Ben Livneh - The Role of Climate on Hydrological Extremes in the Northern Great Plains Region

Gregg Hutchison Author

10/26/2017 Added

105 Plays

Description

The Role of Climate on Hydrological Extremes in the Northern Great Plains Region: Updates from the National Climate Assessment

Searchable Transcript

Search:

[00:00:00.520]Alright, thanks Martha.
[00:00:03.021]Thanks everybody for coming today,
[00:00:04.765]it's an honor to speak here, and to have been invited.
[00:00:08.483]So I'm gonna talk about some of the work I've done
[00:00:11.763]on the impact of climate, on hydrologic
[00:00:15.661]extremes in this region.
[00:00:18.045]As Martha said, my name is Ben Livman,
[00:00:19.863]I'm an Assistant Professor at the University of Colorado.
[00:00:22.445]Can everybody hear me okay in the back?
[00:00:24.823]Yes? Good, alright.
[00:00:27.283]So first, one of the things I was asked
[00:00:30.042]to say a few words about is the National Climate Assessment
[00:00:33.002]that Martha and Mike and I are working on.
[00:00:35.834]So I'll talk a little bit about
[00:00:36.991]the Northern Great Plains chapter, and how that's going.
[00:00:40.397]And then, I'm gonna venture more into some
[00:00:42.237]of the work that I've done in this region,
[00:00:44.595]part of which informs that chapter.
[00:00:47.757]First on the Great Plains drought of 2012,
[00:00:50.557]and then later on the flooding
[00:00:53.256]in the Upper Missouri River basin from 2011.
[00:00:57.434]And both of those, we tried to do,
[00:00:59.955]the team that I was working with,
[00:01:01.117]we tried to do an attribution of some of the drivers
[00:01:03.736]of that, and I'll try to comment on
[00:01:05.517]the role of future climate change, as well.
[00:01:09.997]So a little bit about me, I think Martha
[00:01:11.299]did a really good job telling you the main things.
[00:01:15.517]I've been an Assistant Professor since 2015
[00:01:17.613]at the University of Colorado.
[00:01:19.354]Before that, I got my PhD, and like many people, not all,
[00:01:24.253]but many with the dissertation, I had a very long title.
[00:01:27.651]Okay, and so I've abbreviated it there for you.
[00:01:30.333]And it shows that, yeah, I developed this new land model,
[00:01:35.277]and applied it to hydrology.
[00:01:37.853]Prior to that, I earned two degrees at the University
[00:01:40.755]of Western Ontario in Canada, which is where I grew up.
[00:01:45.075]So if you detect any kind of accent,
[00:01:46.933]that is probably where that's from.
[00:01:49.336]When I'm not working, I play in a jazz
[00:01:52.056]and bluegrass band in Boulder, called Paper Moonshine.
[00:01:56.317]And I have a one year old son at home, named Solomon.
[00:02:02.440]Alright, so talking about the National Climate Assessment,
[00:02:06.656]for those of you that are not familiar,
[00:02:07.894]I'll provide a little bit of background the assessment.
[00:02:10.918]It's actually part of a program called
[00:02:13.094]the US Global Change Research Program, USGCRP.
[00:02:19.078]So the USGCRP was mandated by Congress back in 1990, okay,
[00:02:25.768]in the Global Change Research Act,
[00:02:27.970]to assist the nation and the world to understand,
[00:02:31.026]assess, predict, and respond to human-induced
[00:02:35.688]and natural processes of global change.
[00:02:39.090]Okay, so that's quite a while ago now.
[00:02:41.789]And the National Climate Assessment mandate
[00:02:44.589]is a part of this, and it says
[00:02:46.706]that not less frequently than every four years,
[00:02:50.306]the program shall prepare and submit to the President
[00:02:53.090]and Congress an assessment that does various things.
[00:02:58.690]The one that I'd like to touch on,
[00:03:00.327]which I think is the most salient,
[00:03:02.029]is that it is to analyze current trends in global change,
[00:03:06.290]both human and natural, and projects major trends
[00:03:11.271]for the subsequent 25 to 100 years.
[00:03:14.791]So I thought this was interesting.
[00:03:16.827]One was that this whole program has been around for so long,
[00:03:21.446]and I think that we're on the fourth climate assessment now,
[00:03:24.189]which means we've sort of fallen behind the trajectory
[00:03:26.914]of doing something every four years, but that's okay.
[00:03:31.293]Right, so National Climate Assessment, the NCA four,
[00:03:35.229]is the one that we're involved with,
[00:03:37.650]and it's really a vast report.
[00:03:39.629]So I'm just gonna, I'm laying out here the different parts.
[00:03:42.759]You've got national topics, regional chapters,
[00:03:47.154]and you've got the appendices.
[00:03:49.904]And the things that are new are highlighted here.
[00:03:52.644]So the one that I'll draw your attention to
[00:03:55.079]is the Northern Great Plains regional chapter.
