
If there was weather stations nearby
00:55:14.590 --> 00:55:15.890
that showed some variance.
00:55:24.760 --> 00:55:26.610
Let's look at this damage instance, excuse me
00:55:26.610 --> 00:55:27.830
let's look at this damage incident
00:55:27.830 --> 00:55:30.050
number three, this is one that's more
00:55:30.050 --> 00:55:32.370
in a wild land, urban interface area.
00:55:32.370 --> 00:55:34.820
So isn't out in the wild lands
00:55:34.820 --> 00:55:37.630
and then burned into urban
it's one that started directly
00:55:37.630 --> 00:55:39.550
in a wild land, urban interface.
00:55:39.550 --> 00:55:42.040
This is a description provided, you know
00:55:42.040 --> 00:55:43.390
we still get anything with
00:55:43.390 --> 00:55:44.901
over 900 of bill Higgs is, you know
00:55:44.901 --> 00:55:49.060
it's still significant damage
over 2200 people impacted.
00:55:49.060 --> 00:55:50.780
Yes, size of it it's not very big.
00:55:50.780 --> 00:55:52.970
It's only about 8,000 acres for fire
00:55:54.100 --> 00:55:56.230
considering some of the
monumental size of fires
00:55:56.230 --> 00:55:58.288
particularly in the past year.
00:55:58.288 --> 00:56:02.020
It seems the scope is small,
but this really drives home.
00:56:02.020 --> 00:56:04.700
The fact that small fires, if they started
00:56:04.700 --> 00:56:07.900
in the wrong place could
still have significant impacts.
00:56:07.900 --> 00:56:09.640
As we see this particular fire
00:56:09.640 --> 00:56:12.950
with this deterministic
run started by the hole
00:56:12.950 --> 00:56:14.330
in the top of the green at the top
00:56:14.330 --> 00:56:18.654
of the screen could spread
on adjacent to this community.
00:56:18.654 --> 00:56:23.654
It spread very significantly
into that community.
00:56:24.170 --> 00:56:27.310
You'll notice that there's
not a ton of overlap
00:56:27.310 --> 00:56:30.130
between the fire spread perimeters
00:56:30.130 --> 00:56:32.180
and those dense urban areas represented
00:56:32.180 --> 00:56:34.613
by the dark black fillings.
00:56:36.040 --> 00:56:37.550
We did a lot of calibration
00:56:37.550 --> 00:56:39.600
on what we call those urban encroachment.
00:56:47.010 --> 00:56:49.810
So we made sure we
weren't overestimating those
00:56:49.810 --> 00:56:50.833
potential impacts.
00:56:59.480 --> 00:57:01.523
Here's the probabilistic mode for that.
00:57:02.410 --> 00:57:04.360
And it shows the potential
of where this could
00:57:04.360 --> 00:57:07.313
have gone beyond where
we did determine holistically.
00:57:08.410 --> 00:57:10.028
I'm gonna stop here for a moment
00:57:10.028 --> 00:57:12.630
and just make sure we can catch
00:57:12.630 --> 00:57:17.000
up a little bit really
important that we see some
00:57:17.000 --> 00:57:19.633
of these in concert with
my comments about them.
00:57:21.320 --> 00:57:23.226
So if it's appropriate, I'll just give it
00:57:23.226 --> 00:57:26.620
about a 30 second to 45 second break last.
00:57:26.620 --> 00:57:28.570
Maybe we grab a glass of water as well.
00:58:02.410 --> 00:58:06.330
Okay, I had a glass of
water rejuvenated my voice
00:58:06.330 --> 00:58:09.230
and hopefully allow these
slides to catch up a little bit.
00:58:10.150 --> 00:58:12.670
So from this October 9th to 12th event
00:58:12.670 --> 00:58:17.080
let's look at one more example
of the significant impacts.
00:58:17.080 --> 00:58:19.283
This one here is North of Sacramento,
00:58:20.340 --> 00:58:21.710
maybe a little close to home,
00:58:21.710 --> 00:58:25.490
given some of the fires
around Vacaville this year.
00:58:25.490 --> 00:58:29.663
Again, another example of a WUI type fire.
00:58:32.410 --> 00:58:35.190
We can say this isn't
with the black buildings
00:58:35.190 --> 00:58:37.260
you can see this as right in the middle level
00:58:37.260 --> 00:58:39.833
typical wild land, urban interface area.
00:58:43.110 --> 00:58:45.890
And we see not only the
impacts of those in the WUI
00:58:48.120 --> 00:58:50.993
but also as it impedes on these subdivisions,
00:58:52.260 --> 00:58:55.561
again a small fire in the
wrong place at the wrong time
00:58:55.561 --> 00:58:59.753
still leading the potential
significant consequence.
00:59:02.798 --> 00:59:04.430
There is a probability
00:59:04.430 --> 00:59:07.387
of that probabilistic run of that event.
00:59:07.387 --> 00:59:09.810
And when see again, it aligns really closely
00:59:10.773 --> 00:59:12.607
with the 40 to a hundred percent
00:59:12.607 --> 00:59:16.720
but we also see the
potential of that fire to go well
00:59:16.720 --> 00:59:19.660
beyond if those
conditions varied a little bit.
00:59:19.660 --> 00:59:21.933
That's the beauty of the probabilistic mode.
00:59:26.940 --> 00:59:31.100
Now to sum up the PGE analysis, let's look at
00:59:31.100 --> 00:59:34.923
the most significant events
during the 2019 fire season.
00:59:36.227 --> 00:59:39.270
And that was the PSPS events that occurred
00:59:39.270 --> 00:59:41.903
from October 26 to October 29th.
00:59:45.075 --> 00:59:47.290
In this regard, we saw a significant amount
00:59:47.290 --> 00:59:50.113
of damage incident, 441 reported.
00:59:51.920 --> 00:59:53.823
Once we analyzed them, we found 422
00:59:55.890 --> 00:59:58.390
had the potential to
ignite a fire through arcing.
00:59:59.500 --> 01:00:02.750
And in fact, they were
also within the timeframes
01:00:02.750 --> 01:00:06.493
of the PSPS event and
close enough to burnable fuels.
01:00:07.330 --> 01:00:10.710
So we use those 422 simulations
01:00:10.710 --> 01:00:13.173
or damage incidents to
conduct those simulations.
01:00:18.680 --> 01:00:20.990
The next map shows all those locations.
01:00:20.990 --> 01:00:23.150
We did not label them with numbers
01:00:23.150 --> 01:00:24.750
because there's too many points.
01:00:25.810 --> 01:00:26.800
And if you look closely
01:00:26.800 --> 01:00:29.220
you'll see little dash lines that show some
01:00:29.220 --> 01:00:32.020
of the cluster of points that occurred
01:00:32.020 --> 01:00:36.010
in certain areas throughout
those PSPS event boundaries
01:00:36.010 --> 01:00:38.053
within the PGE service territory.
01:00:41.270 --> 01:00:43.780
Interesting you'll notice some of those
01:00:43.780 --> 01:00:47.700
damage incidents are not within PSPS events.
01:00:47.700 --> 01:00:50.940
These obviously with some
of the ones not modeled,
01:00:50.940 --> 01:00:53.260
there's a dot the upper left in particularly
01:00:53.260 --> 01:00:54.240
in the bottom, right
01:00:54.240 --> 01:00:56.250
you'll see a little cluster of points.
01:00:56.250 --> 01:00:58.550
So when they surveyed, they found locations
01:00:58.550 --> 01:01:01.320
damages that occurred, although they did not
01:01:01.320 --> 01:01:02.943
de-energize those lines.
01:01:08.290 --> 01:01:10.710
Series of maps that zoom in and shows you
01:01:10.710 --> 01:01:12.703
some of those clustering that occur.
01:01:15.210 --> 01:01:19.710
Some of those appear to
be a long different power lines
01:01:19.710 --> 01:01:21.770
some of them are clustered together,
01:01:21.770 --> 01:01:26.770
again when we zoom in the
numbering represents their rank.
01:01:26.870 --> 01:01:28.443
Once we did all the analysis.
01:01:30.770 --> 01:01:32.770
I just walk through a few more of those,
01:01:32.770 --> 01:01:34.010
the second example shots
01:01:34.010 --> 01:01:36.283
of rail clustering up near Calistoga.
01:01:39.010 --> 01:01:40.570
Of course, unfortunately in 2020
01:01:40.570 --> 01:01:45.040
we know the significance
of fires started in that area.
01:01:45.040 --> 01:01:48.563
Not necessarily from IOU damages,
01:01:49.430 --> 01:01:52.790
just fires in that area and
their possibility for damage.
01:01:52.790 --> 01:01:55.210
And of course, with the Tufts fire in 2016
01:01:55.210 --> 01:01:58.380
we saw that significant impact on Santa Rosa
01:01:58.380 --> 01:02:00.163
and then a similar fire this year.
01:02:02.760 --> 01:02:05.270
Again, third map shows
more detailed clustering
01:02:05.270 --> 01:02:10.270
of those locations and the fourth map
01:02:10.670 --> 01:02:13.420
the blue areas representing
the PSPS event boundaries
01:02:21.010 --> 01:02:25.990
over 422 simulations, multiple times run
01:02:25.990 --> 01:02:27.490
through quality assurance on all of them
01:02:27.490 --> 01:02:30.420
looking at fuels and impacting patients.
01:02:30.420 --> 01:02:34.000
Here's the table that
shows the top half of those.
01:02:34.000 --> 01:02:35.430
Of course there's many more.
01:02:35.430 --> 01:02:36.310
Most importantly
01:02:36.310 --> 01:02:38.570
the yellow though is represented the top 10
01:02:40.425 --> 01:02:43.910
and notice in this particular
table under included
01:02:43.910 --> 01:02:45.407
there's a new column that says included
01:02:45.407 --> 01:02:46.813
and it says clusters.
01:02:47.840 --> 01:02:49.760
As we saw in a damage incidents
01:02:49.760 --> 01:02:52.460
there was cluster of points occurring
01:02:52.460 --> 01:02:55.083
a lot of those at the same ignition time.
01:02:55.980 --> 01:02:58.780
So we needed to evaluate what we count
01:02:58.780 --> 01:03:03.163
all of those and what we
consider them as significant fires.
01:03:04.893 --> 01:03:08.580
And we found a whole bunch
of those in this particular table
01:03:08.580 --> 01:03:10.630
from row 11 to 29
01:03:14.360 --> 01:03:16.690
points that we would not consider
01:03:16.690 --> 01:03:18.980
because they would be superseded
01:03:18.980 --> 01:03:22.083
by the top 10 points and would be redundant.
01:03:23.330 --> 01:03:25.290
Let me show you a map
of what that looks like.
01:03:25.290 --> 01:03:26.740
So we better understand that.
01:03:27.880 --> 01:03:30.800
We find this cluster of incident,
01:03:30.800 --> 01:03:33.030
damage incident locations
01:03:33.030 --> 01:03:34.913
consistently in the large events.
01:03:35.940 --> 01:03:40.610
So here we show a fire spread
or ignition point number one
01:03:40.610 --> 01:03:43.310
and then we show other
damage incidents that occurred.
01:03:44.270 --> 01:03:47.050
And they're ranked by where they were based
01:03:47.050 --> 01:03:50.390
on the consequences that involved,
01:03:50.390 --> 01:03:54.840
in this scenario dabbing into now 11 and 12
01:03:55.730 --> 01:03:58.733
are clearly in the path of the fire spread
01:03:58.733 --> 01:04:00.623
for instance one.
01:04:05.590 --> 01:04:10.440
So we do not incorporate
or include the simulation
01:04:10.440 --> 01:04:14.350
for one or 11 based on
an analysis of that spread.
01:04:14.350 --> 01:04:16.110
That would be superseded if even
01:04:16.110 --> 01:04:18.353
if they ignited at the same time.
01:04:36.800 --> 01:04:39.492
Sorry for the break just get a drink of water
01:04:39.492 --> 01:04:41.490
and let us catch up a little bit.
01:04:41.490 --> 01:04:43.610
And this is really important
for these large events.
01:04:43.610 --> 01:04:46.180
We needed to make sure
we weren't double dipping
01:04:46.180 --> 01:04:48.740
if you will for damage incidents
01:04:48.740 --> 01:04:51.600
and recording a significant
what we know it was
01:04:51.600 --> 01:04:53.670
in the path of one that
would have started earlier
01:04:53.670 --> 01:04:56.600
at the same time and would
have been superseded and overrun.
01:04:56.600 --> 01:05:00.000
So the reality is we shouldn't consider it.
01:05:00.000 --> 01:05:02.750
We spent a lot of time
doing this for the large events.
01:05:04.540 --> 01:05:07.259
Ultimately, we came up with our list.
01:05:07.259 --> 01:05:12.259
In this case, it was 10 for
this October 26th to 29th event.
01:05:12.640 --> 01:05:15.560
With these, we see some
very significant large fires.
01:05:15.560 --> 01:05:18.323
The top one and Silvano
could be 50,000 acres.
01:05:21.470 --> 01:05:25.640
Over 13,000 population
potentially impacted in 24 hours.
01:05:25.640 --> 01:05:28.343
And about 7,000 buildings impacted.
01:05:29.360 --> 01:05:32.210
Again look at the initial attack index
01:05:32.210 --> 01:05:35.747
almost all of these are
fives and fours, a three
01:05:35.747 --> 01:05:39.670
and a couple twos to
accommodate significant impacts
01:05:39.670 --> 01:05:41.273
but for much smaller fires.
01:05:43.970 --> 01:05:45.430
The chart on the right again
01:05:45.430 --> 01:05:48.390
shows our distribution of
population in gold buildings
01:05:48.390 --> 01:05:51.620
and gray in acres burn
legend for acres burned
01:05:51.620 --> 01:05:56.120
on the right hand side for the green line,
01:05:56.120 --> 01:05:57.500
legend for total impacts,
01:05:57.500 --> 01:05:59.610
population of buildings on the left.
01:05:59.610 --> 01:06:01.810
Again, we see pretty good adherence.
01:06:01.810 --> 01:06:04.320
If you look of course at number nine
01:06:04.320 --> 01:06:05.920
on that chart, what do we see?
01:06:05.920 --> 01:06:08.860
The acres burn actually dips down really low.
01:06:08.860 --> 01:06:11.320
So we have a scenario where it's a small fire
01:06:11.320 --> 01:06:13.450
but significant impacts.
01:06:13.450 --> 01:06:18.133
Although the IAA initial
attack assessment is moderate.
01:06:24.500 --> 01:06:26.140
Where are those top 10?
01:06:26.140 --> 01:06:29.038
This map shows all the 400
01:06:29.038 --> 01:06:32.490
plus damage incidents
and those light gray dots
01:06:33.470 --> 01:06:35.800
and then in a larger colored dots,
01:06:35.800 --> 01:06:38.660
colored by IAA shows the most significant
01:06:39.970 --> 01:06:43.020
we see number one and
three are close together.
01:06:43.020 --> 01:06:45.440
Three was included because it was not
01:06:45.440 --> 01:06:47.380
in the direct path of one.
01:06:47.380 --> 01:06:50.870
If you remember looking at
that cluster spread a minute ago
01:06:50.870 --> 01:06:52.960
two and six, same scenario
01:06:52.960 --> 01:06:55.653
the other one's a little
more randomly located.
01:06:59.100 --> 01:07:00.600
Let's go ahead and look at
01:07:03.320 --> 01:07:08.310
some of the fire spread results, spatially
01:07:08.310 --> 01:07:10.953
with respect to those key incidents.
01:07:14.718 --> 01:07:17.100
You notice in some of the
narratives that are included
01:07:17.100 --> 01:07:21.360
in here, we actually compare
it to previous fire history.
01:07:21.360 --> 01:07:22.800
Very important.
01:07:22.800 --> 01:07:24.420
We always see that was accommodated
01:07:24.420 --> 01:07:27.450
through the fuels update,
of course, as inputs
01:07:27.450 --> 01:07:29.117
but they're good examples.
01:07:29.117 --> 01:07:31.990
We see history revisiting itself.
01:07:31.990 --> 01:07:34.690
Unfortunately like we did
up near Calistoga this year.
01:07:39.380 --> 01:07:41.520
Here's number one on the hip chart.
01:07:41.520 --> 01:07:44.000
Unfortunately, we're
saying the potential here
01:07:44.000 --> 01:07:47.130
for this to back to communities, again,
01:07:47.130 --> 01:07:48.580
building footprints in black,
01:07:49.830 --> 01:07:52.540
we're seeing the first community being
01:07:52.540 --> 01:07:55.840
impacted relatively
quickly in the first 12 hours.
01:07:55.840 --> 01:07:58.370
And then the second one at the back end of it
01:07:58.370 --> 01:08:02.040
you're seeing that
overlap into those buildings
01:08:02.040 --> 01:08:04.800
to represent the urban encroachment
01:08:09.020 --> 01:08:11.650
which was relatively conservative.
01:08:11.650 --> 01:08:13.750
Again, encroachment into those
01:08:13.750 --> 01:08:16.840
urban areas to calculate impacts
01:08:16.840 --> 01:08:19.960
uses a combination of
the density of the building
01:08:19.960 --> 01:08:23.830
and the fuel load around that density.
01:08:23.830 --> 01:08:26.710
Significant fuel loads have
lead to more severe fires.
01:08:26.710 --> 01:08:29.480
It would be greater
spread into the urban area
01:08:30.980 --> 01:08:34.910
less fuel load, very fuel loads high density.
01:08:34.910 --> 01:08:37.250
You're not gonna get much spread at all.
01:08:37.250 --> 01:08:40.003
It's a very important component
of the impact analysis.
01:08:40.960 --> 01:08:45.960
Here's the probabilistic run
for that community on the right
01:08:46.260 --> 01:08:47.400
to an 80, a hundred percent
01:08:47.400 --> 01:08:51.004
it's gonna get impacted
very quickly on this fire
01:08:51.004 --> 01:08:54.390
less probability for the Southern community.