[00:03:58.602]So this was actually, in the previous climate assessments,
[00:04:01.904]the entire Great Plains was grouped together,
[00:04:04.439]and now we have separated them into two.
[00:04:07.584]And so that's the one that, as I said, we're involved with.
[00:04:13.221]The final publication isn't set to be released
[00:04:15.824]for more than a year from now.
[00:04:18.962]And so the amount that I can actually say about the report
[00:04:22.229]is somewhat limited, because it's still in draft.
[00:04:24.848]Nevertheless, if you see these green circles
[00:04:27.152]all over the place, those were where
[00:04:29.269]we had these regional engagement workshops,
[00:04:31.872]where we I think reached something like 1,000 stakeholders
[00:04:36.112]from across the country, trying to make
[00:04:37.922]the report more usable and sort
[00:04:41.400]of targeting the science where it is needed.
[00:04:45.302]Alright so, our chapter, you know, our chapter,
[00:04:49.702]I can say is gonna contain five parts:
[00:04:53.963]water, agriculture, recreation and tourism,
[00:04:57.707]energy, and indigenous peoples.
[00:05:00.224]So I'm leading the water section,
[00:05:02.966]with the help of others like Martha and Mike.
[00:05:07.126]And when I was making this talk a few days ago,
[00:05:09.926]actually I thought maybe I should reach out
[00:05:11.926]to the higher ups at the National Climate Assessment,
[00:05:14.640]just to make sure that it's okay
[00:05:16.443]that I talk about our chapter.
[00:05:19.338]And so I sent this email, and I basically said
[00:05:24.518]is it okay for me to summarize our chapter in the talk?
[00:05:29.163]And the response was "You shouldn't be citing the report."
[00:05:33.883]So that will limit some of what I can say today,
[00:05:37.359]unfortunately, because it is a draft.
[00:05:40.160]But then, the technical team reached out and said
[00:05:45.563]that actually the graphics that have been produced,
[00:05:48.203]I can use them, because they've been a joint effort.
[00:05:50.923]So I'm just gonna show a little bit of results
[00:05:52.603]from what we may present in the final report.
[00:05:59.424]Right, so if you recall, in the mandate
[00:06:03.147]of the National Climate Assessment,
[00:06:04.886]one of the things we're looking at is what
[00:06:06.730]are the trends for the next 25 to 100 years?
[00:06:09.728]So to kind of comment on the future
[00:06:11.808]of water in the Northern Great Plains,
[00:06:14.411]some analysis was done using the CMIP5 climate projection.
[00:06:19.323]This is an ensemble of general circulation models
[00:06:22.525]that run out to the year 2100.
[00:06:25.392]In order to interpret these results,
[00:06:27.851]the raw climate output has been downscaled,
[00:06:31.451]which I'll talk about in the next couple of slides.
[00:06:33.709]And for some of those, hydrologic simulations
[00:06:36.571]have been done.
[00:06:38.512]And most of what we're looking at is what we're saying,
[00:06:41.389]mid-21st century, so the period around 2050.
[00:06:46.011]So for that, so often what we need to look at is the
[00:06:50.752]historical baseline from which we can infer changes.
[00:06:54.672]So for example, hydrologic simulations were done
[00:06:57.471]with a model called the variable infiltration capacity,
[00:07:01.510]or VIC model, using the CMIP5 projection.
[00:07:08.064]Pardon me, first they simulated the historic snow pack
[00:07:12.222]for the Northern Great Plains region.
[00:07:14.564]And as you can see, you know, you have more snow up
[00:07:16.846]in the high country, as expected, and this is in inches.
[00:07:21.865]The future change out to mid-century is projected
[00:07:24.926]to actually show some increases in the plains.
[00:07:28.526]These are percent changes, but in the mountainous areas
[00:07:31.918]where you have the deepest snow packs,
[00:07:33.284]those are actually projected to decrease by quite a bit.
[00:07:37.662]And so, I realize I have revealed the next panel there.
[00:07:41.625]But one of the hypotheses here is yes,
[00:07:44.139]well temperatures are going to warm,
[00:07:45.566]so we expect the snow packs to be reduced.
[00:07:50.206]And one of the metrics for that is the account
[00:07:52.825]of the number of days, the change
[00:07:56.125]in the number of days below 28 degrees Fahrenheit,
[00:07:59.406]that is one measure of that.
[00:08:00.582]And the graph there may be a little bit difficult to see,
[00:08:03.204]but it basically shows that between,
[00:08:05.464]you know, 10 and maybe 30 fewer days will fall
[00:08:10.484]below 28 degrees Fahrenheit by mid-century,
[00:08:13.305]depending on which emissions scenario you use.
[00:08:16.880]So this would be, you know, one of the primary causes
[00:08:21.364]of reduced snow pack, is warmer temperatures.
[00:08:24.105]And it isn't just limited to the wintertime,
[00:08:26.404]there's also projections that you'll have many more days,
[00:08:29.225]perhaps a comparable number more days
[00:08:32.084]above 90 degrees Fahrenheit.
[00:08:34.361]So there's this potentially longer
[00:08:36.385]growing season, shorter snow season.
[00:08:40.205]Another finding is the, another important
[00:08:42.884]finding for water resources, is stream flow.
[00:08:46.146]Okay, so here's a plot of the spatial,
[00:08:48.029]sort of, stream flow for the region,
[00:08:51.563]based on the historical baseline.
[00:08:52.946]This is a period, a 30 year period,
[00:08:55.287]ending in 2006, I believe.
[00:08:59.743]And interestingly, going forward, the projections suggest,
[00:09:04.087]actually fairly large increases
[00:09:06.226]in flow for much of the region.
[00:09:09.389]But it's a little bit hard to see.
[00:09:11.565]If you look in the highest elevations,
[00:09:13.186]where most of the water originates,
[00:09:15.607]actually there tends to be subtle decrease.
[00:09:17.810]But overall actually, this is something that looks somewhat
[00:09:20.834]encouraging, based on the climate projections.
[00:09:24.535]However, we have to ask ourselves what
[00:09:26.914]is causing the potential increase in stream flow.
[00:09:30.637]And another facet of this, as we've plotted up
[00:09:34.571]the change in the number of days where
[00:09:37.554]you have more than two inches of precipitation per day.
[00:09:41.449]So actually, these are flood events, right?
[00:09:44.374]If you have more than two inches of rain,
[00:09:45.895]you're gonna get a high likelihood of flooding.
[00:09:48.493]So the question then, and these are
[00:09:50.557]for two different scenarios up here,
[00:09:52.573]two different emission scenarios,
[00:09:54.237]you can see large increases in the number of days.
[00:09:57.015]And so you wonder if the increase in stream flow
[00:10:00.754]is necessarily just a good thing,
[00:10:03.010]is it a blessing or a curse?
[00:10:06.578]So that is, you know, the preliminary
[00:10:09.981]kind of outlook for water in the Great Plains.
[00:10:13.761]And I talked a little bit about downscaled
[00:10:15.533]climate simulations.
[00:10:17.522]So let me explain a little bit more about that.
[00:10:20.759]So future climate projections, for most applications,
[00:10:24.941]for most impact assessments, need to be downscaled,
[00:10:28.397]essentially scaled to historical observations.
[00:10:31.981]Because you have these course grid model outputs,
[00:10:36.061]and we actually care about impacts
[00:10:38.279]at a much higher resolution.
[00:10:40.399]So in this way, climate simulations
[00:10:42.381]are often matched to higher resolution data
[00:10:46.843]for an overlapping historical time period,
[00:10:49.703]say from 1950 to 2000 or something like that.
[00:10:53.703]And this is because we need the high resolution
[00:10:55.565]data to understand impacts.
[00:10:59.821]So I've been involved in developing a historical
[00:11:04.461]data set that has been used in this mode.
[00:11:06.801]And actually, the data set that I'm describing here,
[00:11:08.921]turns out, is the official historical data set
[00:11:12.381]for the National Climate Assessment,
[00:11:14.701]which I actually just learned about fairly recently.
[00:11:17.901]Which is a good thing, I suppose, or not,
[00:11:20.381]depending on how you look at it.
[00:11:22.583]And just to explain how this is made,
[00:11:25.087]I started with about 30,000 station data
[00:11:29.261]that have daily observations of temperature
[00:11:32.002]and precipitation, that go back to the year 1950,
[00:11:35.337]or even earlier.
[00:11:38.445]And you put them on a regular grid,
[00:11:40.583]and then you can start to look at the spatial pattern.
[00:11:42.562]So I've just, I've got an animation here
[00:11:45.165]of daily precipitation in April of 1950.
[00:11:49.005]I'm just arbitrarily showing one month
[00:11:51.163]of daily data so you can see the variability.
[00:11:53.725]And so these data can then inform the downscaling,
[00:11:57.661]but also for running historical hydrological analyses.
[00:12:01.581]So as I showed in the previous slide with the snow pack
[00:12:05.327]and stream flow, those were run through a hydrology model.
[00:12:08.121]And just for demonstration purposes,
[00:12:10.601]I've run these data through a hydrological model
[00:12:13.623]and then plotting evapotranspiration
[00:12:15.741]for each day coincident with the precipitation.
[00:12:19.963]So if you're interested, we have a manuscript here
[00:12:22.861]that actually one of my co-authors
[00:12:25.106]is here at UNL, Dr. Francisco Munoz.
[00:12:28.823]And these are freely available, and you can download those.
[00:12:32.082]There's a number of places, you can
[00:12:34.663]check out my webpage if you want more information.
[00:12:38.043]Right, okay, so now we've talked about the data,
[00:12:41.643]and a little bit about the climate assessment.
[00:12:43.623]So I'd like to talk first about some work
[00:12:45.661]we've done the Great Plains drought.
[00:12:49.524]This is work I did with Marty Herling at NOAA,
[00:12:52.989]and we published a paper on this last year.