01:08:54.390 --> 01:08:57.740
Although it's still comes
in between 40 and 60.
01:08:57.740 --> 01:08:58.900
Now the left-hand side
01:08:58.900 --> 01:09:01.330
the West end Northwest
part of that community.
01:09:01.330 --> 01:09:03.823
We talking 60, 80% chance it's gonna hit.
01:09:09.330 --> 01:09:11.590
The same fire behavior on weather charts
01:09:12.490 --> 01:09:14.440
on the left to look at the rate of spread.
01:09:14.440 --> 01:09:16.300
We're seeing extreme rates of spread
01:09:17.390 --> 01:09:20.410
very high rates of spread rate from the gecko
01:09:20.410 --> 01:09:25.340
particularly an hours 15 through 20.
01:09:25.340 --> 01:09:28.840
We're seeing very extreme
rates of spread takeover.
01:09:28.840 --> 01:09:31.560
As we saw in the maps, climb length,
01:09:31.560 --> 01:09:34.492
moderate intensity about hour nine
01:09:34.492 --> 01:09:37.100
it kicks in to higher intensity.
01:09:37.100 --> 01:09:39.003
So we see those significant impacts.
01:09:39.900 --> 01:09:41.560
Weather charts show the combination
01:09:41.560 --> 01:09:45.430
of the predicted any observed data,
01:09:45.430 --> 01:09:47.320
which we considered of course,
01:09:47.320 --> 01:09:49.793
in determining which weather data to use.
01:09:51.410 --> 01:09:53.703
Let's have a look at incident number two,
01:09:55.410 --> 01:09:58.510
again, significant first
one was 50,000 acres.
01:09:58.510 --> 01:10:02.680
This is about 37,000
slight reduction, you know
01:10:02.680 --> 01:10:07.100
13,000 to that's like to
about 5,000 buildings.
01:10:07.100 --> 01:10:09.130
It's still a significant population.
01:10:09.130 --> 01:10:11.350
Again, the alignment of how many buildings
01:10:11.350 --> 01:10:13.030
and population is gonna be based
01:10:13.030 --> 01:10:17.270
on the density of that type of building.
01:10:17.270 --> 01:10:20.010
High rises are gonna, areas that are impacted
01:10:20.010 --> 01:10:22.300
are gonna have obviously a lot of population
01:10:22.300 --> 01:10:23.650
but only one building.
01:10:23.650 --> 01:10:25.380
And so they don't always align.
01:10:25.380 --> 01:10:29.673
It depends on what type
of building it's impacting.
01:10:33.760 --> 01:10:37.401
This one we see it go
across this fire, actually spread
01:10:37.401 --> 01:10:42.340
across water and actually
spot across and continue.
01:10:42.340 --> 01:10:44.700
We see lots of urban interface here, again
01:10:45.690 --> 01:10:49.740
as expected, pretty
linear winds from the North.
01:10:49.740 --> 01:10:51.593
And we see this one really take off.
01:10:55.810 --> 01:10:58.850
I'll just go back to that slide for a minute.
01:10:58.850 --> 01:11:01.250
Realize you're probably catching up a little
01:11:01.250 --> 01:11:05.090
notice in the West hand
side, the two communities
01:11:05.090 --> 01:11:07.637
the impact into those and the spread
01:11:07.637 --> 01:11:10.623
and particularly, in the Southeast part
01:11:10.623 --> 01:11:14.210
in the later hours of this fire simulation,
01:11:14.210 --> 01:11:17.070
we see significant encroachment.
01:11:17.070 --> 01:11:19.700
Why, because of the fuel
load surrounding those
01:11:19.700 --> 01:11:21.373
buildings were very significant.
01:11:22.410 --> 01:11:25.953
And so we get greater
encroachments and possible impacts.
01:11:29.160 --> 01:11:33.493
If we look at the
consequence, similar patterns,
01:11:34.440 --> 01:11:37.540
again, the deterministic lines up very well
01:11:37.540 --> 01:11:40.643
at the top 60 to 80% of probability.
01:11:47.657 --> 01:11:50.260
And I stop here just for a few minutes
01:11:50.260 --> 01:11:53.110
make sure that we're caught
up, get myself another drink.
01:12:18.340 --> 01:12:19.970
Let's jump ahead to damage incident
01:12:19.970 --> 01:12:21.603
number four, on a number of lists here
01:12:21.603 --> 01:12:24.430
this one has slightly
different characteristics.
01:12:24.430 --> 01:12:29.430
Yet again, potential impacts
more closer to the Bay area.
01:12:33.180 --> 01:12:36.410
Again, lots of wild land urban interface.
01:12:36.410 --> 01:12:39.500
Two key communities to
the North and the South.
01:12:39.500 --> 01:12:41.670
We see significant encroachment on the WUI
01:12:41.670 --> 01:12:44.370
because of the adjacent fuel loads
01:12:48.350 --> 01:12:49.870
in community in the Northwest.
01:12:49.870 --> 01:12:51.650
We see a real nice spread
01:12:51.650 --> 01:12:54.760
into that representing the fire behavior.
01:12:54.760 --> 01:12:56.810
It could be very damaging, unfortunately.
01:12:58.050 --> 01:13:00.890
And then in the South, we see because of the,
01:13:00.890 --> 01:13:02.637
again proximity of different fuel loads
01:13:02.637 --> 01:13:06.373
and those WUI areas, significant damages.
01:13:08.510 --> 01:13:11.680
Again, depends where the fire's located.
01:13:11.680 --> 01:13:13.140
We look at the probabilistic map.
01:13:13.140 --> 01:13:16.821
We see consider an
opportunity in the last 20%
01:13:16.821 --> 01:13:20.130
for spread but not a lot of impacts.
01:13:20.130 --> 01:13:22.730
Unfortunately, most of the
impacts hit pretty early on
01:13:22.730 --> 01:13:24.030
with those two communities
01:13:26.100 --> 01:13:29.813
and both its deterministic
and probabilistic capture that.
01:13:32.600 --> 01:13:36.793
Let's summarize up the
different advanced for PG and E.
01:13:41.560 --> 01:13:44.157
This table here shows you, of course
01:13:44.157 --> 01:13:47.634
the different events that occurred in 2019.
01:13:47.634 --> 01:13:49.850
Shows you the little damages reported
01:13:49.850 --> 01:13:51.840
and sounded in the field.
01:13:51.840 --> 01:13:55.280
How many of those damages
would have ignited a fire?
01:13:55.280 --> 01:13:57.370
I.e did we model, and then
01:13:57.370 --> 01:13:59.460
on the right hand side with the red headers
01:13:59.460 --> 01:14:02.923
we actually see the primary
consequence about impact.
01:14:04.150 --> 01:14:06.413
Population buildings and total acres.
01:14:10.050 --> 01:14:12.500
If we look at the two big events on the ninth
01:14:12.500 --> 01:14:16.083
that we looked at and the 26th, significant
01:14:16.930 --> 01:14:19.789
obviously the 26 being the primary event,
01:14:19.789 --> 01:14:22.900
large event with a potential of
01:14:22.900 --> 01:14:24.393
over 3 million acres burned.
01:14:27.710 --> 01:14:30.440
While undertaking this analysis
01:14:30.440 --> 01:14:35.440
in 2019 and early, 2020, we would look
01:14:35.880 --> 01:14:38.150
at a number is 3 million acres is being well.
01:14:38.150 --> 01:14:40.620
That's pretty extreme,
is that really possible?
01:14:40.620 --> 01:14:43.260
We haven't had a fire
season like that for awhile.
01:14:43.260 --> 01:14:47.780
Unfortunately, 2020 came in like a line
01:14:47.780 --> 01:14:49.850
and showed us it's quite possible
01:14:49.850 --> 01:14:53.610
on a very short timeframe as
we exceeded 4 million acres.
01:14:53.610 --> 01:14:57.163
So when you look
retroactively at these numbers,
01:14:59.950 --> 01:15:01.340
they're very sobering
01:15:01.340 --> 01:15:02.910
but we can really understand the potential
01:15:02.910 --> 01:15:05.880
of these events to occur from fires
01:15:05.880 --> 01:15:10.880
whether they're caused by
damage incidents on lines of it
01:15:12.280 --> 01:15:16.600
not de-energize or
primarily from other sources.
01:15:16.600 --> 01:15:20.430
As we know, electric utility
ignitions are relatively rare
01:15:20.430 --> 01:15:23.053
and about 8% only of all fires.
01:15:25.570 --> 01:15:28.823
That's a quick summary of PG and E.
01:15:29.927 --> 01:15:32.510
It was the majority of
the event analyzed simply
01:15:32.510 --> 01:15:34.493
because of the volume of
that particular fire season
01:15:34.493 --> 01:15:37.207
and the number of PSPS events.
01:15:37.207 --> 01:15:40.503
But let's finish up by
looking at the So Cal Edison
01:15:40.503 --> 01:15:42.740
and the SDG events,
because they did have events
01:15:42.740 --> 01:15:43.663
that did occur.
01:15:46.160 --> 01:15:47.420
Southern California Edison
01:15:47.420 --> 01:15:50.113
we see a significant
number of events compared
01:15:50.113 --> 01:15:54.610
to PSPS events compared to the other IOUs,
01:15:54.610 --> 01:15:59.340
many of which reported no
damages afterwards and therefore
01:15:59.340 --> 01:16:03.083
no simulations to see what
those damages might've done.
01:16:04.170 --> 01:16:06.530
And this could have been analysis.
01:16:06.530 --> 01:16:10.390
They're identified in gray and the black,
01:16:10.390 --> 01:16:14.747
we see where damages
were recorded after the event.
01:16:16.430 --> 01:16:17.770
And then how many of those
01:16:19.120 --> 01:16:22.980
like the PGE events late October seems to be
01:16:22.980 --> 01:16:25.630
the timeframe for a lot of these to hit
01:16:25.630 --> 01:16:27.650
and the biggest advance for so Cal Edison
01:16:27.650 --> 01:16:30.190
were on 21st through the 27th
01:16:30.190 --> 01:16:32.563
each representing multiple days.
01:16:35.350 --> 01:16:37.803
Let's look at these events.
01:16:39.360 --> 01:16:41.330
In this case because of the volume of events,
01:16:41.330 --> 01:16:44.774
weren't significant ie
only 64 damage incidents
01:16:44.774 --> 01:16:47.970
across all PSPS events.
01:16:47.970 --> 01:16:49.950
They were modeled individually
01:16:49.950 --> 01:16:52.541
but bundled together for summarizing.
01:16:52.541 --> 01:16:53.740
So you'll see
01:16:53.740 --> 01:16:57.240
we've done that here in
this, in the same example.
01:16:57.240 --> 01:17:00.540
Now this requires a good
analysis and comparison
01:17:00.540 --> 01:17:03.930
of those results when
evaluating significant impacts
01:17:03.930 --> 01:17:06.307
to look at the different times of events.
01:17:06.307 --> 01:17:09.560
But we actually saw similar damage incidents
01:17:09.560 --> 01:17:13.340
of potential fires across multiple events.
01:17:13.340 --> 01:17:15.710
You'll see what I mean as I go through this.
01:17:15.710 --> 01:17:19.260
55 were identified as a
potential to cause an ignition
01:17:19.260 --> 01:17:23.259
and of those 54 were found
within the PSPS boundaries
01:17:23.259 --> 01:17:28.259
and located near burnable
fuels that could ignite a fire
01:17:28.820 --> 01:17:30.920
and therefore use to conduct the analysis.
01:17:35.220 --> 01:17:39.600
Here's a map throughout
the FTE service territory
01:17:39.600 --> 01:17:41.347
we identified two clusters.
01:17:41.347 --> 01:17:44.736
I'll show you a detailed
maps of those clusters.
01:17:44.736 --> 01:17:47.160
Again, this is a little
different than the PG and E
01:17:47.160 --> 01:17:50.620
because it's not, it's
multiple events remodeling
01:17:50.620 --> 01:17:52.990
and you see the color coding of those areas.
01:17:52.990 --> 01:17:55.100
We'll zoom in to give
you a little more detail
01:17:55.100 --> 01:17:57.690
of those events, the one
in September, and then
01:17:57.690 --> 01:18:01.540
the ones throughout
October, the darker, the orange
01:18:01.540 --> 01:18:03.040
the closer to the end of October
01:18:03.040 --> 01:18:06.340
which are the events that
had most significant damages
01:18:06.340 --> 01:18:07.423
and then potential.
01:18:08.400 --> 01:18:10.210
If we zoom into that cluster
01:18:11.210 --> 01:18:13.760
what we start to see is
polygon boundaries that
01:18:13.760 --> 01:18:17.200
represent the events i.e the
lines that would be energized
01:18:18.070 --> 01:18:21.013
and these overlap between different events.
01:18:22.590 --> 01:18:27.590
So for example, all of
the colored polygon shows
01:18:28.620 --> 01:18:31.500
with a number one box represents an area
01:18:31.500 --> 01:18:36.100
where de-energization
occurred to two different events,
01:18:36.100 --> 01:18:39.690
October 24th and also on October 30th,
01:18:39.690 --> 01:18:44.053
two separate PSPS events
similar lines de-energize.
01:18:45.610 --> 01:18:47.280
You look at area four that shows
01:18:47.280 --> 01:18:50.777
that kind of tan, muddy color same thing
01:18:50.777 --> 01:18:54.130
de-energized during October 10th
01:18:54.130 --> 01:18:56.570
but also on October 30th.
01:18:56.570 --> 01:18:59.800
So we see a temple overview.
01:18:59.800 --> 01:19:03.840
You look in the center above area three
01:19:03.840 --> 01:19:05.880
we see three different events
01:19:05.880 --> 01:19:08.320
or sorry in the Southwest that's
01:19:08.320 --> 01:19:12.160
a steel blue color area, two
01:19:12.160 --> 01:19:14.439
we see three different events.
01:19:14.439 --> 01:19:16.050
De-energization from those damage incidents
01:19:16.050 --> 01:19:18.800
impacted the same circuits
over three different events.
01:19:23.590 --> 01:19:25.550
Zoom in a little bit. so we can see
01:19:25.550 --> 01:19:28.743
for those events where those
damage incidents occurred.
01:19:30.890 --> 01:19:32.050
This is an interesting one
01:19:32.050 --> 01:19:34.350
because we're obviously seeing a whole series
01:19:34.350 --> 01:19:37.493
of damages along a particular circuit.
01:19:38.420 --> 01:19:40.160
The numbers represent their ranking
01:19:40.160 --> 01:19:45.160
as the result of the fire modeling analysis.
01:19:45.830 --> 01:19:48.640
So we're seeing number one seven, four, six
01:19:48.640 --> 01:19:50.723
and nine all along the same area.
01:19:53.393 --> 01:19:55.153
And you'll see that come out in the results.
01:19:56.340 --> 01:19:58.480
We're seeing a whole bunch here
01:19:58.480 --> 01:20:03.480
and the game area
North of LA to the table up.
01:20:04.080 --> 01:20:06.448
So you can also see the total damages
01:20:06.448 --> 01:20:09.740
the number of events for each of those.
01:20:09.740 --> 01:20:12.023
These are the events of the 10th and 11th.
01:20:18.410 --> 01:20:22.180
I'll continue to show you a
few maps to zoom in here
01:20:22.180 --> 01:20:23.880
and then we'll get to the results.
01:20:26.182 --> 01:20:28.727
This one shows one
damage incident from the 16th
01:20:29.650 --> 01:20:31.053
and what was de-energized.
01:20:35.270 --> 01:20:39.190
Here's the 24th showing
you the damage incidents
01:20:39.190 --> 01:20:42.323
and a PSPS event boundaries.
01:20:50.730 --> 01:20:55.207
And lastly, we advance
to the 28th to the 30th
01:20:55.207 --> 01:20:57.180
shown on this yellow and orange colors
01:20:57.180 --> 01:20:58.620
where we get a significant amount
01:20:58.620 --> 01:21:01.443
of damages occur and captured.
01:21:05.780 --> 01:21:07.800
I'm just going to stop here for a second
01:21:07.800 --> 01:21:10.853
make sure we catch up and we'll finish off.
01:21:29.820 --> 01:21:32.800
This table like the PGE events I showed you.
01:21:32.800 --> 01:21:33.820
It shows a summary
01:21:35.341 --> 01:21:38.430
of all the 54 simulations that were run again
01:21:38.430 --> 01:21:41.470
representing multiple PSPS events.
01:21:41.470 --> 01:21:44.580
So you'll notice the ignition
time is included here.
01:21:44.580 --> 01:21:48.400
Identified in yellow are
the ones that we identified
01:21:48.400 --> 01:21:52.900
as most significant
you'll notice ranking four
01:21:52.900 --> 01:21:55.210
through seven were events that all occurred
01:21:55.210 --> 01:21:59.220
on 10 24, same ignition
time, similar impacts.
01:22:01.890 --> 01:22:06.870
So they're all clustered
closely to the actual ranking
01:22:06.870 --> 01:22:11.480
number one simulation and
therefore were not accommodated
01:22:11.480 --> 01:22:14.980
or included in the most significant incidents
01:22:14.980 --> 01:22:17.903
because we deemed them to be redundant.
01:22:18.950 --> 01:22:21.270
You see that with nine and 11
01:22:21.270 --> 01:22:23.063
and we also see with 18,
01:22:24.890 --> 01:22:28.890
again to demonstrate that
that cluster I showed you
01:22:28.890 --> 01:22:31.073
on one of the early maps on the 24th,
01:22:32.420 --> 01:22:33.730
along a single line here
01:22:33.730 --> 01:22:35.743
we get significant damages occurred.
01:22:38.810 --> 01:22:41.860
Incident number one was the most significant
01:22:41.860 --> 01:22:44.303
with respect to spread demonstrated here.
01:22:45.440 --> 01:22:48.036
Therefore those other ones were eliminated
01:22:48.036 --> 01:22:53.036
and not incorporated in
the final most significant
01:22:53.700 --> 01:22:57.700
because they were so close
that they wouldn't have ignited.