[00:12:56.071]So I'm just gonna touch on a few key findings
[00:12:59.090]that came out of this paper today.
[00:13:03.927]Alright, so there was this, as everyone here
[00:13:07.010]is probably well aware, there was this large drought event,
[00:13:09.453]and kind of rapid onset drought event,
[00:13:12.950]in 2012 here in the Great Plains.
[00:13:15.533]And that prompted a report to come out of NOAA,
[00:13:19.111]this is sort of before I got there,
[00:13:21.810]on diagnosing the origins of the 2012 drought.
[00:13:26.274]And the preliminary work was done
[00:13:28.230]looking at things like ocean atmospheric coupling,
[00:13:31.095]and kinda big picture global causes and drivers.
[00:13:34.615]The distinction with what I'm gonna show today
[00:13:36.834]is that the focus is here specifically on the land surface.
[00:13:39.569]And among the experiments that we did,
[00:13:41.727]I'm gonna focus on those experiments
[00:13:43.933]that try to relate drought sensitivities directly
[00:13:46.514]to either temperature or to precipitation,
[00:13:49.954]because those are the two kind of
[00:13:51.090]major things we observe well.
[00:13:54.354]So it can be difficult not just to infer the causes,
[00:14:00.253]or even the impacts of drought,
[00:14:02.375]especially when we go back in time
[00:14:04.087]and have fewer observations.
[00:14:05.830]So many of you may be familiar with this graphic here.
[00:14:09.415]This a plot of the historic average grain yields,
[00:14:13.154]corn grain yields, going back to the 1860s.
[00:14:16.653]And it's a way in which we can potentially
[00:14:19.167]use grain yields as a proxy for hydrologic conditions.
[00:14:23.244]So obviously, you know, we didn't have remote sensing,
[00:14:28.410]or even a large in situ network back in the 1800s.
[00:14:31.685]So grain yields, and you can see that there
[00:14:33.743]are these different trends lines,
[00:14:36.348]these have nothing to do with climate change,
[00:14:38.186]but rather denote different systemic
[00:14:41.402]improvements in agricultural practices.
[00:14:44.538]And if you look at departures from these trend lines,
[00:14:48.181]those can tell you about if there was a large extreme event,
[00:14:52.341]if there was a flood or a drought.
[00:14:55.439]And since 2012, the drought affected
[00:14:57.839]a large swath of agricultural land.
[00:15:01.297]It actually shows up as a pretty large departure here,
[00:15:04.375]in terms of grain yields, and it had quite
[00:15:06.021]a large price tag estimated to go along with that.
[00:15:09.599]So that's one way to estimate these things.
[00:15:13.258]So in our analysis, the first step was to understand first,
[00:15:18.437]what are the meteorological conditions?
[00:15:20.159]Before we can start attributing anything,
[00:15:22.158]we have to understand kinda the baseline situation.
[00:15:24.634]So I took this same data set that I was referring to before,
[00:15:28.476]and I'm showing here the May through August
[00:15:33.441]standardized anomaly for precipitation and temperature.
[00:15:40.847]And what you can see is that, so for those of you
[00:15:43.988]that are not familiar with the standardized anomaly,
[00:15:46.548]what I'm showing basically are the May through August
[00:15:49.681]values minus the historical mean,
[00:15:54.815]and divided by the standard deviation
[00:15:56.665]for this reference period, this 30 year period.
[00:15:59.204]So what you're essentially seeing, for the central plains,
[00:16:02.009]is the number of standard deviations
[00:16:04.409]above or below the mean for that event.
[00:16:08.105]So you can see that the darker red values are one sigma,
[00:16:12.287]one standard deviation, or even more
[00:16:13.689]departures from historical.
[00:16:17.583]And so that begins to tell us a little bit
[00:16:18.945]about how severe the meteorological conditions are,
[00:16:23.385]but it doesn't tell us which of these two facets,
[00:16:27.126]of course they are related to one another, but which
[00:16:29.924]of them was really more responsible than the other.
[00:16:34.687]So to try to get at this, a lot of the work
[00:16:37.485]I do in my research group involves hydrologic modeling,
[00:16:40.662]as you may have inferred so far.
[00:16:43.417]And so, we ran this fully distributed water
[00:16:47.225]and energy balance model called the VIC hydrology model.
[00:16:51.924]We focused on the top one meter of soil.
[00:16:54.465]Okay so, here's a snapshot from August of 2012,
[00:16:59.083]where we see the standardized soil moisture anomaly here,
[00:17:03.945]and we started thinking well okay, so we have a model,
[00:17:06.865]how do we know if the model is actually telling us
[00:17:09.508]something that is remotely accurate to what happened?
[00:17:12.425]So one of the first things that we looked at,
[00:17:14.527]which is relevant to this institution, is we looked at
[00:17:17.049]a similar time slice from the US drought monitor,
[00:17:20.265]more so just to give sort of a qualitative comparison.
[00:17:25.225]And what we found is that, you know, in general,
[00:17:28.649]the model seems to capture the major features.
[00:17:31.764]You may have this large negative anomaly
[00:17:33.924]in the center of the country, but also local minima
[00:17:37.065]in northwest Nevada, and in the Southeast.
[00:17:40.447]So we thought okay, the model is doing a pretty good job,
[00:17:44.449]certainly it's a difficult, this is not
[00:17:46.505]like an apples to apples comparison,
[00:17:48.287]but this is our first comparison.
[00:17:50.763]And then we thought maybe there's
[00:17:52.004]something else that we could look at.
[00:17:53.807]So we looked at the GRACE data.
[00:17:59.268]So maybe, show of hands, how many people
[00:18:00.564]are familiar with the GRACE date?
[00:18:03.001]Okay, quite a few, great.
[00:18:04.476]So for those that are not, the GRACE data measures changes
[00:18:08.018]in the Earth's gravity, and relates
[00:18:09.800]that to movements of water, because water has a lot of mass,
[00:18:12.898]which contributes to gravity.
[00:18:14.952]And so what I'm plotting here an animation
[00:18:17.261]of changes in liquid water thickness over the globe,
[00:18:22.300]and there's basically a monthly time step,
[00:18:26.189]and you can see, you know, negative anomalies
[00:18:28.509]in places like Greenland and Antarctica,
[00:18:31.431]where you maybe have mass loss.
[00:18:33.960]And as you get closer to 2012, and you may have
[00:18:36.657]just missed a big, the drought kinda goes by there.
[00:18:41.523]This course scale is about a 300 kilometer resolution,
[00:18:45.101]is actually probably commensurate
[00:18:47.043]with the scale of the drought.
[00:18:48.621]So it's actually a useful, potentially useful observation.
[00:18:51.421]So zooming in on just 2012, and just on North America,
[00:18:57.301]you can see this kind of rapid onset of the drought,
[00:19:01.747]where things went from near normal around April,
[00:19:05.923]and then really dried out.
[00:19:08.161]Okay so, we can do the same thing with the model now,
[00:19:12.648]simulating just 2012 in just the Central Plains here.
[00:19:17.402]And the model can tell us a little bit
[00:19:19.783]more about some of the dynamics.
[00:19:21.448]Obviously these are simulated, but we can start
[00:19:23.242]to learn a little bit about how things evolve.
[00:19:25.943]So if I take just that box, the Central Plains,
[00:19:29.928]and I show the model simulation average for that,
[00:19:33.987]so I've got the months of 2012 here.
[00:19:36.984]This is depth, this is called a contour plot.
[00:19:39.565]And through time, you can see that soil moisture,
[00:19:42.744]which is plotted, are the colors,
[00:19:45.441]was near normal in April, okay, of 2012.
[00:19:49.843]Then you can really see this rapid onset
[00:19:51.565]of drying as it descended down
[00:19:54.365]into the soil column and caused this drought.
[00:19:57.725]And one of the unique features
[00:19:59.267]of this drought was the rapid onset.
[00:20:02.840]So then, we can take our model, the VIC model.
[00:20:07.603]Also, I ran another model, this is the unified land model,
[00:20:10.643]the one I developed for my dissertation, so two models,
[00:20:14.184]and compare them through time with GRACE.
[00:20:19.964]So we have quite independent information
[00:20:22.401]here that we're comparing.
[00:20:24.163]And when you get to 2012, you can see
[00:20:26.541]that they all kinda drop off a cliff together.
[00:20:31.144]So this is another type of validation
[00:20:33.123]that we could do with the model to make sure
[00:20:35.101]that we were doing, you know, that we could actually
[00:20:37.617]make some sort of informative statements
[00:20:42.977]about the drought, and about the attribution.
[00:20:47.626]Okay so, we know that drought integrates
[00:20:51.929]weather over long time periods.
[00:20:54.429]And so what I'm showing here is standardized temperature
[00:20:58.473]anomalies for the growing season of like 64 years,
[00:21:02.633]and standardized precipitation anomalies for 64 years.
[00:21:06.929]And I'm gonna highlight here the 2012 drought,
[00:21:11.033]which what I'm trying to show
[00:21:12.669]is that it was a remarkably dry, maybe two or three,
[00:21:15.593]almost two and a half sigma standard deviation departure
[00:21:19.913]in precipitation, and two departures in temperature,
[00:21:24.109]to show that it was a remarkable event.
[00:21:27.169]Another drought in recent memory was the 1988 drought,
[00:21:29.528]which was not as severe for that time period.
[00:21:33.107]And just to say that, you know,
[00:21:34.029]this is the kind of observational data that we have,
[00:21:36.333]but it doesn't tell us which one
[00:21:38.429]of these was necessarily more important.
[00:21:42.689]So what we did is we made the same plot with the model,
[00:21:46.232]where we have simulated soil moisture
[00:21:49.773]standardized anomalies, versus precipitation here.
[00:21:54.371]And what's interesting is, you know, you've got 1988
[00:21:57.395]and 2012 showing up here, two pretty important droughts,
[00:22:00.728]but they're not actually the most severe drought.