01:22:57.700 --> 01:23:00.210
They would have been burnt over very quickly
01:23:00.210 --> 01:23:02.813
by the fire spread for
the first damage incident.
01:23:05.100 --> 01:23:08.750
An attempt of course, to make
sure that we don't double dip
01:23:08.750 --> 01:23:11.723
and account redundantly across those.
01:23:15.440 --> 01:23:18.965
So across the three the five different events
01:23:18.965 --> 01:23:21.770
where we had damage incidents
01:23:23.960 --> 01:23:26.430
we're seeing the distribution
01:23:26.430 --> 01:23:29.250
of the most significant across those events.
01:23:29.250 --> 01:23:31.480
10 were out of the October 10th, four
01:23:31.480 --> 01:23:35.683
out of the October 24th and
two out of the October 30th.
01:23:36.670 --> 01:23:38.960
There's some summaries on the maximum average
01:23:38.960 --> 01:23:43.090
and standard deviation of
the potential impacts as well.
01:23:43.090 --> 01:23:45.600
Let's just summarize those up and have a look
01:23:45.600 --> 01:23:48.923
at some of the impact analysis.
01:23:50.040 --> 01:23:51.557
For the October 10th event these were
01:23:51.557 --> 01:23:55.960
the five most significant from that event.
01:23:55.960 --> 01:23:58.260
Notice all of them with the exception
01:23:58.260 --> 01:24:00.493
of the bottom one have IAA is a five.
01:24:03.370 --> 01:24:05.780
Significant amount of
population of buildings impacted.
01:24:05.780 --> 01:24:07.940
Obviously the density of population
01:24:07.940 --> 01:24:11.333
around Los Angeles area leads to this.
01:24:12.520 --> 01:24:14.723
We see similar for October 24th,
01:24:15.820 --> 01:24:19.710
again significant impact
fires and not super large
01:24:19.710 --> 01:24:21.360
but obviously in the wrong place.
01:24:24.880 --> 01:24:27.140
Lastly, on October 28th to the 30th
01:24:28.250 --> 01:24:30.350
we see two of the most significant events.
01:24:31.310 --> 01:24:33.220
The chart at the bottom as seen
01:24:33.220 --> 01:24:36.243
in the other PSPS analysis
shows that distribution.
01:24:37.930 --> 01:24:39.285
Interestingly, in this one
01:24:39.285 --> 01:24:43.070
when you look at acres
burn across the population
01:24:43.070 --> 01:24:45.780
character buildings impacted characteristics
01:24:45.780 --> 01:24:49.100
we see that really vary widely.
01:24:49.100 --> 01:24:52.650
In the first one significant impacts
01:24:52.650 --> 01:24:54.780
particularly the population
due to the density
01:24:54.780 --> 01:24:59.130
of the type of housing
yet, not an overly large fire.
01:24:59.130 --> 01:25:02.061
The second one, we see a large fire,
01:25:02.061 --> 01:25:06.395
third one, not so large fire, big impact.
01:25:06.395 --> 01:25:10.443
And we see that trend across the SCE
01:25:11.310 --> 01:25:13.010
because of its unique characteristics
01:25:13.010 --> 01:25:15.930
and spatial variability and the landscape,
01:25:15.930 --> 01:25:18.320
and particularly where the values are risk.
01:25:18.320 --> 01:25:20.913
Distinctly different than the PG and E.
01:25:23.810 --> 01:25:25.480
Here's where the most significant
01:25:25.480 --> 01:25:28.630
where most were North of that LA area.
01:25:28.630 --> 01:25:29.680
You see that clustering
01:25:29.680 --> 01:25:34.680
of the dark red IAA 5 so
extreme spread scenarios.
01:25:34.757 --> 01:25:36.193
You see one up North.
01:25:41.210 --> 01:25:43.780
Let's look at some of those most significant
01:25:43.780 --> 01:25:44.743
here's number one,
01:25:50.840 --> 01:25:53.450
significant amount of population impacted
01:25:53.450 --> 01:25:57.493
almost 8,000 relatively small fire, 15,000.
01:25:58.960 --> 01:26:02.510
Once we see the
deterministic simulation spread
01:26:02.510 --> 01:26:03.823
becomes really obvious.
01:26:05.640 --> 01:26:08.430
The particularly impacts in this WUI area
01:26:10.040 --> 01:26:12.120
San Anna condition, no doubt
01:26:12.120 --> 01:26:15.577
posted from the East into the West.
01:26:15.577 --> 01:26:17.080
And we see that encroachment
01:26:17.080 --> 01:26:21.430
into this area over that time period.
01:26:21.430 --> 01:26:24.310
Notice a lot of that happens
early on in the green.
01:26:24.310 --> 01:26:29.310
So pretty early any impact.
01:26:29.840 --> 01:26:31.690
We look at the probabilistic.
01:26:31.690 --> 01:26:33.480
We see 80 to a hundred percent.
01:26:33.480 --> 01:26:35.680
A lot of those impacts happen early on
01:26:35.680 --> 01:26:36.880
in this particular fire.
01:26:51.506 --> 01:26:54.270
Again, the fire behavior on weather charts
01:26:54.270 --> 01:26:57.820
as we just know that you
look at the fire behavior
01:26:57.820 --> 01:26:59.390
look at the extreme rates of spread
01:26:59.390 --> 01:27:02.896
in those first seven, eight hours
01:27:02.896 --> 01:27:04.660
we've got very high rates
of spread represented
01:27:04.660 --> 01:27:06.913
by the orange on the top of the chart.
01:27:08.520 --> 01:27:10.870
We also see very high flame length.
01:27:10.870 --> 01:27:13.510
So fire intensity and
the first eight hours as
01:27:13.510 --> 01:27:16.963
well represented by the
goal on the flame life chart.
01:27:18.450 --> 01:27:21.210
We're seeing a close
association routine predicted
01:27:21.210 --> 01:27:23.420
and observed, which is really good
01:27:23.420 --> 01:27:24.950
the density of weather stations
01:27:24.950 --> 01:27:27.430
and that data used as part of the modeling.
01:27:27.430 --> 01:27:30.815
Predictive modeling gives
us a good comfort level
01:27:30.815 --> 01:27:34.513
with the weather data
and therefore the results.
01:27:38.600 --> 01:27:43.323
Let's move on to number
two, De Mar area, larger fire,
01:27:50.730 --> 01:27:52.350
again San Anna condition,
01:27:52.350 --> 01:27:54.373
significant winded spread very quickly.
01:27:56.540 --> 01:27:59.150
We see a classic Rui scenario in the North.
01:27:59.150 --> 01:28:02.180
It gets impacted pretty
early on in this fire.
01:28:02.180 --> 01:28:04.700
And then the urban area on the left
01:28:04.700 --> 01:28:09.570
gets impacted much farther along.
01:28:09.570 --> 01:28:12.090
We see significant
encroachment into those areas.
01:28:12.090 --> 01:28:14.363
Of course, again, surrounding fuel load.
01:28:21.840 --> 01:28:23.460
Here's the deterministic, sorry.
01:28:23.460 --> 01:28:25.110
The probabilistic spread of that
01:28:28.080 --> 01:28:29.290
similar representation
01:28:32.300 --> 01:28:35.800
look at the zero to 20% probability
01:28:36.809 --> 01:28:38.210
it cause significantly more damage
01:28:38.210 --> 01:28:40.970
if it were to spread
beyond the deterministic,
01:28:40.970 --> 01:28:44.270
particularly in the Northwest
and also in the South.
01:28:44.270 --> 01:28:47.980
Possible but not incorporated directly
01:28:47.980 --> 01:28:50.530
in our analysis because
of the lower probabilities.
01:28:51.780 --> 01:28:53.780
This is the beauty of the probabilistic.
01:29:00.590 --> 01:29:02.160
Let's jump ahead here and look
01:29:02.160 --> 01:29:03.933
at one that's farther along.
01:29:06.010 --> 01:29:08.120
Smaller populations different location
01:29:10.800 --> 01:29:12.703
out East, away from Los Angeles.
01:29:14.610 --> 01:29:17.510
Again, same type of wind scenario
01:29:17.510 --> 01:29:21.321
an example of a fire in the
wrong place, unfortunately
01:29:21.321 --> 01:29:25.713
and surrounded by people and buildings,
01:29:27.200 --> 01:29:29.010
not a lot of classic Rui
01:29:29.010 --> 01:29:31.230
dotted homes throughout their fairly
01:29:31.230 --> 01:29:33.530
significant subdivisions and urban areas.
01:29:33.530 --> 01:29:37.683
And you can see the impacts,
particularly in the South park.
01:29:41.210 --> 01:29:43.610
Examples, I've chosen
really helped to demonstrate
01:29:43.610 --> 01:29:45.450
that not all fires and in this regard
01:29:45.450 --> 01:29:47.653
fire simulations are created equal.
01:29:49.020 --> 01:29:51.030
The location of those
01:29:51.030 --> 01:29:53.610
the conditions around it,
and that spatial variability
01:29:53.610 --> 01:29:56.310
on where people live is really critical.
01:29:56.310 --> 01:30:00.030
And that's that variability
we see across the state.
01:30:00.030 --> 01:30:03.690
It's not the same, it's
different in different areas.
01:30:03.690 --> 01:30:05.340
Those of us in the fire business know
01:30:05.340 --> 01:30:08.310
that significant difference
from Southern California
01:30:08.310 --> 01:30:11.770
with its fuels versus Northern California.
01:30:11.770 --> 01:30:15.200
And so get different types of
buyers in different scenarios.
01:30:15.200 --> 01:30:19.080
None the less in this
regard for the CPUC analysis
01:30:19.080 --> 01:30:21.153
we have a framework for comparing those.
01:30:24.010 --> 01:30:25.743
Here's our probabilistic run.
01:30:28.760 --> 01:30:31.750
We see a lot of potential
for it to spread up North
01:30:33.220 --> 01:30:37.840
not a lot in those final
20%, but the deterministic
01:30:37.840 --> 01:30:41.203
and the top 80% match 60% match really well.
01:30:45.550 --> 01:30:48.330
Again, that data is
available for all of that.
01:30:48.330 --> 01:30:50.680
We focus with the probabilistic diagnose most
01:30:50.680 --> 01:30:53.910
significant here's the summary of results
01:30:56.310 --> 01:30:59.437
for the Edison PSPS events.
01:31:01.570 --> 01:31:05.253
We see that October 2nd,
that's real significant potential.
01:31:06.580 --> 01:31:11.580
Second highest number of
damage incidents with 17 expected.
01:31:11.653 --> 01:31:13.970
But when you compare it to the October 27th
01:31:13.970 --> 01:31:18.970
where we had 20 damage
incidents is about double the impact.
01:31:19.670 --> 01:31:22.430
A good indicator of the really matters again
01:31:22.430 --> 01:31:23.900
where those incidents occur
01:31:25.010 --> 01:31:26.710
and they're not all created equal.
01:31:29.900 --> 01:31:30.733
But we're gonna shop
01:31:30.733 --> 01:31:34.780
by looking at the SDG&E
events substantially smaller
01:31:36.070 --> 01:31:38.343
than a PGD or Edison.
01:31:39.960 --> 01:31:42.130
We had three PSPS events.
01:31:42.130 --> 01:31:43.850
One, there was no damages reported.
01:31:43.850 --> 01:31:45.683
That's shown in gray in this table,
01:31:46.610 --> 01:31:48.977
October 24th we had six
01:31:48.977 --> 01:31:53.977
and October 28th, we
had nine damage incidents.
01:31:57.330 --> 01:31:58.590
When we looked at the data
01:31:58.590 --> 01:32:01.460
of those 15 damage incidents reported
01:32:01.460 --> 01:32:04.270
only 13 were identified as a potential
01:32:04.270 --> 01:32:09.270
to cause that ignition and
were within the PSPS boundaries
01:32:09.310 --> 01:32:12.020
and were in close enough proximity
01:32:12.020 --> 01:32:14.390
to actually ignite a fire.
01:32:14.390 --> 01:32:17.593
So 13 simulations were conducted.
01:32:22.540 --> 01:32:24.420
As with the other events here's a map
01:32:24.420 --> 01:32:29.290
of those PSPS event boundaries,
like the Edison analysis.
01:32:29.290 --> 01:32:31.250
We bundled them all together
01:32:31.250 --> 01:32:34.486
because of the small amount
of actual damage ignition.
01:32:34.486 --> 01:32:39.220
You can see color coded
with the blue beam 1024 events
01:32:39.220 --> 01:32:43.720
rates through the reddy
orange areas being 10 30,
01:32:43.720 --> 01:32:46.121
10 29 events important to note
01:32:46.121 --> 01:32:48.150
since he's represent different times
01:32:48.150 --> 01:32:50.866
there is overlap between them.
01:32:50.866 --> 01:32:52.885
And so certain areas pull up
01:32:52.885 --> 01:32:57.193
consistently on top of each other for events.
01:33:00.870 --> 01:33:04.263
Although not necessarily were
damages found in those areas.
01:33:07.410 --> 01:33:12.410
There's a summary of the 13
simulations that were conducted.
01:33:14.340 --> 01:33:18.093
In this case, we saw a
clean break at the top five.
01:33:22.780 --> 01:33:25.390
Interestingly, not only do we see a break
01:33:25.390 --> 01:33:29.020
in the population impacted
buildings impact in firesides
01:33:29.020 --> 01:33:32.537
but if you look at the IAA
they're all fives and fours
01:33:32.537 --> 01:33:36.555
and a boom, we see a big
drop with lower spread fires.
01:33:36.555 --> 01:33:39.434
Although still again, wrong location.
01:33:39.434 --> 01:33:42.800
Look at fire number seven there
01:33:44.100 --> 01:33:48.520
only 2000 acres yet could impact 900 people
01:33:48.520 --> 01:33:49.913
and 900 buildings.
01:33:53.920 --> 01:33:56.370
The table and charts
showing the summary of those
01:34:02.040 --> 01:34:03.680
incident damage incident number one
01:34:03.680 --> 01:34:06.400
clearly stands above, particularly
01:34:06.400 --> 01:34:09.810
in acres burn 96,000 compared to number two
01:34:09.810 --> 01:34:11.770
that's on the second highest one
01:34:11.770 --> 01:34:14.390
which had actually answered
the three only has 18,000
01:34:14.390 --> 01:34:17.680
roughly so significantly larger fires
01:34:17.680 --> 01:34:22.680
and it's expected significantly more impacts,
01:34:23.130 --> 01:34:27.900
10,000 people 11,000
buildings stands alone by itself.
01:34:27.900 --> 01:34:29.290
You know, one of the things we noticed
01:34:29.290 --> 01:34:31.900
when you looked at the chart
again is for damage incident.
01:34:31.900 --> 01:34:35.267
One, the size of the fire
is large and large impacts.
01:34:35.267 --> 01:34:38.560
The rest of them are smaller fires.
01:34:38.560 --> 01:34:40.613
Yet large impacts two through five.
01:34:45.000 --> 01:34:48.043
This is where the five most
significant were located.
01:34:51.360 --> 01:34:53.660
The other ones are
showing a small black dots.
01:34:56.810 --> 01:34:59.680
The other eight let's have a look
01:34:59.680 --> 01:35:01.543
at those most significant.
01:35:06.990 --> 01:35:10.270
Again, number one stands
up starting near Ramona
01:35:11.610 --> 01:35:14.793
and then spreading in a Santa Ana scenario.
01:35:15.840 --> 01:35:17.340
Very similar to the Cedar fire
01:35:17.340 --> 01:35:19.083
that actually occurred in 2003,
01:35:25.880 --> 01:35:27.880
deterministic shows a significant spread
01:35:27.880 --> 01:35:31.380
not only into the classic WUI,
01:35:31.380 --> 01:35:34.283
but into those denser populated areas,
01:35:35.420 --> 01:35:38.730
obviously mostly centrally
and then to the North
01:35:38.730 --> 01:35:40.670
and then really on the, the West
01:35:40.670 --> 01:35:41.953
as you get towards Poway.
01:35:44.740 --> 01:35:46.790
Considerable rates of spread in this one,
01:35:49.210 --> 01:35:51.663
here's the probabilistic, very similar,
01:35:53.120 --> 01:35:57.203
again aligning right 40%
to a hundred percent.
01:35:58.630 --> 01:36:02.037
The conditions were
very significant here in it.
01:36:02.037 --> 01:36:04.263
And modeling really, really captures that.
01:36:06.910 --> 01:36:09.260
We bought the fire
behavior and weather charts.
01:36:10.910 --> 01:36:12.090
What did we see for this fire?
01:36:12.090 --> 01:36:14.250
If we look at the rates of spread very high
01:36:14.250 --> 01:36:16.886
to extreme rates of spreads throughout
01:36:16.886 --> 01:36:19.320
the fire for the first
particularly over 20 hours
01:36:19.320 --> 01:36:22.540
in particular, the first 15
01:36:23.610 --> 01:36:26.100
and an hour six, seven
eight is when it really kicks
01:36:26.100 --> 01:36:28.320
off our forest land really kicks off.
01:36:28.320 --> 01:36:32.668
We get significant spread
as represented by the maps
01:36:32.668 --> 01:36:36.220
but we also get intensity
here, look at the plane line.
01:36:36.220 --> 01:36:38.852
We see very high even extreme intensity,
01:36:38.852 --> 01:36:41.810
notice we have not seen
much extreme intensity
01:36:41.810 --> 01:36:46.170
and some of the others example simulations
01:36:46.170 --> 01:36:48.230
but this one again, starting I guess.
01:36:48.230 --> 01:36:49.710
So we have a combination and this one
01:36:49.710 --> 01:36:54.490
of not only significant
spread, but significant intensity
01:36:54.490 --> 01:36:56.440
which can lead to that
damage that we saw some
01:36:56.440 --> 01:37:00.023
of the largest damages we
saw in any potential event.