[00:22:03.853]And in fact, the most severe one
[00:22:06.071]since 1950 turns out to be 1956, which doesn't
[00:22:10.551]show up as particularly anomalous for this time period
[00:22:14.071]in terms of temperature and precipitation.
[00:22:16.675]But if you look at the other low years,
[00:22:20.317]what you see is that many of the dry years,
[00:22:23.453]kind of in a row, were preceding 1956.
[00:22:27.757]So it's not just an acute temperature
[00:22:29.795]or precipitation anomaly that causes a severe drought,
[00:22:33.117]but it's a sustained anomaly.
[00:22:35.493]And we have these other indices that get at this,
[00:22:37.656]something that the model is integrating through time.
[00:22:42.915]So this highlights the need for,
[00:22:44.589]or the utility of having a model to integrate
[00:22:47.592]these things and balance mass and energy in the process.
[00:22:56.392]Okay so, the main experiment that I'm gonna show for this,
[00:23:00.232]for the drought, is this synthetic experiment we did,
[00:23:03.992]in which we tried to isolate the impact
[00:23:05.576]of temperature and precipitation.
[00:23:07.736]So here's this historical kinda baseline run.
[00:23:10.515]We found about a 1.5 standardized sigma
[00:23:14.456]anomaly for the growing season for 2012.
[00:23:18.573]So to try to isolate temperature and precipitation,
[00:23:20.872]we did something very simple, which was we essentially
[00:23:23.469]reran the model, and we used all the conditions as before,
[00:23:29.656]except we set the precipitation value
[00:23:32.632]to the historical climatology.
[00:23:35.555]So we had the high temperatures of 2012,
[00:23:37.731]but not the precipitation depths.
[00:23:39.933]And we assimilate only a very modest drying.
[00:23:44.172]Okay, so this is just that the high temperatures
[00:23:45.555]only had a modest effect on soil moisture.
[00:23:49.132]When we do the same thing, but we now isolate precipitation,
[00:23:53.311]so we remove the high temperatures,
[00:23:55.075]the anomalously high temperatures, from the event,
[00:23:57.815]we get a much, to the eye a much closer match,
[00:24:01.773]and actually numerically a much larger contribution,
[00:24:04.536]which may not be all that surprising.
[00:24:07.610]But I think it was useful to try to quantify
[00:24:09.976]the contribution of each of these components.
[00:24:14.797]And depending on the model you run or the setup
[00:24:17.677]that you use, we ended up with something like
[00:24:20.195]between three to one or four to one the contribution
[00:24:23.693]of precipitation to temperature in the drying event.
[00:24:27.459]So we tried to really break these apart,
[00:24:29.255]which was not actually easy to do.
[00:24:31.635]But this is sort of the result that we found there.
[00:24:35.976]So from this drought, a few key points.
[00:24:38.989]As I just said, the signature was primarily
[00:24:42.611]driven by precipitation, which isn't, you know, there's no,
[00:24:45.507]that isn't necessarily newsworthy in and of itself,
[00:24:49.565]other than we tried to actually attribute
[00:24:51.891]or quantify how much attribution that was.
[00:24:55.907]Now in terms of future climate, well temperature
[00:24:59.309]sensitivity here was seen to be relatively low.
[00:25:02.568]However, the future is projected
[00:25:04.568]to be much warmer than the past.
[00:25:07.331]And so even a small temperature sensitivity
[00:25:09.469]could make future droughts more severe,
[00:25:12.589]particularly something I didn't really talk about
[00:25:14.792]is the low precipitation tends to be coupled
[00:25:17.834]with high temperatures, and so that's something important.
[00:25:20.675]Other work that I did not show suggests
[00:25:24.671]that there is some potential for predictability
[00:25:27.080]of these types of droughts, if you knew something
[00:25:30.243]about the antecedent soil moisture and temperature.
[00:25:33.941]And for that, those details, I would encourage you
[00:25:37.603]to check out this manuscript, or you can email me for that.
[00:25:42.061]Alright so, the first part of the talk was on drought.
[00:25:47.608]Alright, so the second part I wanted to
[00:25:49.751]talk a little bit about the work
[00:25:51.271]I was involved with on understanding,
[00:25:54.492]on the opposite end of the hydrologic
[00:25:56.674]extremes in the Upper Missouri River Basin.
[00:25:59.430]So specifically, looking at the 2011
[00:26:02.071]Upper Missouri River Basin flooding.
[00:26:04.775]And this is work I did also with NOAA,
[00:26:06.914]as well as with the Army Corps of Engineers' office
[00:26:09.271]in Omaha, that helped to fund this work,
[00:26:13.186]trying to diagnose the drivers of this event.
[00:26:16.514]So another, you know, event that's close to home here,
[00:26:21.212]the Missouri River Basin conditions in June of 2011,
[00:26:25.175]when the flood occurred, it was the highest flow
[00:26:27.575]ever recorded since records began in 1898.