01:37:02.060 --> 01:37:06.620
We all know FTG (indistinct)
have working heavily
01:37:06.620 --> 01:37:10.270
on its advanced weather
system and it really performed.
01:37:10.270 --> 01:37:13.123
And in this regard we're seeing a close match
01:37:13.123 --> 01:37:14.860
with the density of the weather stations
01:37:14.860 --> 01:37:19.860
as well throughout that again
giving us good confidence
01:37:19.960 --> 01:37:21.163
in that prediction data.
01:37:22.750 --> 01:37:23.900
Let's look at the second one here
01:37:23.900 --> 01:37:26.640
because there's a large
drop-off, but nonetheless
01:37:26.640 --> 01:37:29.030
still a potential for significant impacts.
01:37:35.230 --> 01:37:37.450
Deterministic significant WUI area
01:37:42.090 --> 01:37:45.920
some dense urban there,
encroaches on that dense urban
01:37:45.920 --> 01:37:48.180
particularly on what we call the urban fringe
01:37:50.160 --> 01:37:53.910
and a fairly dense WUI scenario.
01:37:53.910 --> 01:37:56.480
And it hits up pretty
quickly and pretty early on
01:37:56.480 --> 01:37:58.053
in this particular simulation.
01:37:59.800 --> 01:38:02.750
Probabilistic shows the same pattern.
01:38:02.750 --> 01:38:05.025
Probabilistic really identifies the potential
01:38:05.025 --> 01:38:07.933
for this to be way more significant.
01:38:10.210 --> 01:38:13.960
If you look at that zero to 20%
01:38:13.960 --> 01:38:17.053
we really start to encroach
on major urban areas.
01:38:19.140 --> 01:38:21.897
Again, the overlap of
the deterministic is black
01:38:21.897 --> 01:38:24.180
and you can see it doesn't touch those areas
01:38:24.180 --> 01:38:29.180
but anything from 40, even
60% on this could be a really
01:38:29.843 --> 01:38:33.190
really bad one, much worse
than we actually predict.
01:38:33.190 --> 01:38:36.000
If things got out of control
and the winds perhaps picked
01:38:36.000 --> 01:38:37.953
up or got worse than predicted.
01:38:39.970 --> 01:38:43.190
Again, here's the summary of the PSPS events
01:38:43.190 --> 01:38:48.190
for STG&E a again the
October 28th event is the one
01:38:48.660 --> 01:38:50.960
with the most significant impacts.
01:38:50.960 --> 01:38:52.140
That's where obviously
01:38:52.140 --> 01:38:54.400
where our fire number one damage incident
01:38:54.400 --> 01:38:55.653
number one was located.
01:38:57.060 --> 01:38:58.710
What we're seeing is due to the fuels.
01:38:58.710 --> 01:39:01.140
We're seeing significant acres burn
01:39:01.140 --> 01:39:03.423
for a small amount of damage incidents,
01:39:04.425 --> 01:39:06.750
primarily because the shrub grass
01:39:06.750 --> 01:39:09.040
and particularly the shrub fuels well-known
01:39:09.040 --> 01:39:12.080
to Southern California,
particularly in San Diego County
01:39:12.080 --> 01:39:14.713
that really facilitate significant growth.
01:39:18.370 --> 01:39:20.883
All right, got you a ton of stuff.
01:39:22.860 --> 01:39:26.273
Let's look at some of the
findings we found to finish up.
01:39:28.180 --> 01:39:32.000
Well, obviously when we
look at all the events here
01:39:34.133 --> 01:39:36.510
we see some significant events jump
01:39:36.510 --> 01:39:38.643
out at us by the potential impacts.
01:39:40.130 --> 01:39:42.954
Obviously the PG&E event of the 26th
01:39:42.954 --> 01:39:45.863
could have resulted in significant damages.
01:39:47.007 --> 01:39:49.030
So this is what could have been
01:39:50.290 --> 01:39:54.370
but was a verdict due to the de-energization.
01:39:54.370 --> 01:39:57.530
We see the next most
significant event actually being
01:39:57.530 --> 01:40:00.860
at the GME, although not
a lot of damage incidents
01:40:00.860 --> 01:40:03.270
right on October 28th, the end
01:40:03.270 --> 01:40:05.730
of October was not a good time in 2019
01:40:05.730 --> 01:40:10.127
as is often the case in many
years for PSPS well understood.
01:40:10.127 --> 01:40:13.733
And that the October 2nd event with Edison.
01:40:21.360 --> 01:40:23.630
What kind of summary findings
01:40:23.630 --> 01:40:25.380
and recommendations we can up lift?
01:40:26.800 --> 01:40:28.530
Well, obviously we saw a large amount
01:40:28.530 --> 01:40:31.023
of damages for some PSPS events.
01:40:32.790 --> 01:40:35.250
The first time we've been
able to actually characterize
01:40:35.250 --> 01:40:37.801
and quantify what could have been.
01:40:37.801 --> 01:40:42.374
Now, it's important to know
that this study did not include
01:40:42.374 --> 01:40:46.320
or incorporate what actually was,
01:40:46.320 --> 01:40:51.290
what were the impacts due
to de-energization and PSPS.
01:40:51.290 --> 01:40:53.450
We focused on what could have been
01:40:53.450 --> 01:40:56.140
there's another side of the
equation that was not addressed
01:40:56.140 --> 01:40:58.240
and was not within scope for this project.
01:41:00.580 --> 01:41:02.530
We consistently see specific areas
01:41:02.530 --> 01:41:05.223
and assets identified as high risk.
01:41:07.210 --> 01:41:11.040
This area did not compare
these to the mitigation priorities
01:41:11.040 --> 01:41:13.543
or the wildfire mitigation plans at the IOUs.
01:41:15.540 --> 01:41:18.317
Many of those I expect would
be located or accommodated
01:41:18.317 --> 01:41:22.000
in those, through their
mitigation priorities.
01:41:22.000 --> 01:41:23.653
However, we did not compare that.
01:41:25.250 --> 01:41:26.960
We consistently see de-energization
01:41:26.960 --> 01:41:31.324
and similar locations and
those damage incidents
01:41:31.324 --> 01:41:32.903
in similar locations.
01:41:35.530 --> 01:41:39.180
Impacts are highly dependent
on that spatial variability.
01:41:39.180 --> 01:41:43.010
So they're varying environment
and the landscape conditions
01:41:43.010 --> 01:41:44.600
part of that is fuels.
01:41:44.600 --> 01:41:46.843
Part of it is unique weather conditions
01:41:46.843 --> 01:41:49.740
that occurs in certain
valleys in certain areas
01:41:54.930 --> 01:41:56.673
and also where people are located.
01:41:57.600 --> 01:42:00.810
Fundamentally, this
consequence analysis is all
01:42:00.810 --> 01:42:04.363
about where people live
and what density they live.
01:42:05.274 --> 01:42:08.070
And the fuel loads that are around them.
01:42:09.765 --> 01:42:11.765
That, that leads to what level of impact
01:42:12.800 --> 01:42:14.403
or encroachment could occur.
01:42:15.480 --> 01:42:18.220
We found a lot of variability, the input data
01:42:19.310 --> 01:42:22.830
particularly the data provided
by the IOUs early days
01:42:22.830 --> 01:42:27.480
of PSPS and early days
of collection of this data.
01:42:27.480 --> 01:42:31.310
And we dive deeply into
those collection methods
01:42:31.310 --> 01:42:34.770
and the IOUs were excellent
and provide detailed information
01:42:34.770 --> 01:42:37.620
about their collection
methods, their analysis
01:42:37.620 --> 01:42:40.693
of that data and the decisions they made.
01:42:42.320 --> 01:42:43.300
It's also pretty clear
01:42:43.300 --> 01:42:47.490
that fire spread modeling
prior to events can aid in PSPS
01:42:47.490 --> 01:42:51.070
decision-making if we know what's possible
01:42:51.070 --> 01:42:52.330
the consequence side
01:42:52.330 --> 01:42:56.823
can we believe really aid
that PSPS decision-making?
01:42:59.208 --> 01:43:02.490
The IOUs would benefit from
conducting similar analysis
01:43:02.490 --> 01:43:05.193
for damaged data collected for future events.
01:43:08.550 --> 01:43:10.350
It's clear we need to standardize
01:43:10.350 --> 01:43:13.350
the post event data collection process.
01:43:13.350 --> 01:43:16.830
So there's a consistency in schema format
01:43:16.830 --> 01:43:19.400
and the criteria used to make decisions.
01:43:19.400 --> 01:43:22.340
So that data provided by each of the IOUs
01:43:22.340 --> 01:43:25.353
is standardized and consistent.
01:43:27.910 --> 01:43:30.260
We also need to standardize the data format
01:43:30.260 --> 01:43:34.700
and the criteria used for
reporting the events as well.
01:43:34.700 --> 01:43:36.991
You could see similar variation
01:43:36.991 --> 01:43:41.991
and the input data in polygons from the IOUs.
01:43:42.250 --> 01:43:45.530
And that's also leads that if
this type of modeling was to
01:43:45.530 --> 01:43:49.580
be implemented as a
standard, perhaps in future
01:43:49.580 --> 01:43:52.290
for helping to quantify what have been,
01:43:52.290 --> 01:43:56.396
you need to standardize
criteria around the input data
01:43:56.396 --> 01:43:59.875
the assumptions associated with that modeling
01:43:59.875 --> 01:44:02.023
and also the openness.
01:44:09.230 --> 01:44:11.880
That's the final slide in our presentation
01:44:11.880 --> 01:44:15.413
of the fire modeling for the 2019 PSPS event.
01:44:17.310 --> 01:44:19.940
I'll throw it back over the fence to you Tony
01:44:19.940 --> 01:44:22.103
accommodate any public comments or a Q&A.
01:44:24.660 --> 01:44:28.579
Great thank you very much,
David, just before we get
01:44:28.579 --> 01:44:30.730
into the public comment or
question and answer period
01:44:30.730 --> 01:44:32.100
just to note any members
01:44:32.100 --> 01:44:34.820
of the public may make public comments
01:44:34.820 --> 01:44:37.030
comments may not address issues pertaining
01:44:37.030 --> 01:44:39.420
to the 2019 PSPS events that are subject
01:44:39.420 --> 01:44:42.110
to and within the scope of a Judy Curry
01:44:42.110 --> 01:44:44.840
or rate setting proceedings
pursuant to Commission rules
01:44:44.840 --> 01:44:49.290
of practice and procedures,
article eight ex parte rules
01:44:49.290 --> 01:44:54.290
and government code section 11125.7H
01:44:54.600 --> 01:44:55.710
and the Commission's rules
01:44:55.710 --> 01:44:58.193
of public comment at Commission meetings.
01:44:59.610 --> 01:45:04.423
So given that operator we
will open it up for any questions.
01:45:06.010 --> 01:45:07.263
Thank you.
01:45:07.263 --> 01:45:08.880
The public comment line is now open.
01:45:08.880 --> 01:45:11.530
If you wish to speak during
the public comment period
01:45:11.530 --> 01:45:14.180
or you have any questions,
please press star one
01:45:14.180 --> 01:45:17.270
unmute your phone and
clearly record your name
01:45:17.270 --> 01:45:19.124
and organization when prompted.
01:45:19.124 --> 01:45:20.684
Our first public comment
01:45:20.684 --> 01:45:23.250
or question comes from Kate Woodford.
01:45:23.250 --> 01:45:25.003
Your line is open, please go ahead.
01:45:26.330 --> 01:45:27.540
Thank you.
01:45:27.540 --> 01:45:29.760
My name is Kate Woodford and I work for,
01:45:29.760 --> 01:45:31.550
and I'm commenting on behalf of
01:45:31.550 --> 01:45:34.300
the Center for Accessible Technology.
01:45:34.300 --> 01:45:36.170
We appreciate the efforts the Commission has
01:45:36.170 --> 01:45:38.570
made to establish potential losses due
01:45:38.570 --> 01:45:40.280
to fires that might have happened.
01:45:40.280 --> 01:45:43.223
If the IOUs had not de-energized their lines
01:45:43.223 --> 01:45:46.957
in anticipation of a higher
high fire threat conditions
01:45:46.957 --> 01:45:51.520
it is an important
calculation of potential harm.
01:45:51.520 --> 01:45:53.330
At the same time, we know
01:45:53.330 --> 01:45:56.450
that customers and communities
experienced direct harm.
01:45:56.450 --> 01:45:59.110
Each time the power is turned off.
01:45:59.110 --> 01:45:59.943
We would like to know
01:45:59.943 --> 01:46:02.390
if the Commission is
conducting a similar evaluation
01:46:02.390 --> 01:46:05.666
of the actual harms caused
by actual de-energization
01:46:05.666 --> 01:46:09.210
events that took place in 2019.
01:46:09.210 --> 01:46:12.270
Actual harms include direct
and immediate health impacts
01:46:12.270 --> 01:46:15.280
for those on respirators
or other medical devices
01:46:15.280 --> 01:46:18.240
who lose power and don't have backup devices
01:46:18.240 --> 01:46:21.360
as well as the long term health impacts.
01:46:21.360 --> 01:46:25.220
They also include financial
harm, including the loss of food
01:46:25.220 --> 01:46:28.614
and medication due to the lack of power,
01:46:28.614 --> 01:46:32.530
or it could include the loss
of wages because they place
01:46:32.530 --> 01:46:37.460
of work might be closed due
to the power that was shut off.
01:46:37.460 --> 01:46:39.880
Is there any calculation of risk and harm
01:46:39.880 --> 01:46:42.005
that also should include traffic risks
01:46:42.005 --> 01:46:45.320
to traffic signals that were without power
01:46:45.320 --> 01:46:47.570
during a de-energization event
01:46:47.570 --> 01:46:50.070
or possibly what the injury or loss
01:46:50.070 --> 01:46:51.780
of life or damage to property
01:46:51.780 --> 01:46:54.200
and curve from a fire that might be caused
01:46:54.200 --> 01:46:56.920
by generators are
customers using fires to cook
01:46:57.851 --> 01:46:59.430
because it made the de-energization event.
01:46:59.430 --> 01:47:03.150
The analysis of potential
harm had the utilities refrain
01:47:03.150 --> 01:47:07.040
from shutting off power in
2019 is a hypothetical estimate.
01:47:07.040 --> 01:47:09.599
These actual harms suffered were real
01:47:09.599 --> 01:47:11.930
and should be analyzed the damages
01:47:11.930 --> 01:47:14.000
to the citizens of California from repetitive
01:47:14.000 --> 01:47:16.750
and extensive power shut offs are measurable
01:47:16.750 --> 01:47:19.503
and for way too many devastatingly real,
01:47:19.503 --> 01:47:21.960
there are consequences of losing power
01:47:21.960 --> 01:47:24.150
that deeply affect vulnerable people.
01:47:24.150 --> 01:47:27.570
The reality also is that most
of these vulnerable people
01:47:27.570 --> 01:47:29.925
are those that are lower income elderly,
01:47:29.925 --> 01:47:33.800
disabled non-English speaking and et cetera.
01:47:33.800 --> 01:47:37.280
And they are the ones who
suffer the heaviest burden
01:47:37.280 --> 01:47:38.533
of these power shut off.
01:47:39.810 --> 01:47:42.932
Sifrei T has repeatedly
argued with the Commission
01:47:42.932 --> 01:47:45.180
that they must look at the harms associated
01:47:45.180 --> 01:47:46.960
with shutting the power off as well
01:47:46.960 --> 01:47:50.950
as those associated
with keeping the power on.
01:47:50.950 --> 01:47:53.220
We asked again here
for this to be considered.
01:47:53.220 --> 01:47:54.053
Thank you.
01:47:57.550 --> 01:47:59.810
Thank you, miss Woodford for that.
01:47:59.810 --> 01:48:01.720
As of right now, there is that I'm not aware
01:48:01.720 --> 01:48:04.530
of any actual harm
analysis as part of this study
01:48:06.100 --> 01:48:09.010
but I will take a note on that
and I appreciate a comment.
01:48:09.010 --> 01:48:10.780
Thank you, operator.
01:48:10.780 --> 01:48:11.613
Next person.
01:48:11.613 --> 01:48:12.490
Next caller, please.
01:48:13.410 --> 01:48:16.210
The next caller is Ian Fisher with the public
01:48:16.210 --> 01:48:17.230
advocate's office.
01:48:17.230 --> 01:48:18.063
Your line is open.
01:48:18.063 --> 01:48:18.896
Please go ahead.
01:48:20.520 --> 01:48:22.190
Hello there, my name is Ian Fisher with
01:48:22.190 --> 01:48:23.390
the public advocate's office,
01:48:23.390 --> 01:48:24.610
I've got two or three questions
01:48:24.610 --> 01:48:26.620
I'd like to ask you specific to this study
01:48:26.620 --> 01:48:27.520
if you don't mind.
01:48:28.910 --> 01:48:30.330
I'll just jump straight into it.
01:48:30.330 --> 01:48:33.210
So my first question is what are you doing
01:48:33.210 --> 01:48:36.250
to assess the missing
potential addition data?
01:48:36.250 --> 01:48:38.441
We all know that you can detect the damage
01:48:38.441 --> 01:48:41.262
to the lines and the damage to the assets
01:48:41.262 --> 01:48:45.640
on some of the actual vegetation hazards.
01:48:45.640 --> 01:48:48.470
However, things like wind
Slack, and some of the kind
01:48:48.470 --> 01:48:51.570
of more temporal vegetation
hazards are not detectable
01:48:51.570 --> 01:48:54.020
on the inspection by the utilities.
01:48:54.020 --> 01:48:56.033
Have you done some sorts of assessments
01:48:56.033 --> 01:48:57.767
of that missing information
01:48:57.767 --> 01:49:00.860
and the potential bias in your models?
01:49:00.860 --> 01:49:04.290
And my second question speaks
01:49:04.290 --> 01:49:06.191
to the previous call as issue.
01:49:06.191 --> 01:49:09.455
I'm seeing only two measures of value
01:49:09.455 --> 01:49:12.320
as far as this is concerned.
01:49:12.320 --> 01:49:15.630
And the risk of one is acreage
under the risk population.
01:49:15.630 --> 01:49:19.289
Oh, sorry and buildings
apologies three acreage population
01:49:19.289 --> 01:49:24.289
and buildings through
additional issues and needs and
01:49:24.550 --> 01:49:26.020
Kobos that need to be assessed here
01:49:26.020 --> 01:49:28.130
including the evacuation od (indistinct)
01:49:28.130 --> 01:49:30.600
the level of elderly and disabled
01:49:30.600 --> 01:49:33.080
population within a particular location
01:49:33.080 --> 01:49:36.030
and this socioeconomic
groups within a particular location
01:49:36.030 --> 01:49:39.323
all of these will actually
affect the consequence
01:49:39.323 --> 01:49:42.123
and the risk associated with your model.
01:49:42.123 --> 01:49:44.560
How have you considered how you would build
01:49:44.560 --> 01:49:48.377
in those small detailed
assessments of the demographics?
01:49:48.377 --> 01:49:50.773
Thank you, I'll just
leave those two question.
01:49:53.160 --> 01:49:54.150
Thank you.
01:49:54.150 --> 01:49:57.150
I'll start and I'm not sure if
David has any to chime in too
01:49:58.330 --> 01:50:02.020
as far as data points beyond
what was studied here.
01:50:02.020 --> 01:50:05.410
This was just an exploration
of the fire modeling software.
01:50:05.410 --> 01:50:09.000
So the known damage
locations were used as again
01:50:09.000 --> 01:50:13.030
known points that we can do
apply the Technosilva model to
01:50:13.030 --> 01:50:16.136
to get those, get the
study in the exploration.
01:50:16.136 --> 01:50:19.149
That's the three main topics that were picked
01:50:19.149 --> 01:50:22.430
by Technosilva are the three largest ones
01:50:22.430 --> 01:50:24.120
the three largest concerns.
01:50:24.120 --> 01:50:26.500
There are many more, as you mentioned
01:50:26.500 --> 01:50:30.350
but the model was not,
that was not set up to do that.
01:50:30.350 --> 01:50:31.800
And David I'll let you chime in
01:50:31.800 --> 01:50:33.750
on whether that's even feasible or not.
01:50:35.010 --> 01:50:36.449
Sure, great question.
01:50:36.449 --> 01:50:39.420
But with respect to biases of the model
01:50:39.420 --> 01:50:42.969
I'm not sure that would
require a little more investigation
01:50:42.969 --> 01:50:46.377
but with respect to the values
of risk, a really good point
01:50:46.377 --> 01:50:51.210
as we found in doing this
work over the many years
01:50:51.210 --> 01:50:55.760
as the priority tends to be,
you know, lives, public safety
01:50:55.760 --> 01:50:58.410
and then buildings and financial,
01:50:58.410 --> 01:51:00.330
we fully appreciate those other factors,
01:51:00.330 --> 01:51:02.340
such as the liability.
01:51:02.340 --> 01:51:04.700
Again, in this regard, we're
looking at the fire spread
01:51:04.700 --> 01:51:08.060
from our particular ignition
source, which is that asset.
01:51:08.060 --> 01:51:11.281
And it identifies where that fire is going.
01:51:11.281 --> 01:51:13.690
We're fully aware that there's other factors
01:51:13.690 --> 01:51:16.120
on the impacted landscape.
01:51:16.120 --> 01:51:17.610
And you mentioned some of them, Ian
01:51:17.610 --> 01:51:20.615
such as social vulnerability can actually
01:51:20.615 --> 01:51:25.615
increase particular risk because of the lack
01:51:26.960 --> 01:51:29.820
of resources or inability a particular parts
01:51:29.820 --> 01:51:34.450
of the population to
evacuate in a timely basis.
01:51:34.450 --> 01:51:37.214
We also know that factors such as egress
01:51:37.214 --> 01:51:40.580
the density of population in combination
01:51:40.580 --> 01:51:44.440
with access routes can come to play
01:51:44.440 --> 01:51:48.310
with respect to eat people's ability to get
01:51:48.310 --> 01:51:52.420
out of the way and thirdly,
suppression difficulty
01:51:52.420 --> 01:51:55.100
and the ability of the characteristics
01:51:55.100 --> 01:51:57.860
of the train and the
resources to actually get in
01:51:57.860 --> 01:51:59.123
to fight the fire.
01:52:01.440 --> 01:52:02.800
Social vulnerability egress
01:52:02.800 --> 01:52:07.199
and those characteristics
of the landscape are part
01:52:07.199 --> 01:52:11.654
of the model and
incorporated for calculating risk.
01:52:11.654 --> 01:52:14.640
We were doing that on
an active basis right now
01:52:14.640 --> 01:52:18.910
for the entire state of
California that focuses more
01:52:18.910 --> 01:52:23.090
on forecasting risk in the
consequences of forecasting risk
01:52:23.090 --> 01:52:24.240
particularly with looking
01:52:24.240 --> 01:52:28.370
at increasing risk from
particular ignition points
01:52:28.370 --> 01:52:30.570
such as electric utility assets
01:52:30.570 --> 01:52:32.740
so that we consider those characteristics
01:52:32.740 --> 01:52:35.056
because I agree they're real.
01:52:35.056 --> 01:52:38.780
They were not incorporated
in this study simply
01:52:38.780 --> 01:52:42.940
because the focus was
on unique ignition locations
01:52:43.975 --> 01:52:45.710
and where that would be we to make sure
01:52:45.710 --> 01:52:48.000
we have a standard for comparison.
01:52:48.000 --> 01:52:50.500
I think definitely a opportunity
01:52:50.500 --> 01:52:54.204
for improvement is to add
those landscape characteristics
01:52:54.204 --> 01:52:58.110
and susceptibility items, to the analysis.
01:52:58.110 --> 01:53:01.420
Again, social vulnerability,
suppression difficulty
01:53:01.420 --> 01:53:04.370
and egress being the most significant ones.
01:53:04.370 --> 01:53:09.360
Lastly, with respect to
incorporating other values at risk
01:53:09.360 --> 01:53:14.360
beyond buildings that
can easily be added to it
01:53:14.680 --> 01:53:17.130
because its analysis has conducted
01:53:17.130 --> 01:53:21.420
the integration of values at risk,
01:53:21.420 --> 01:53:26.420
such as highly sensitized landscapes
01:53:26.420 --> 01:53:27.903
that could be easily damaged,
01:53:29.010 --> 01:53:33.160
considerations of flood
or mudslides after fires.
01:53:33.160 --> 01:53:36.040
Those types of criteria could be added.
01:53:36.040 --> 01:53:37.810
And it's really a post-processing
01:53:37.810 --> 01:53:39.600
to the actual fire modeling.
01:53:39.600 --> 01:53:41.190
So they are things that could be added
01:53:41.190 --> 01:53:42.740
and leverage this data.
01:53:42.740 --> 01:53:44.840
Hopefully this helped answer the question.
01:53:46.140 --> 01:53:47.394
Thank you.
01:53:47.394 --> 01:53:48.670
That answers my second question.
01:53:48.670 --> 01:53:51.290
But my first question is still out there.
01:53:51.290 --> 01:53:54.570
Your ignition data is fundamentally biased
01:53:54.570 --> 01:53:57.410
because of the consequences of PSPS events,
01:53:57.410 --> 01:54:00.250
things like line slap and some vegetation
01:54:00.250 --> 01:54:03.050
hazard events are undetectable.
01:54:03.050 --> 01:54:05.670
And so if this has to be translated
01:54:05.670 --> 01:54:08.640
into some sort of assessment
and some sort of policy
01:54:08.640 --> 01:54:12.170
we need to get a handle on what those biases
01:54:12.170 --> 01:54:15.640
and where those errors
are in your input data.
01:54:15.640 --> 01:54:17.137
And my question to you
01:54:17.137 --> 01:54:21.430
and I guess more widely is
how are we getting a handle
01:54:21.430 --> 01:54:23.733
on those biases in the input data?
01:54:26.435 --> 01:54:30.400
With respect to the
regulations for data collectioning
01:54:30.400 --> 01:54:33.121
I'm gonna have to defer to the folks at CPUC
01:54:33.121 --> 01:54:36.890
that's not in our Bailey wick to respond to
01:54:38.420 --> 01:54:41.690
our focus was on the data that was available
01:54:41.690 --> 01:54:45.291
was to provide interrogate
the data as much as possible
01:54:45.291 --> 01:54:47.931
to make sure that it was as accurate
01:54:47.931 --> 01:54:51.020
not only positionally, but
particularly with respect
01:54:51.020 --> 01:54:55.650
to whether it would
actually arc and cause a fire
01:54:55.650 --> 01:54:59.640
and had several iterations
with engineering on our side
01:54:59.640 --> 01:55:02.749
IAA side for that with respect to others
01:55:02.749 --> 01:55:06.023
other data sets are gonna have to defer to.
01:55:06.960 --> 01:55:08.300
No I think you misunderstand me.
01:55:08.300 --> 01:55:09.240
This is not another dataset.
01:55:09.240 --> 01:55:11.360
This is just missing data that
01:55:11.360 --> 01:55:15.030
your universe of ignition
points is missing data
01:55:15.030 --> 01:55:18.134
because of the consequences of a PSPS events.
01:55:18.134 --> 01:55:22.686
That's my point and the
needs to be some handle.
01:55:22.686 --> 01:55:24.010
You need to recognize that
01:55:24.010 --> 01:55:26.210
and do a universe of ignition points.
01:55:26.210 --> 01:55:28.060
I would ask that you recognize
01:55:28.060 --> 01:55:30.083
the university's missing data.
01:55:31.280 --> 01:55:32.480
You see what I'm saying?
01:55:33.527 --> 01:55:34.360
I can't comment on what data
01:55:34.360 --> 01:55:35.620
we don't have and that's missing.
01:55:37.489 --> 01:55:39.843
I would, you know, probably
a deeper discussion on
01:55:39.843 --> 01:55:42.130
what data would potentially be missing
01:55:42.130 --> 01:55:44.640
with respect to ignition we could only model
01:55:44.640 --> 01:55:47.150
what was available from those surveys.
01:55:47.150 --> 01:55:48.450
So I think it would come back
01:55:48.450 --> 01:55:51.430
to perhaps some of those
findings and recommendations
01:55:51.430 --> 01:55:53.808
about the standardization
and collection of data
01:55:53.808 --> 01:55:57.760
with respect to the possible ignitions.
01:55:57.760 --> 01:55:59.490
The fundamental issue is that data is
01:55:59.490 --> 01:56:01.600
uncollectible because it's line slap.
01:56:01.600 --> 01:56:04.390
So it's transients and it's vegetation
01:56:04.390 --> 01:56:06.240
infants that are transient.
01:56:06.240 --> 01:56:09.570
And so the line man cannot collect that data.
01:56:09.570 --> 01:56:11.830
So you need to do some sort of
01:56:11.830 --> 01:56:13.730
experimental analysis of what's going on.
01:56:13.730 --> 01:56:16.580
I think there is technologies that I've seen
01:56:16.580 --> 01:56:19.880
from some providers that do allow
01:56:19.880 --> 01:56:22.100
the collection of that type of data
01:56:22.100 --> 01:56:25.710
when that would be information that should be
01:56:25.710 --> 01:56:27.711
used for this type of modeling.
01:56:27.711 --> 01:56:31.680
Again, have to defer to the
regulatory agency on that.
01:56:31.680 --> 01:56:33.420
But some of that data is collectible
01:56:33.420 --> 01:56:35.670
by certain technologies I've seen it in,
01:56:35.670 --> 01:56:38.633
out in the marketplace, not all of them.
01:56:38.633 --> 01:56:41.003
Only if the launch is switched on.
01:56:43.320 --> 01:56:45.400
Yeah I'm not an engineer so I can't...
01:56:45.400 --> 01:56:47.116
If the launch is switched
off, it's not collectible.
01:56:47.116 --> 01:56:49.340
That's the issue anyway, okay.
01:56:49.340 --> 01:56:51.483
I'll see at this point, by that point.
01:56:53.270 --> 01:56:54.690
Fair enough.
01:56:54.690 --> 01:56:55.745
Thank you.
01:56:55.745 --> 01:56:56.610
Thank you very much, Mr. Fisher,
01:56:56.610 --> 01:56:58.490
operator next caller, please.
01:57:00.580 --> 01:57:03.140
Our next call is John Mader, you line is open
01:57:03.140 --> 01:57:03.973
please go ahead.
01:57:05.664 --> 01:57:06.510
Yes this is John Mader
01:57:06.510 --> 01:57:09.410
with a Wildfire Safety Advisory Board.
01:57:09.410 --> 01:57:12.763
I have two questions and one comment.
01:57:13.600 --> 01:57:15.470
First of all, I want to say that, you know
01:57:15.470 --> 01:57:18.460
consequence mapping tools absolutely vital
01:57:18.460 --> 01:57:20.970
to being able to solve
these problems that are
01:57:20.970 --> 01:57:23.040
facing the people of California.
01:57:23.040 --> 01:57:25.894
So it's great to see this work being done.
01:57:25.894 --> 01:57:28.600
One is the model.
01:57:28.600 --> 01:57:31.037
You talk about 20 different models
01:57:31.037 --> 01:57:33.120
in what we'd like to see more detailed
01:57:33.120 --> 01:57:35.190
about what models are used.
01:57:35.190 --> 01:57:40.070
Like naturally you probably
using weather models
01:57:40.070 --> 01:57:43.760
ignition models, propagation
models, but you know
01:57:43.760 --> 01:57:46.163
how are those, how are those built?
01:57:47.120 --> 01:57:50.590
And then my second
question is the projections
01:57:50.590 --> 01:57:52.180
that come out of those models.
01:57:52.180 --> 01:57:55.170
You know, what degree
of confidence do you have?
01:57:55.170 --> 01:57:59.790
Like you've shown where
you've got a 50% chance
01:57:59.790 --> 01:58:04.050
that the fire spreads to
this part of the topography.
01:58:04.050 --> 01:58:08.080
You know, what degree of
confidence does your models have
01:58:08.080 --> 01:58:10.580
for that kind of those projections?
01:58:10.580 --> 01:58:14.200
And then lastly, my
comment would be, you know
01:58:14.200 --> 01:58:15.460
where you were looking
01:58:15.460 --> 01:58:20.460
at from utility asset
ignitions, what would happen?
01:58:21.140 --> 01:58:24.180
You could conversely you
could start looking at typography
01:58:24.180 --> 01:58:28.630
and say, if you put utility
assets in this particular place
01:58:28.630 --> 01:58:32.154
you have a high degree
of a probability of having
01:58:32.154 --> 01:58:35.120
a very high consequence fire.
01:58:35.120 --> 01:58:37.210
And we could start creating maps
01:58:37.210 --> 01:58:40.643
of where perhaps not to put utility assets.
01:58:41.590 --> 01:58:42.893
So I'll leave it at that.
01:58:44.550 --> 01:58:46.950
Sure Tony, I can
address these if appropriate.
01:58:48.532 --> 01:58:49.860
Go ahead David.
01:58:49.860 --> 01:58:52.860
Yeah, with respect to the
models, you know, good question.
01:58:54.210 --> 01:58:56.130
The models that we have taken are based
01:58:56.130 --> 01:58:59.770
on published science, that's adopted
01:58:59.770 --> 01:59:03.450
and accepted by fire
agencies across the world.
01:59:03.450 --> 01:59:05.210
In particularly in North America, the basis
01:59:05.210 --> 01:59:09.388
of the propagation model
is the rock mill model.
01:59:09.388 --> 01:59:12.453
We've added significant
enhancements to do that.
01:59:12.453 --> 01:59:14.790
Some of those are very specific and detailed
01:59:14.790 --> 01:59:18.650
under crown fire analysis,
encroachment life fuel marches
01:59:18.650 --> 01:59:21.750
those kinds of things to
enhance those models.
01:59:21.750 --> 01:59:24.500
Those enhancements that are built on top
01:59:24.500 --> 01:59:27.728
of the core published science
are actually based on research
01:59:27.728 --> 01:59:30.220
and then operational testing that
01:59:30.220 --> 01:59:33.570
we've conducted in
concert with our customers.
01:59:33.570 --> 01:59:37.072
For example, the models
applied for this were calibrated
01:59:37.072 --> 01:59:41.970
against 10,000 fires in the 2020 fire season
01:59:41.970 --> 01:59:45.670
working directly with CAL Fire where over 500
01:59:45.670 --> 01:59:49.126
of those were very advanced
analysis of the model
01:59:49.126 --> 01:59:51.552
all those models working together
01:59:51.552 --> 01:59:54.486
with what was really observed and expected in
01:59:54.486 --> 01:59:58.040
particularly there's certain
data sources now available
01:59:58.040 --> 02:00:01.000
from the national guard and the military
02:00:01.000 --> 02:00:05.110
that allow us to capture
effected areas every 15 minutes.
02:00:05.110 --> 02:00:09.040
And we built machine
learning algorithms that model
02:00:09.040 --> 02:00:13.954
that tie the spread predictions
directly to what's observed.
02:00:13.954 --> 02:00:15.810
It's a revolutionary time.
02:00:15.810 --> 02:00:17.860
We've never had that in the past
02:00:17.860 --> 02:00:20.490
and our models have
been calibrated against that.
02:00:20.490 --> 02:00:21.995
It's an ongoing process.
02:00:21.995 --> 02:00:24.961
With relate and that addresses one, I think
02:00:24.961 --> 02:00:29.961
and two in with respect to that, you know
02:00:30.609 --> 02:00:35.410
this is as much about
operationalizing fire science
02:00:35.410 --> 02:00:37.237
as it is technology implementation.