[00:26:32.178]So it was a very large event.
[00:26:33.580]Some numbers here demonstrate that.
[00:26:35.938]You had above Sioux City, which is where we were concerned,
[00:26:43.319]you had almost four times the usual flow.
[00:26:46.199]Other places had even higher values.
[00:26:48.162]But in general this was, you know, a record-setting
[00:26:51.420]kind of flood event, and these are numbers
[00:26:54.076]of the different major reservoirs upstream.
[00:26:58.153]So similar to the 2012 drought, actually,
[00:27:01.399]right after the event, a NOAA climate assessment report
[00:27:04.775]came out looking at the drivers of this.
[00:27:08.898]You know, ocean atmosphere coupling, that kind of thing.
[00:27:11.897]And one of the things that emerged
[00:27:13.260]was this recent variability in annual flow.
[00:27:17.420]So what I'm showing here is a bar plot of,
[00:27:21.938]each bar represents one year of flow in the upper basin.
[00:27:26.699]And these are what are known as naturalized flows,
[00:27:29.596]that the Army Corps actually reconstructed
[00:27:31.676]accounting for the reservoirs and other things.
[00:27:35.575]And what you see is, actually we have
[00:27:37.676]these sort of historic drought periods interspersed
[00:27:40.976]with kind of increasing flood events.
[00:27:44.316]So this really interesting variability and oscillation
[00:27:47.575]that led us to inquire further,
[00:27:50.736]and that's gonna be the topic of these next few slides.
[00:27:55.676]Okay so, what we've got here on the upper right,
[00:27:59.260]this is the same plot as previous.
[00:28:01.441]You've got annual stream flow since 1898,
[00:28:05.281]and below that you've got annual precipitation falling over
[00:28:09.678]the Upper Missouri River Basin for the same time period.
[00:28:13.718]So these, the red line shows a 20 year
[00:28:17.537]moving window of standard deviation.
[00:28:20.417]And what we find is that runoff exhibits
[00:28:23.921]this rising interannual variability,
[00:28:27.278]particularly almost like a step change since 1975.
[00:28:31.120]And the same thing you see in precipitation,
[00:28:34.300]although not quite as much, but you definitely
[00:28:36.198]see this interesting change in variability,
[00:28:38.979]perhaps not as much as runoff.
[00:28:41.916]And so we're wondering, what is responsible for this large,
[00:28:45.192]what is the mechanism behind this disparity?
[00:28:48.481]And a few things we looked at,
[00:28:49.760]I won't show all of those today, are changes
[00:28:52.060]in precipitation intensity, seasonality changes.
[00:28:56.220]So it was the precipitation being delivered
[00:28:57.681]at the same time, and is potentially the sequencing
[00:29:01.280]of the wetting, the antecedent moisture,
[00:29:04.019]is that changing in some way?
[00:29:07.702]Right so, a couple of kind of big picture features
[00:29:11.740]of this region is, if you look at the two primary modes
[00:29:15.361]of variability, this was done using something
[00:29:18.380]called principle components analysis.
[00:29:20.678]But the important thing here is you have the primary signal,
[00:29:24.901]and the precipitation is the sort wet growing season.
[00:29:27.738]So we already know this.
[00:29:29.158]And if you just knew about the wet growing season,
[00:29:30.998]that would explain about 90 percent
[00:29:33.004]of precipitation variability in the region.
[00:29:36.138]Another seven percent of the variability
[00:29:37.884]can be explained by this kind of
[00:29:39.500]montane wet signal during the winter and spring.
[00:29:44.172]So that's the high country snow pack.
[00:29:46.833]So those are kind of the two key meteorological features
[00:29:49.190]that one would need to know if they were going
[00:29:51.699]to do a hydrological analysis as we did.
[00:29:56.188]Right so, we started first with station data.
[00:29:59.347]So these points plotted here represent stations
[00:30:02.305]where we had reasonably complete data coverage,
[00:30:05.729]going back to the year 1901.
[00:30:08.972]Okay so, a very long record,
[00:30:10.408]and there is I think over a hundred stations,
[00:30:12.412]and this is the entire Missouri basin, of course.
[00:30:15.190]So on the top here, we just show for those stations
[00:30:19.590]what has been the seasonal change in cold season,
[00:30:22.785]October through March, precipitation versus warm season.
[00:30:27.180]And we're comparing the recent 40 years
[00:30:29.868]with the past 75, or almost 80 years,
[00:30:33.845]trying to say how has this recent 40 year period
[00:30:36.566]been different than the past?
[00:30:38.860]And you actually see that there have been
[00:30:40.348]steady increases in many places, these dark blue colors
[00:30:43.782]in winter, and not so much in summer.
[00:30:49.125]And of course, it's always a bit of a mixed bag,
[00:30:52.097]you see browns and blues mixed together,
[00:30:54.312]which is you know, a puzzle for a different talk.
[00:30:59.116]So then we thought well maybe it's
[00:31:00.118]the extremes that are changing.