02:00:37.237 --> 02:00:41.010
And in that regard, that's
something we take very seriously.
02:00:41.010 --> 02:00:42.873
I think we had nine published
02:00:42.873 --> 02:00:46.210
in accredited generals,
scientific publications
02:00:46.210 --> 02:00:50.360
and the past year alone
as a company to ensure
02:00:50.360 --> 02:00:54.450
that the science we're
implementing technology wise
02:00:54.450 --> 02:00:57.780
is robust and leading
edge and addressing some
02:00:57.780 --> 02:01:00.611
of those deficiencies
that we know in the model.
02:01:00.611 --> 02:01:05.611
And we use to adjust to calibration
02:01:08.630 --> 02:01:11.496
in the real world such
as the 2020 fire season
02:01:11.496 --> 02:01:14.990
that is an ongoing process
and continues to happen.
02:01:14.990 --> 02:01:17.707
A model is as good as, you know,
02:01:18.930 --> 02:01:20.630
like they say all models are wrong
02:01:21.520 --> 02:01:24.340
but it's understanding
the consistency of them.
02:01:24.340 --> 02:01:26.380
And so with a degree of confidence
02:01:26.380 --> 02:01:30.218
just to reiterate that point,
we very feel very good.
02:01:30.218 --> 02:01:32.380
Although refinements ongoing
02:01:32.380 --> 02:01:34.641
in an ongoing basis with new research
02:01:34.641 --> 02:01:38.120
our degree of competence is quite comfortable
02:01:38.120 --> 02:01:39.630
because this isn't a model that came
02:01:39.630 --> 02:01:40.980
out of the academic arena.
02:01:40.980 --> 02:01:43.380
This is a model that's been applied worldwide
02:01:43.380 --> 02:01:44.990
over the last 20 years.
02:01:44.990 --> 02:01:46.530
And in particularly, like I said
02:01:46.530 --> 02:01:50.101
in the 2020 fire season
really was the proving point
02:01:50.101 --> 02:01:52.216
that the technology works.
02:01:52.216 --> 02:01:55.440
It hits the mark, despite
propaganda out there
02:01:55.440 --> 02:01:57.720
in the press by some, the reality
02:01:57.720 --> 02:02:00.540
is CAL Fire can tell you
this model hit the mark.
02:02:00.540 --> 02:02:02.713
And so God has a level of credibility
02:02:02.713 --> 02:02:05.423
that comes back to this analysis.
02:02:05.423 --> 02:02:09.550
I'm gonna ask my partner, Dr. Ramirez Joaquin
02:02:09.550 --> 02:02:11.740
if there's anything else
you would like to add
02:02:11.740 --> 02:02:14.620
you're really the fire behavior
and modeling specialist.
02:02:14.620 --> 02:02:16.220
Are you able to respond to this?
02:02:17.210 --> 02:02:18.313
Yeah thank you.
02:02:18.313 --> 02:02:19.920
I just wanna emphasize
02:02:19.920 --> 02:02:24.230
and really reinforce what you just spread.
02:02:24.230 --> 02:02:26.420
And the amount of models that are included
02:02:26.420 --> 02:02:28.767
are really pretty significant.
02:02:28.767 --> 02:02:32.547
As implied in the question,
where are you seeing
02:02:32.547 --> 02:02:36.210
the leading leaving the best data possible?
02:02:36.210 --> 02:02:38.579
So the models from
example into on the weather
02:02:38.579 --> 02:02:42.210
is the best available weather
model, probably in the world.
02:02:42.210 --> 02:02:47.210
We are running at 108
hours during the last two years
02:02:48.580 --> 02:02:51.200
every day we test and
improve it in the field.
02:02:51.200 --> 02:02:52.851
We have two kilometer resolution.
02:02:52.851 --> 02:02:55.206
We use models to evaluate the
02:02:55.206 --> 02:02:59.310
what is the availability
of the vegetation to burn.
02:02:59.310 --> 02:03:01.490
And we found an analysis of 30 years
02:03:01.490 --> 02:03:06.490
of 40 years, actually, of
vegetation measurements
02:03:07.520 --> 02:03:09.420
on the field, coupled with
02:03:09.420 --> 02:03:11.700
the remote sensing imagery that
02:03:11.700 --> 02:03:13.680
we capture every day we have the load
02:03:13.680 --> 02:03:16.690
we have all those to
calculate the light moisture.
02:03:16.690 --> 02:03:18.800
As (indistinct) mentioned, they're probably
02:03:18.800 --> 02:03:23.607
and reviewed and their
new for this technology.
02:03:23.607 --> 02:03:24.780
And also on the fire spread
02:03:24.780 --> 02:03:26.180
we know that there are limitations.
02:03:26.180 --> 02:03:28.860
We know that in the future
we'll have new models
02:03:28.860 --> 02:03:31.320
hopefully that could, for example
02:03:31.320 --> 02:03:33.463
explain competitive situations
that are not addressed
02:03:33.463 --> 02:03:37.480
by the ExxonMobil's, but for our one day ran
02:03:37.480 --> 02:03:40.310
for the first day run that
these models have done
02:03:40.310 --> 02:03:43.830
we're very comfortable on the use.
02:03:43.830 --> 02:03:46.760
And obviously being able to be battle proven
02:03:46.760 --> 02:03:49.580
mentioned in the field we, during these year
02:03:49.580 --> 02:03:52.379
it really provides the
full credibility to the work
02:03:52.379 --> 02:03:54.445
knowing that there are limitations.
02:03:54.445 --> 02:03:56.707
And one thing I will say also is that
02:03:56.707 --> 02:04:00.421
during the last year and (indistinct)
02:04:00.421 --> 02:04:02.460
the we have what the Kincade fire
02:04:02.460 --> 02:04:04.220
on this model, where use on site
02:04:04.220 --> 02:04:07.166
on making Kincade fire
with extremely good results
02:04:07.166 --> 02:04:10.590
supporting operations from the common center.
02:04:10.590 --> 02:04:13.233
We know that there is a ton of work to do
02:04:13.233 --> 02:04:15.429
we're really anxious to get more
02:04:15.429 --> 02:04:17.809
better models from the scientists.
02:04:17.809 --> 02:04:20.450
But the reality is that we
need to provide answers
02:04:20.450 --> 02:04:23.750
today to prevent these situations to happen.
02:04:23.750 --> 02:04:27.220
And this is a very solid, solid
02:04:27.220 --> 02:04:29.810
consolidated modeling capabilities
02:04:29.810 --> 02:04:32.573
to provide some analysis to this problem.
02:04:35.040 --> 02:04:37.890
If I may, I'm gonna add one more point here,
02:04:37.890 --> 02:04:40.611
John you brought up an excellent point up in,
02:04:40.611 --> 02:04:43.410
independent of this analysis.
02:04:43.410 --> 02:04:46.120
Could we not use these models, you know
02:04:46.120 --> 02:04:47.973
in more of a proactive sense?
02:04:49.650 --> 02:04:51.050
Really I think what you're getting at
02:04:51.050 --> 02:04:54.160
is the ability to forecast risk in advance.
02:04:54.160 --> 02:04:56.544
So in this regard, we applied it directly
02:04:56.544 --> 02:04:59.880
to ignition points from our electric utility.
02:04:59.880 --> 02:05:01.283
However, for CAL Fire
02:05:01.283 --> 02:05:05.088
we run about 144 million simulations nightly
02:05:05.088 --> 02:05:07.810
with hourly weather data, looking
02:05:07.810 --> 02:05:11.710
out over a hundred hours with ignitions,
02:05:11.710 --> 02:05:14.070
a thousand meters across all of California
02:05:14.070 --> 02:05:17.400
that allows us to forecast risk in advance,
02:05:17.400 --> 02:05:18.820
independent, nothing to do
02:05:18.820 --> 02:05:21.980
with electric utility as any ignition source.
02:05:21.980 --> 02:05:23.710
CAL Fire has adopted this
02:05:23.710 --> 02:05:26.960
over the 2020 fire season is using it as part
02:05:26.960 --> 02:05:30.930
of their input to set readiness
levels, resource allocation
02:05:30.930 --> 02:05:33.500
independent of when, before fires occur.
02:05:33.500 --> 02:05:35.010
So your point's very well made.
02:05:35.010 --> 02:05:37.160
It is being applied for risk forecasting
02:05:37.160 --> 02:05:39.410
which I think is the most critical component.
02:05:41.017 --> 02:05:42.028
Thank you.
02:05:42.028 --> 02:05:45.020
I appreciate those answers.
02:05:45.020 --> 02:05:47.804
When it's specifically to
the degree of confidence
02:05:47.804 --> 02:05:52.340
you showed a couple of
fire propagation projections
02:05:52.340 --> 02:05:57.000
for during certain PSPS events as a result
02:05:57.000 --> 02:05:59.300
of damaged utility equipment.
02:05:59.300 --> 02:06:03.570
And you had a line where
you were like 50% chance
02:06:03.570 --> 02:06:06.330
that it will reach this
line and these buildings
02:06:06.330 --> 02:06:10.230
would be impacted like,
is it 50% plus or minus 5%,
02:06:10.230 --> 02:06:15.230
50% plus or minus 20% of that kind of degree
02:06:15.240 --> 02:06:19.953
of confidence in those type
of propagation projections.
02:06:21.040 --> 02:06:22.960
I'll let Dr. Ramirez reviewed that
02:06:22.960 --> 02:06:25.250
with respect to the probabilistic.
02:06:25.250 --> 02:06:28.376
Yeah the probabilistic is basically
02:06:28.376 --> 02:06:30.440
trying to get someone to cover approach.
02:06:30.440 --> 02:06:33.530
Basically we're running from
the same admission point.
02:06:33.530 --> 02:06:34.870
We change the conditions
02:06:34.870 --> 02:06:38.140
because there's potentially some trinity
02:06:38.140 --> 02:06:41.200
in the input data regarding
the weather, right?
02:06:41.200 --> 02:06:43.330
So basically the result that we provide is
02:06:43.330 --> 02:06:48.320
when we saved this a 50% of
impact on them, a specific area.
02:06:48.320 --> 02:06:51.020
It means that the 100 simulations that we
02:06:51.020 --> 02:06:55.160
created that area falls into what's impacted
02:06:55.160 --> 02:06:58.440
and at least 50 of them.
02:06:58.440 --> 02:07:01.320
And we'll go to the 5% area and the largest
02:07:01.320 --> 02:07:03.836
that is covering the latest fire sheds,
02:07:03.836 --> 02:07:06.696
it means that only five of those simulations
02:07:06.696 --> 02:07:09.330
where we change say which is the conditions
02:07:09.330 --> 02:07:12.370
were impacting was impact in that area.
02:07:12.370 --> 02:07:13.657
It's not a level of confidence.
02:07:13.657 --> 02:07:16.250
It's a level of uncertainty that we tried
02:07:16.250 --> 02:07:18.493
to accomplish with this approach.
02:07:19.770 --> 02:07:21.170
Understood, thank you.
02:07:22.470 --> 02:07:25.303
Thank you very much
operator next caller, please.
02:07:26.520 --> 02:07:28.510
Our next color is Jacqueline Err,
02:07:28.510 --> 02:07:29.380
your line is open.
02:07:29.380 --> 02:07:30.213
Please go ahead.
02:07:31.420 --> 02:07:32.253
Yes, thank you.
02:07:32.253 --> 02:07:34.810
This is Jacquelyn Err,
with the acting town council,
02:07:34.810 --> 02:07:37.320
I'm struggling to reconcile your statements
02:07:37.320 --> 02:07:39.060
regarding confidence in your model
02:07:39.060 --> 02:07:40.800
with very recent instances
02:07:40.800 --> 02:07:43.109
which fire consequence models run by Edison
02:07:43.109 --> 02:07:45.450
and SDG&E were highly over predictive
02:07:45.450 --> 02:07:50.020
and they substantially
overestimated actual fire spread.
02:07:50.020 --> 02:07:51.450
Much of the error pertains
02:07:51.450 --> 02:07:53.610
to assumptions regarding input factors
02:07:53.610 --> 02:07:56.770
such as firefighting
capability and other elements.
02:07:56.770 --> 02:08:00.070
Given this, my first
question is how can we stay
02:08:00.070 --> 02:08:01.720
Calders have confidence in the use
02:08:01.720 --> 02:08:04.900
of your model to make
decisions on PSPS events
02:08:04.900 --> 02:08:07.650
that continuously turn our lives upside down
02:08:07.650 --> 02:08:09.509
and cause devastating impacts.
02:08:09.509 --> 02:08:12.970
The second question is
you showed many simulations
02:08:12.970 --> 02:08:14.380
of wildfires that could have occurred
02:08:14.380 --> 02:08:17.290
on equipment that was reported as damaged.
02:08:17.290 --> 02:08:20.270
One of your simulations
addressed a de-energization event
02:08:20.270 --> 02:08:23.115
by Edison October 2nd,
but the post event reports
02:08:23.115 --> 02:08:25.665
filed with the Commission in 2019
02:08:25.665 --> 02:08:27.811
do not indicate that Edison actually
02:08:27.811 --> 02:08:30.085
de-energized any circuits on October 2nd.
02:08:30.085 --> 02:08:31.643
So that's troubling.
02:08:31.643 --> 02:08:34.161
And also Edison's reports have shown that
02:08:34.161 --> 02:08:37.258
its equipment has been
damaged at very low wind speeds.
02:08:37.258 --> 02:08:40.807
So do you have data describing the damages
02:08:40.807 --> 02:08:43.910
on the equipment
involved in your simulations,
02:08:43.910 --> 02:08:46.717
for example what was
damaged, how it was damaged
02:08:46.717 --> 02:08:49.370
and what would wind speed caused the damage
02:08:49.370 --> 02:08:51.490
and also will the maps and conclusions
02:08:51.490 --> 02:08:54.370
and data that you've presented
be compiled into a report.
02:08:54.370 --> 02:08:55.513
It could be accessed by the public so
02:08:55.513 --> 02:08:58.330
that we can take our time
to consider it more fully.
02:08:58.330 --> 02:08:59.330
Thank you very much.
02:09:04.400 --> 02:09:05.766
Thank you.
02:09:05.766 --> 02:09:06.870
Quick, real quick.
02:09:06.870 --> 02:09:08.300
As far as the reports,
02:09:08.300 --> 02:09:09.940
yes the intent is to post them
02:09:09.940 --> 02:09:13.343
in the near future on the CPUC website.
02:09:14.210 --> 02:09:15.570
Can't give you an exact date right now
02:09:15.570 --> 02:09:17.880
but that is the intent
is to post the reports.
02:09:17.880 --> 02:09:19.593
David, do you have something?
02:09:21.130 --> 02:09:23.930
Yeah, with respect to the STG and SCE
02:09:23.930 --> 02:09:27.322
and their decisions,
we're not actively involved
02:09:27.322 --> 02:09:29.132
in making those decisions.
02:09:29.132 --> 02:09:33.010
So I can't really comment
on what level of the fire
02:09:33.010 --> 02:09:36.210
spread modeling they use
to make those decisions.
02:09:36.210 --> 02:09:40.090
You know, that'd be a question
we'd have to pose to them
02:09:40.090 --> 02:09:42.583
with respect to confidence in the model,
02:09:42.583 --> 02:09:47.030
we have significant
confidence in the model again,
02:09:47.030 --> 02:09:51.960
to reiterate the model has not just been used
02:09:51.960 --> 02:09:53.520
for this analysis.
02:09:53.520 --> 02:09:55.740
It's consistently been
adopted across the state
02:09:55.740 --> 02:09:59.450
of California is the best
available model at this time
02:09:59.450 --> 02:10:02.522
and endorsed authoritatively by CAL Fire.
02:10:02.522 --> 02:10:06.040
And I think as Dr. Ramirez
said, battle-tested
02:10:06.040 --> 02:10:08.930
over the 2020 fire season from now,
02:10:08.930 --> 02:10:11.810
we have good confidence in the model.
02:10:11.810 --> 02:10:15.180
Having said that we
continually find opportunities
02:10:15.180 --> 02:10:17.150
for improvement and refinement
02:10:17.150 --> 02:10:20.553
with respect to estimation impacts.
02:10:20.553 --> 02:10:22.867
Part of that has been significant.
02:10:22.867 --> 02:10:26.569
The increase in the
ability to define better fuels
02:10:26.569 --> 02:10:28.761
data using LIDAR information.
02:10:28.761 --> 02:10:32.439
That's very active with us again,
02:10:32.439 --> 02:10:37.439
better data outputs
focusing on that in particular
02:10:37.700 --> 02:10:40.423
and in particular, the
encroachment algorithms,
02:10:40.423 --> 02:10:44.920
we're working on a project
now with two of our scientists
02:10:44.920 --> 02:10:48.749
to really better deal with matching
02:10:48.749 --> 02:10:51.080
what's observed and been expected
02:10:51.080 --> 02:10:53.217
through a lot of the NIST and other analysis
02:10:53.217 --> 02:10:56.580
and particularly using the damage data
02:10:56.580 --> 02:11:01.290
by CAL Fire to better match the encroachment
02:11:01.290 --> 02:11:05.240
and the actual potential
of consequence out there.
02:11:05.240 --> 02:11:08.647
With regards to the other
impacts, which I fully understand
02:11:08.647 --> 02:11:12.020
and do not disagree
with, with respect to PSPS
02:11:12.020 --> 02:11:14.320
again, not part of the scope of the contract
02:11:14.320 --> 02:11:17.550
and not something to be incorporated by us.
02:11:17.550 --> 02:11:20.840
Although I can appreciate
that in evaluation or analysis
02:11:20.840 --> 02:11:25.120
of those impacts is perhaps
the other side of the equation.
02:11:25.120 --> 02:11:27.020
It's not just about the consequences of what
02:11:27.020 --> 02:11:31.853
could have been, hopefully
that provides some answers.
02:11:33.446 --> 02:11:35.046
I would like just to add value
02:11:37.395 --> 02:11:39.730
or even clarification also in
the comment on the question
02:11:39.730 --> 02:11:43.890
about the potential estimation of the results
02:11:43.890 --> 02:11:46.300
of the impact analysis, because yes,
02:11:46.300 --> 02:11:49.707
we haven't included in this simulations,
02:11:49.707 --> 02:11:52.830
there are not fires they're simulations.
02:11:52.830 --> 02:11:55.850
We haven't included any
kind of suppression activities
02:11:55.850 --> 02:11:58.970
or complexity of the
development of this suppression
02:11:58.970 --> 02:12:01.160
activity that may happen in all of them.
02:12:01.160 --> 02:12:03.183
We wanna I compare apples to apples,
02:12:03.183 --> 02:12:06.097
because the reality as we saw these year
02:12:06.097 --> 02:12:10.464
the potential activity
of our ground resources
02:12:10.464 --> 02:12:17.360
is heavily also not
modified but the number
02:12:17.360 --> 02:12:19.650
up talking to this that may be taking place
02:12:19.650 --> 02:12:22.210
we'll just have to remember
that these fire season
02:12:22.210 --> 02:12:25.979
in our (indistinct) fires that
started at the same time.
02:12:25.979 --> 02:12:29.670
This has been absolutely
extraordinary in terms
02:12:29.670 --> 02:12:33.402
of productivity and discuss their resources,
02:12:33.402 --> 02:12:36.287
start to happen when we have some
02:12:36.287 --> 02:12:38.286
(indistinct) of event and these situations
02:12:38.286 --> 02:12:43.286
that we've already build in 2019 (indistinct)
02:12:43.820 --> 02:12:46.235
are telling us that the
times that we're living
02:12:46.235 --> 02:12:48.973
in terms of conditions
are pretty extraordinary.
02:12:48.973 --> 02:12:52.080
And that's the new time that we live in.
02:12:52.080 --> 02:12:54.910
I agree that obviously we add
02:12:54.910 --> 02:12:59.060
if we may add suppression
activities on all of them
02:12:59.060 --> 02:13:01.775
those potential impacts could be reduced
02:13:01.775 --> 02:13:03.640
but that's (indistinct).
02:13:03.640 --> 02:13:05.180
Okay, wait.
02:13:05.180 --> 02:13:06.600
So just so I understand these models
02:13:06.600 --> 02:13:10.270
the simulations you've shown us do not factor
02:13:10.270 --> 02:13:13.590
in suppression, but that's the element
02:13:13.590 --> 02:13:15.380
of all the modeling that's being done
02:13:15.380 --> 02:13:17.760
in the wildfire mitigation plan.
02:13:17.760 --> 02:13:20.850
So, okay.
02:13:20.850 --> 02:13:23.450
And how about the damage to the equipment
02:13:23.450 --> 02:13:26.101
data pertaining to the damaged equipment
02:13:26.101 --> 02:13:31.101
that you used to determine where fires,
02:13:31.690 --> 02:13:34.663
the fire do the simulations.
02:13:36.020 --> 02:13:40.762
So that data was obtained
by the CPUC from the IOUs
02:13:40.762 --> 02:13:45.660
based on my understanding
of their regulatory requirements
02:13:45.660 --> 02:13:50.040
of conducting surveys after PSPS events
02:13:50.040 --> 02:13:54.040
that data is done by field engineers, GPS.
02:13:54.040 --> 02:13:56.410
We had each of the IOUs describing detailed
02:13:56.410 --> 02:13:58.447
or data collection methods
02:13:58.447 --> 02:14:01.400
and their analysis methods of that data.
02:14:01.400 --> 02:14:03.950
One part of it is location was very accurate
02:14:03.950 --> 02:14:05.630
of course cause it was GPS.
02:14:05.630 --> 02:14:09.530
The other factor is when did the damage occur
02:14:09.530 --> 02:14:14.530
and that's where we got
estimations from the IOU engineers.
02:14:14.900 --> 02:14:18.790
And then we conducted our
own detailed weather analysis
02:14:18.790 --> 02:14:21.344
working with San Jose state's fire weather
02:14:21.344 --> 02:14:23.930
to identify if they made sense
02:14:23.930 --> 02:14:25.740
or whether they needed to be adjusted.
02:14:25.740 --> 02:14:27.820
In some scenarios they were adjusted
02:14:27.820 --> 02:14:30.045
to reflect worst case scenarios.
02:14:30.045 --> 02:14:31.760
But the reality is
02:14:31.760 --> 02:14:34.150
that they don't collect data are unable to
02:14:34.150 --> 02:14:37.650
I believe collect data exactly
when the damage occurs.
02:14:37.650 --> 02:14:39.420
And so that is an estimation
02:14:39.420 --> 02:14:41.260
but we did receive all that detailed data
02:14:41.260 --> 02:14:43.997
of all the locations and
all the characteristics
02:14:43.997 --> 02:14:47.630
along with photographs of the damage as well.
02:14:47.630 --> 02:14:49.260
Can you make that data available
02:14:49.260 --> 02:14:52.090
as part of what you post for public review?
02:14:52.090 --> 02:14:54.883
I would really appreciate
it, including pictures.
02:14:58.180 --> 02:15:00.227
Not up to me to provide really up to the CPUC
02:15:00.227 --> 02:15:03.740
of the data obtained by them from the IOU.
02:15:03.740 --> 02:15:05.853
So I'll push that over the fence to Tony.
02:15:07.400 --> 02:15:09.110
And this damage should be reported
02:15:09.110 --> 02:15:10.779
as part of the post event report
02:15:10.779 --> 02:15:12.900
but we'll follow up.
02:15:12.900 --> 02:15:13.733
Thank you very much.
02:15:13.733 --> 02:15:15.700
The confusion because it doesn't look
02:15:15.700 --> 02:15:18.518
like Edison de-energized on October 2nd.
02:15:18.518 --> 02:15:20.723
So that's why I'm asking the question.
02:15:22.400 --> 02:15:24.610
I can't answer the
right I'll have to follow up.
02:15:24.610 --> 02:15:25.443
Thank you.
02:15:28.080 --> 02:15:29.680
Operator, next caller, please.
02:15:32.930 --> 02:15:35.090
Our next caller is Joseph Mitchell.
02:15:35.090 --> 02:15:37.700
Your line is open, please go ahead.
02:15:37.700 --> 02:15:39.373
Hi, good morning.
02:15:40.705 --> 02:15:45.577
I have a couple of questions
regarding the fire growth
02:15:47.660 --> 02:15:52.050
on the, the large end of
the fire, the catastrophic
02:15:52.050 --> 02:15:57.050
end of the fire most utility losses are due
02:15:57.290 --> 02:16:02.190
to catastrophic fires and
it would have been useful.
02:16:03.820 --> 02:16:06.430
I was wondering if you
had any supplemental slides
02:16:06.430 --> 02:16:11.430
or anything to see some either examples
02:16:11.900 --> 02:16:15.515
or statistical validation of how your
02:16:15.515 --> 02:16:20.515
your models work on the catastrophic end
02:16:21.470 --> 02:16:26.470
a related question, you
terminate your runs at 24 hours.
02:16:28.490 --> 02:16:31.800
There have been a
number of catastrophic fires
02:16:31.800 --> 02:16:36.690
which continued to do catastrophic damage
02:16:36.690 --> 02:16:40.950
in the time after 24 hours.
02:16:40.950 --> 02:16:45.950
So do you have any
supplemental information you
02:16:46.130 --> 02:16:48.640
could share regarding the calibration?
02:16:48.640 --> 02:16:52.550
Because I know you're
calibrating on the, you know
02:16:52.550 --> 02:16:57.550
the eight hour or 24 hour
level for immediate fire risk
02:16:58.300 --> 02:17:00.869
but this is more because the utilities
02:17:00.869 --> 02:17:04.379
and the CPUC are looking at
how much damage can be done.
02:17:04.379 --> 02:17:07.298
Where do we have to be worried about damaging
02:17:07.298 --> 02:17:09.710
if we're gonna be doing a shut off
02:17:09.710 --> 02:17:12.001
or if we're gonna be hardening lines
02:17:12.001 --> 02:17:15.023
what kind of information
can you share on that?
02:17:19.540 --> 02:17:24.540
Yes, Joe, with respect to
the 24 hours was identified
02:17:24.540 --> 02:17:28.150
as an extent that we thought was reasonable
02:17:28.150 --> 02:17:32.080
to accommodate those
types of destructive fires.
02:17:32.080 --> 02:17:35.250
That could be longer duration well understood
02:17:35.250 --> 02:17:37.210
that many of those fires
02:17:37.210 --> 02:17:39.550
when they become
extended attack fires can grow
02:17:39.550 --> 02:17:42.338
as we saw this year well beyond that amounts
02:17:42.338 --> 02:17:45.990
but we needed some consistent comparison.
02:17:45.990 --> 02:17:48.773
The reality is in an operational context,
02:17:50.208 --> 02:17:53.640
fires are run using real-time observation
02:17:53.640 --> 02:17:56.960
and short term prediction
data and consistently run.
02:17:56.960 --> 02:17:59.260
So for example, when an agency such
02:17:59.260 --> 02:18:01.140
as CAL Fire uses our software
02:18:01.140 --> 02:18:03.570
the model they're modeling all the time
02:18:03.570 --> 02:18:05.190
because we're using predictive data
02:18:05.190 --> 02:18:07.280
but we're receiving observations
02:18:07.280 --> 02:18:11.860
and prediction often as good,
sometimes misses the mark.
02:18:11.860 --> 02:18:13.920
And so we need to be able to calibrate
02:18:13.920 --> 02:18:15.640
on the fly and adjust on the fly.
02:18:15.640 --> 02:18:17.451
So in an operational scenario
02:18:17.451 --> 02:18:20.843
we don't go modeling
out 72 hours or something.
02:18:20.843 --> 02:18:25.680
It's typically a burning period,
eight hours of that nature.
02:18:25.680 --> 02:18:29.840
We may do some
probabilistic runs over 24, 48, 72
02:18:29.840 --> 02:18:32.260
to get a feeling aware of going, but we model
02:18:32.260 --> 02:18:33.700
on the fly all the time.
02:18:33.700 --> 02:18:35.790
'Cause we have data coming in our model
02:18:35.790 --> 02:18:38.950
accommodates those adjustments
based on what's observed
02:18:38.950 --> 02:18:41.490
in the field and can adjust
rate a spread Packers
02:18:41.490 --> 02:18:44.570
on the fly with a machine
learning capability.
02:18:44.570 --> 02:18:47.590
So that's a little different
that operational scenario
02:18:47.590 --> 02:18:49.570
than what we're doing here, here we're really
02:18:49.570 --> 02:18:52.270
taking advantage of the best available,
02:18:52.270 --> 02:18:55.485
predictive weather data
and fuels data that we have
02:18:55.485 --> 02:19:00.485
over a horizon we felt moving beyond 24 hours
02:19:01.460 --> 02:19:04.440
would really be stretching perhaps
02:19:04.440 --> 02:19:09.440
the reliability of that data
beyond 70 to 80, 48 hours.
02:19:10.410 --> 02:19:12.110
And that decision was made collectively
02:19:12.110 --> 02:19:13.960
within the CPUC and the modeling team.
02:19:13.960 --> 02:19:18.173
The 24 hours represented
a long enough scenario.
02:19:18.173 --> 02:19:20.700
So that's why it was chosen.
02:19:20.700 --> 02:19:23.301
And that's how it's distinction between
02:19:23.301 --> 02:19:27.647
how the modeling is used
operationally versus a sense here
02:19:27.647 --> 02:19:30.590
where we get have a consistent comparison.
02:19:31.634 --> 02:19:34.940
Okay, so do you have the, you know
02:19:34.940 --> 02:19:36.870
something where you can show, you know
02:19:36.870 --> 02:19:39.992
at 24 hours even going out 24 hours,
02:19:39.992 --> 02:19:43.698
here's how we compare
against other real fires
02:19:43.698 --> 02:19:46.000
that have burned for 24 hours?
02:19:46.000 --> 02:19:47.683
Sure, sure.
02:19:48.723 --> 02:19:51.445
If I may add suddenly on that regard part
02:19:51.445 --> 02:19:55.709
of analysis obviously was
trying to set up the scenario
02:19:55.709 --> 02:20:00.709
setting up by looking up the
past events that happened
02:20:01.850 --> 02:20:06.010
that cross disruptive situations
in a fresh burning period.
02:20:06.010 --> 02:20:08.580
And we calibrated them all day long.
02:20:08.580 --> 02:20:12.060
We displayed the comparison
with events, like for example
02:20:12.060 --> 02:20:15.983
Debra Fields fire 1991,
one that was very significant
02:20:15.983 --> 02:20:19.240
and was also the tops fire,
which in the first six hours
02:20:19.240 --> 02:20:21.855
of the run costs dramatically impacts
02:20:21.855 --> 02:20:24.853
and the campfire obviously.
02:20:26.458 --> 02:20:29.003
During this year 2020 in scenario
02:20:29.003 --> 02:20:32.272
that we are my life in the past,
02:20:32.272 --> 02:20:36.670
we found that first hump of
the glass fire was really much
02:20:36.670 --> 02:20:40.304
in what we were predicting
in a situation like that.
02:20:40.304 --> 02:20:42.170
Where in the first round
02:20:42.170 --> 02:20:43.860
we could have some significant impact.
02:20:43.860 --> 02:20:47.090
So yes, we be with the files and we'll try to
02:20:47.090 --> 02:20:50.923
that's one of the reasons that
we do go beyond the 24 hours.
02:20:53.910 --> 02:20:54.810
Okay, thank you.
02:20:57.660 --> 02:21:00.050
I forgot to mention that
which fire in San Diego
02:21:00.050 --> 02:21:02.100
by the way was also a subject
02:21:02.100 --> 02:21:05.370
of analysis witch (indistinct)
02:21:08.870 --> 02:21:09.703
Thanks.
02:21:12.660 --> 02:21:15.240
Once again, if you would
like to make a public comment
02:21:15.240 --> 02:21:17.820
please press star one, unmute your phone
02:21:17.820 --> 02:21:20.734
and clearly record your
name and organization.
02:21:20.734 --> 02:21:23.740
Our next comment comes from Ian Fisher.
02:21:23.740 --> 02:21:24.670
Your line is open.
02:21:24.670 --> 02:21:25.503
Please go ahead.
02:21:26.420 --> 02:21:27.740
Hi, it's me again.
02:21:27.740 --> 02:21:28.940
CAl advocate's office.
02:21:28.940 --> 02:21:33.090
I just want to clarify something
about this 24 hour run.
02:21:33.090 --> 02:21:38.090
So I'm trying to understand,
I totally appreciate the
02:21:38.208 --> 02:21:41.950
old model is working very
effective and fully calibrated.
02:21:41.950 --> 02:21:45.060
What I'm trying to
understand is its relationship
02:21:45.060 --> 02:21:49.130
to actual public policy and
the decisions to make PSPS
02:21:49.130 --> 02:21:52.903
you know to initiate a PSPS event.
02:21:54.782 --> 02:21:58.010
Were you given any scope in
making the decisions in setting
02:21:58.010 --> 02:22:02.970
up these scenarios by the
PUC as far as what measures
02:22:02.970 --> 02:22:06.430
should be used and what
scenario should be used
02:22:06.430 --> 02:22:10.810
to determine the criticality of an event?
02:22:10.810 --> 02:22:13.150
I'm guessing I'm trying
to work out why 24 hours
02:22:13.150 --> 02:22:16.310
and not six hours and four
hours for your model run.
02:22:16.310 --> 02:22:20.660
Surely it's about saving
life and land as a priority.
02:22:20.660 --> 02:22:24.925
And so the four hour and
six hour runs would be way
02:22:24.925 --> 02:22:29.760
more important to my mind than a 24 hour run.
02:22:29.760 --> 02:22:32.723
Can you kind of explain
that a little more again?
02:22:34.740 --> 02:22:35.573
Sure.
02:22:35.573 --> 02:22:38.810
Related to to public policy
02:22:38.810 --> 02:22:42.890
and how this might be used
to make PSPS decisions.
02:22:42.890 --> 02:22:45.630
No, we, that was not part of the scope.
02:22:45.630 --> 02:22:46.463
There was no discussion
02:22:46.463 --> 02:22:48.390
and we were provided no guidance on that.
02:22:48.390 --> 02:22:50.760
Now we were really focusing on applying
02:22:50.760 --> 02:22:54.360
the modeling to try to
quantify potential impacts.
02:22:54.360 --> 02:22:56.376
Of course, with running a 24 hour
02:22:56.376 --> 02:23:00.160
we just don't have the results for a 24 hour.
02:23:00.160 --> 02:23:03.520
We have the results for like every 10 minutes
02:23:03.520 --> 02:23:06.210
or at least every hour for the 24.
02:23:06.210 --> 02:23:10.942
So we can extract any
subset of that, that we want.
02:23:10.942 --> 02:23:15.942
We felt 24 hours represented more of a
02:23:15.982 --> 02:23:18.010
I believe the term gentleman
use was catastrophic.
02:23:18.010 --> 02:23:21.930
We'd like to use the term
destructive would allow us to
02:23:21.930 --> 02:23:24.260
accommodate what would be destructive
02:23:24.260 --> 02:23:27.650
beyond an eight hour
period of first burning period
02:23:27.650 --> 02:23:31.980
particularly beyond,
especially beyond initial attack.
02:23:31.980 --> 02:23:35.360
And that's where our initial
attack assessment index
02:23:35.360 --> 02:23:38.350
comes to play to help us better understand
02:23:38.350 --> 02:23:42.090
the 24 hours of impact valid.
02:23:42.090 --> 02:23:44.120
And should we ignore those
02:23:44.120 --> 02:23:47.760
because we'll never get to
24 hours again, not having data
02:23:47.760 --> 02:23:51.373
on suppression effectiveness or availability.
02:23:52.676 --> 02:23:56.290
We had the rather
consistent, no, there's a policy
02:23:56.290 --> 02:23:59.330
no guidance on that, but
hopefully that helped a little bit
02:23:59.330 --> 02:24:01.563
on the why 24 and not eight.
02:24:02.430 --> 02:24:05.050
So for a follow up question
02:24:05.050 --> 02:24:09.990
would your ranking change
if it was six or eight hours?
02:24:09.990 --> 02:24:12.727
So you rank and present
the worst case scenarios
02:24:12.727 --> 02:24:15.660
and the worst fires with that ranking change,
02:24:15.660 --> 02:24:20.390
if you change the actual
the run length so to speak.
02:24:20.390 --> 02:24:22.640
Absolutely remember again, we have data
02:24:22.640 --> 02:24:27.050
for all of the hours as
you saw, if a visual like that
02:24:27.050 --> 02:24:29.970
remember the flame length
and the rate of spread charts.
02:24:29.970 --> 02:24:31.450
While we have all that information
02:24:31.450 --> 02:24:33.680
about how much burned each hour, we asked you
02:24:33.680 --> 02:24:35.280
we have all the consequence
02:24:35.280 --> 02:24:37.310
of what burned each hour as well.
02:24:37.310 --> 02:24:38.832
And so yes, absolutely that data
02:24:38.832 --> 02:24:43.480
the beauty of this is to look
02:24:43.480 --> 02:24:47.440
at different periods and identify, you know
02:24:47.440 --> 02:24:49.330
we just wanted to look
at the first burning period
02:24:49.330 --> 02:24:51.840
which were potentially the worst ones,
02:24:51.840 --> 02:24:54.440
but again we want it to look at events
02:24:54.440 --> 02:24:55.880
played out a little longer and say
02:24:55.880 --> 02:24:57.890
what could be catastrophic things
02:24:57.890 --> 02:25:00.930
any of those simulations could be evaluated
02:25:00.930 --> 02:25:03.150
within and statistical analyze
02:25:03.150 --> 02:25:05.440
with respect to shorter
timeframes and say, well
02:25:05.440 --> 02:25:07.912
how many of those impacts
for us, you know, for me
02:25:07.912 --> 02:25:10.040
period of eight hours, absolutely.
02:25:10.040 --> 02:25:11.670
That's a possible opportunity
02:25:11.670 --> 02:25:14.173
for improvement and use of the data.
02:25:15.540 --> 02:25:17.800
So just to clarify, okay,
thank you very much.
02:25:17.800 --> 02:25:19.300
That generally answers the question.
02:25:19.300 --> 02:25:21.660
So as you understand,
we are in a quandary here
02:25:21.660 --> 02:25:24.580
we trapped between one catastrophic event
02:25:24.580 --> 02:25:26.140
and another potentially catastrophic event.
02:25:26.140 --> 02:25:29.660
PSPS does have very serious
economic conceptual packs
02:25:29.660 --> 02:25:31.080
as do wildfires.
02:25:31.080 --> 02:25:33.950
And so we really have to strike that balance
02:25:33.950 --> 02:25:36.423
about which one you want to live with.
02:25:37.830 --> 02:25:41.900
And so it's kind of, the
immediacy is more important
02:25:41.900 --> 02:25:45.610
to me in your report are you being asked
02:25:45.610 --> 02:25:49.170
to provide the four hour or 8 hour data
02:25:49.170 --> 02:25:52.812
and that the rank order of
events as part of that, as well
02:25:52.812 --> 02:25:57.493
as a 24 hour data, 'cause
the 24 hour data just strikes me
02:25:57.493 --> 02:25:58.630
that you're locking yourself
02:25:58.630 --> 02:26:00.530
into a dark room and scaring yourself.
02:26:02.170 --> 02:26:05.520
Yeah again, it's not
one block of 24 hour data.
02:26:05.520 --> 02:26:07.170
It's a series of observations
02:26:07.170 --> 02:26:10.310
throughout all the hours that make up a 24.
02:26:10.310 --> 02:26:14.962
So by providing the data
set for each simulation
02:26:14.962 --> 02:26:19.962
you inherently obtain all the
data on a per hourly basis.
02:26:20.150 --> 02:26:23.202
So any of those.
Understood.
02:26:23.202 --> 02:26:24.810
So yeah it is a rich data set
02:26:24.810 --> 02:26:28.490
that could be mined and
every all the core data
02:26:28.490 --> 02:26:31.280
that we used input and
output a great analysis
02:26:31.280 --> 02:26:33.872
as a course of deliverable to the CPUC.
02:26:33.872 --> 02:26:38.360
I fully appreciate your
comments about the policy.
02:26:38.360 --> 02:26:42.380
Again, our scope was
specifically on a fire analysis
02:26:42.380 --> 02:26:45.870
so I'll have to defer any
feedback to that, to the CPUC.
02:26:45.870 --> 02:26:48.900
Sure, so, I mean, but you are being asked to
02:26:48.900 --> 02:26:50.793
produce a report, is that correct?
02:26:52.660 --> 02:26:55.967
Yes, we've generated
reports in concert with CPUC
02:26:55.967 --> 02:26:58.867
and I think Tony can provide
more information about those.
02:26:59.750 --> 02:27:03.590
Okay and do those
reports include six hour burn
02:27:03.590 --> 02:27:05.340
or four hour burn data?
02:27:05.340 --> 02:27:08.450
I mean, do they include
summaries of that information?
02:27:08.450 --> 02:27:12.310
So you can contrast it with the 24 hour data
02:27:13.290 --> 02:27:14.890
without having to mine the data?
02:27:15.833 --> 02:27:18.180
The analysis focused
on the results comparison
02:27:18.180 --> 02:27:22.410
of the results of a full 24
hours, although the data does.
02:27:22.410 --> 02:27:23.602
So there is no summary
02:27:23.602 --> 02:27:27.358
so there's no summary reports of 6 hour data.
02:27:27.358 --> 02:27:29.358
I'd have to go in myself and analyze it.
02:27:30.730 --> 02:27:33.320
Yes, there's that was not a agreed upon,
02:27:33.320 --> 02:27:36.080
is that the scope the focus
was on the summarizing
02:27:36.080 --> 02:27:38.843
and analyzing the results
for 24 hours as we showed.
02:27:39.943 --> 02:27:41.443
Okay, thank you.
02:27:45.970 --> 02:27:48.530
Our next comment comes from Janet Cohen.
02:27:48.530 --> 02:27:49.460
Your line is open.
02:27:49.460 --> 02:27:50.293
Please go ahead.
02:27:51.310 --> 02:27:52.870
Well, this is Janice Cohen.
02:27:52.870 --> 02:27:57.770
I'm a fire behavior,
atmospheric scientist researcher.
02:27:57.770 --> 02:28:00.920
I've been working on developing a couple
02:28:00.920 --> 02:28:04.813
of weather fire models for 25 years.
02:28:05.950 --> 02:28:08.880
This work, the simulation of hypothetically
02:28:08.880 --> 02:28:12.837
missions only really has
meaning or value to the audience.
02:28:12.837 --> 02:28:16.133
If the accuracy of the fire growth model
02:28:16.133 --> 02:28:19.750
on actual events is publicly shown.
02:28:19.750 --> 02:28:21.700
This has not been done.
02:28:21.700 --> 02:28:23.760
I wanna repeat that the public
02:28:23.760 --> 02:28:27.315
and research community
have been shown no evidence
02:28:27.315 --> 02:28:31.380
about how accurate this model is as applied
02:28:31.380 --> 02:28:34.090
to growth of many actual fires
02:28:34.090 --> 02:28:37.522
across varying conditions, including today.
02:28:37.522 --> 02:28:41.480
In fact, these antiquated modeling methods
02:28:41.480 --> 02:28:45.380
are well-known to provide poor representation
02:28:45.380 --> 02:28:47.160
of actual fire growth
02:28:47.160 --> 02:28:50.488
because they fail to
include two crucial factors
02:28:50.488 --> 02:28:55.488
micro-scale that is hundreds
of meters, wind circulations
02:28:55.890 --> 02:28:59.158
and topographic effects
that very second by second
02:28:59.158 --> 02:29:04.158
not hour by hour on kilometers scales.
02:29:04.540 --> 02:29:08.570
And they also, secondly,
don't capture the effects
02:29:08.570 --> 02:29:13.210
of the fire itself on the
wind fire induced winds.
02:29:13.210 --> 02:29:15.290
A dramatic example of that was
02:29:15.290 --> 02:29:17.400
the Creek fire this past summer
02:29:17.400 --> 02:29:19.160
which was driven almost entirely
02:29:19.160 --> 02:29:24.053
by fire induced winds this
fiddling with modeling results
02:29:24.053 --> 02:29:28.756
which we describe as calibration
has been used to disguise
02:29:28.756 --> 02:29:33.700
and bury these model
weaknesses for over four decades
02:29:33.700 --> 02:29:37.360
the calibration can't
identify conditions leading
02:29:37.360 --> 02:29:40.110
to nonlinear explosive fire growth that we've
02:29:40.110 --> 02:29:42.940
seen this summer from the data available
02:29:42.940 --> 02:29:45.320
or from looking at past events.
02:29:45.320 --> 02:29:47.953
They don't simply act the same way
02:29:47.953 --> 02:29:50.230
even if they're in the same condition
02:29:50.230 --> 02:29:54.021
because they vary according
to atmospheric structure.
02:29:54.021 --> 02:29:58.760
So my question, when
more forecasts such as this
02:29:58.760 --> 02:30:01.530
and other models out there in the marketplace
02:30:01.530 --> 02:30:04.700
be evaluated and compared against each other
02:30:04.700 --> 02:30:09.700
in an openly, publicly accessible model
02:30:09.840 --> 02:30:12.043
bake-off if we want to call it that?
02:30:14.590 --> 02:30:16.737
Okay, I can address this.
02:30:19.580 --> 02:30:21.710
Well, Johnny is something that first of all
02:30:21.710 --> 02:30:23.061
I would say that your statement
02:30:23.061 --> 02:30:27.830
I disagree strongly with
your personal statement
02:30:29.730 --> 02:30:33.167
the model that we're
using applied operationally
02:30:33.167 --> 02:30:36.607
that's been tested and
proven in the field here.
02:30:36.607 --> 02:30:41.420
And many places in the
world they providing really
02:30:41.420 --> 02:30:44.580
successful operational answers.
02:30:44.580 --> 02:30:47.290
We obviously are waiting
for the science community.
02:30:47.290 --> 02:30:48.760
You as a relevant research
02:30:48.760 --> 02:30:51.389
in this arena are working on that.
02:30:51.389 --> 02:30:53.214
We obviously are aware
02:30:53.214 --> 02:30:55.860
that there will be better
models in the future.
02:30:55.860 --> 02:30:57.743
We are still waiting for them.
02:30:57.743 --> 02:30:59.763
That will help us launch our questions
02:30:59.763 --> 02:31:01.770
like the ones that you addressed.
02:31:01.770 --> 02:31:03.080
Which I totally agree.
02:31:03.080 --> 02:31:05.000
I think that we really need to know
02:31:05.000 --> 02:31:09.550
a better understanding of how the complexity
02:31:09.550 --> 02:31:12.560
of the interaction with atmosphere
02:31:12.560 --> 02:31:16.033
and putting them on, on the fire situations.
02:31:16.033 --> 02:31:18.976
We know that we know that
the science take some time
02:31:18.976 --> 02:31:22.530
and to prove it that's the new model.
02:31:22.530 --> 02:31:24.920
It will take time, but with that
02:31:24.920 --> 02:31:26.780
we have to leave with the original models
02:31:26.780 --> 02:31:31.320
and the ones that we're using
again are proven and tested.
02:31:31.320 --> 02:31:34.560
Are using everyday and we
have our sample just as far
02:31:34.560 --> 02:31:36.620
as it's on 500 fires that were
02:31:36.620 --> 02:31:37.870
in our (indistinct) based
02:31:39.253 --> 02:31:41.480
on observation data that
wasn't available before.
02:31:41.480 --> 02:31:43.150
And probably the biggest largest
02:31:43.150 --> 02:31:45.410
among of observational data modeling.
02:31:45.410 --> 02:31:47.673
So pretty again, pretty solid on the model
02:31:47.673 --> 02:31:50.670
definitely waiting for the
scientist to create a new one.
02:31:50.670 --> 02:31:54.880
And again, it should be model
that this thing is important.
02:31:54.880 --> 02:31:55.760
That should be operational.
02:31:55.760 --> 02:31:59.453
I mean, it's useless to have
a model to analyze a fire.
02:32:00.390 --> 02:32:03.800
You take two or three, four
months just analyze one.
02:32:03.800 --> 02:32:05.670
And you mentioned the Creek fire.
02:32:05.670 --> 02:32:07.040
Well, we know that the Creek fire
02:32:07.040 --> 02:32:09.640
was absolutely an extraordinary event
02:32:09.640 --> 02:32:12.870
that is beyond the actual
science to understand
02:32:12.870 --> 02:32:16.330
but a fire like the Creek
fire doesn't stop like that.
02:32:16.330 --> 02:32:18.910
As a matter of fact, the Creek fire
02:32:18.910 --> 02:32:20.604
it started like it was like about 10 days
02:32:20.604 --> 02:32:24.870
before undetermined
conditions started to change.
02:32:24.870 --> 02:32:28.160
It's a different type of
problem that we wanna solve.
02:32:28.160 --> 02:32:31.480
And hopefully the model that you and others,
02:32:31.480 --> 02:32:32.730
create to help us.
02:32:32.730 --> 02:32:36.880
And actually we're
working with the (indistinct)
02:32:41.798 --> 02:32:45.270
in putting together the
state to introduce these
02:32:45.270 --> 02:32:46.290
in the future.
02:32:46.290 --> 02:32:48.450
But again, while that
happens, we have to work
02:32:48.450 --> 02:32:51.483
with the busier possible
with the best results possible
02:32:51.483 --> 02:32:55.130
and try to solve answers
in a (indistinct) way.
02:32:55.130 --> 02:32:57.430
Because if we wait for
the new science to come
02:32:57.430 --> 02:33:00.777
when we'll be in years
waiting to have an answer.
02:33:36.220 --> 02:33:37.870
There are no other questions
02:33:37.870 --> 02:33:39.580
in the comment line right now.
02:33:39.580 --> 02:33:41.670
If you would like to make a
comment or have a question
02:33:41.670 --> 02:33:44.400
please press star one and mute your phone
02:33:44.400 --> 02:33:47.400
and clearly record your name
and organization when prompted.
02:34:13.190 --> 02:34:14.180
We have no other questions
02:34:14.180 --> 02:34:16.180
or comments coming through at this time.
02:34:43.750 --> 02:34:47.210
As far as there's no
definitive plan to incorporate this
02:34:47.210 --> 02:34:50.290
in any policies or
procedures, it was like I said
02:34:50.290 --> 02:34:53.040
an exploration of new software.
02:34:53.040 --> 02:34:54.709
So pending that David,
02:34:54.709 --> 02:34:57.560
do you have anything before we sign off?
02:34:57.560 --> 02:34:58.710
I appreciate your time.
02:35:00.740 --> 02:35:01.573
But this is the operator.
02:35:01.573 --> 02:35:03.430
I apologize, but when you dial back in,
02:35:03.430 --> 02:35:05.980
you were on mute for
most of what you just said.
02:35:05.980 --> 02:35:07.480
I just opened your line.
02:35:07.480 --> 02:35:09.783
You may wanna repeat what you had just said.
02:35:11.860 --> 02:35:13.122
I just heard I was on mute.
02:35:13.122 --> 02:35:16.520
Can anybody hear David, can you hear me now?
02:35:16.520 --> 02:35:17.353
I can.
02:35:18.708 --> 02:35:20.410
Thank you.
02:35:20.410 --> 02:35:22.630
Make the operator for me,
so thank you everybody.
02:35:22.630 --> 02:35:24.840
Appreciate your time today.
02:35:24.840 --> 02:35:27.840
This was just a presentation
of a an exploration
02:35:27.840 --> 02:35:30.255
of this new fire modeling software.
02:35:30.255 --> 02:35:32.860
So I appreciate it big time, David
02:35:32.860 --> 02:35:35.870
did you have anything you
wanted to close out with?
02:35:35.870 --> 02:35:37.673
Just thanks everybody for setting in
02:35:37.673 --> 02:35:39.970
particularly with some of the delays
02:35:39.970 --> 02:35:42.000
that were occurring for some folks
02:35:42.000 --> 02:35:44.187
and I really appreciate the comments.
02:35:44.187 --> 02:35:48.452
We fully endorsed the
feedback from the community.
02:35:48.452 --> 02:35:51.202
It is a complex issue and we're tying
02:35:51.202 --> 02:35:55.380
to make use of the latest
science learn and move forward.
02:35:55.380 --> 02:35:57.540
And there's lots of moving parts.
02:35:57.540 --> 02:35:59.558
I really appreciate the questions
02:35:59.558 --> 02:36:02.794
and the interrogation that's, what's required
02:36:02.794 --> 02:36:05.455
to push this forward to save lives and homes.
02:36:05.455 --> 02:36:06.700
We appreciate it.
02:36:06.700 --> 02:36:07.533
Thank you.
02:36:08.610 --> 02:36:09.861
All right.
02:36:09.861 --> 02:36:10.967
Thank you so much, David
02:36:10.967 --> 02:36:13.040
and the rest of the Technosilva team.
02:36:13.040 --> 02:36:15.640
And with that, we will close the meeting.
02:36:15.640 --> 02:36:17.490
I appreciate everybody's
time to be here today
02:36:17.490 --> 02:36:19.510
and participate and I hope you have it all.
02:36:19.510 --> 02:36:21.440
They have a safe and good day.
02:36:21.440 --> 02:36:22.440
Thank you very much.
02:36:25.300 --> 02:36:26.133
Thank you.
02:36:26.133 --> 02:36:26.966
That concludes...