[00:31:01.777]So we looked at only the top one percent
[00:31:04.636]of precipitation events, and we said how
[00:31:06.598]is the fraction of precipitation
[00:31:10.070]that falls in those one percent events,
[00:31:13.043]how is the fraction of that relative to the entire season?
[00:31:16.163]How is that changing?
[00:31:17.223]Are those one percent events getting larger?
[00:31:20.825]And it turns out that in many places, yes.
[00:31:23.822]You see these dark blue colors,
[00:31:25.155]which suggest 20 or 25 percent changes.
[00:31:28.323]And that is more so in the cold season than the warm season.
[00:31:33.719]But again, there's a lot of variability here.
[00:31:35.875]Okay so, that was one thing we looked at.
[00:31:39.342]Temperatures for the region, these are,
[00:31:42.195]I'm showing maximum daily temperatures.
[00:31:45.619]And those have generally increased,
[00:31:48.379]especially in the northwestern part of the basin.
[00:31:51.075]In the southeast part, you actually see
[00:31:54.519]this sort of cooling from maximum daily temperatures,
[00:31:57.481]and this we understand is actually potentially
[00:31:59.945]an artifact of the data set that you use.
[00:32:02.963]So the prism data versus NCEI data.
[00:32:07.061]And there's also potentially a phenomena known
[00:32:09.283]as a warming hole in the southeast of this area,
[00:32:12.519]which I would be very interested
[00:32:14.083]to talk to anybody here about afterwards.
[00:32:16.905]So there's the data quality potentially,
[00:32:18.686]whether or not that's physically real
[00:32:21.742]or not is a different topic.
[00:32:24.622]But in general, things have gotten a little warmer,
[00:32:26.265]especially in the upper basin.
[00:32:28.661]Alright, so we have the forcing data,
[00:32:32.185]we have the temperature, the precipitation.
[00:32:33.861]To understand the role of precipitation versus antecedent
[00:32:38.960]moisture, we did some hydrologic modeling experimentation.
[00:32:43.006]So I took this variable infiltration capacity VIC model,
[00:32:47.625]which is the same model that was used
[00:32:49.840]in the National Climate Assessment.
[00:32:51.523]And I used by data set of stations to create a gridded
[00:32:56.286]simulation for 64 years, okay, going back to 1950.
[00:33:02.366]And we divided the upper basin into these six sub-basins
[00:33:05.982]that correspond with the major reservoirs in those areas.
[00:33:11.241]One of the notable features of the Upper
[00:33:12.841]Missouri River Basin is it's extremely arid
[00:33:15.643]from a hydrologic perspective, its runoff ratio.
[00:33:18.483]So stream flow over precipitation
[00:33:20.279]of only nine percent, which is extremely arid.
[00:33:22.999]And for those of us that have done hydrologic analyses,
[00:33:26.501]we know that these arid regions are extremely difficult
[00:33:30.098]to model accurately, simply because
[00:33:32.222]there's so many ways that the analysis can go wrong.
[00:33:35.203]So we had to do this long calibration,
[00:33:37.523]and because there's about 25,000 grid cells
[00:33:40.595]in the upper basin, and it takes a long time to run,
[00:33:43.378]we could only do maybe a 50 or 100
[00:33:46.003]different sensitivity analyses.
[00:33:48.446]So we ended up with a simulated hydrograph,
[00:33:51.395]shown here in red, that we can pair with the Army Corps
[00:33:55.257]of Engineers in black, this is the natural stream flow.
[00:33:59.561]And while we capture some of the main features,
[00:34:02.547]we miss a few of the early peaks in the season,
[00:34:07.177]and we don't capture everything.
[00:34:08.601]And at some point, we had to sort of move on
[00:34:10.937]and say that we were, you know,
[00:34:13.859]this was good enough to proceed to the next step.
[00:34:16.958]The metric that we were most concerned
[00:34:19.817]with was the ratio of the modeled standard deviation
[00:34:22.659]to the observed, because that's a measure
[00:34:25.417]of the sensitivity of the model relative to observations.
[00:34:29.715]And we got that to be really close
[00:34:31.177]to one, which was heartening.
[00:34:33.417]Just so you're aware, this is kind of the
[00:34:34.971]model setup that we used going into this.
[00:34:38.398]So now we ran a few sensitivities to try
[00:34:41.152]to get at the attribution here.
[00:34:45.293]And so the question we asked, and this is a question
[00:34:47.897]that the Army Corps folks were asking,
[00:34:50.318]and actually a lot of people were asking.
[00:34:52.813]In 2011, were the antecedent moisture conditions,
[00:34:57.072]so the soil moisture and the snow pack,
[00:34:59.417]were they such that a flood could
[00:35:00.878]have been predicted at the start of the year?
[00:35:04.192]Alternatively, were the forcings,
[00:35:06.211]and by forcings I'm primarily talking about precipitation,
[00:35:09.411]was this the key factor driving the flood?
[00:35:12.073]In which case, this would have been difficult to predict.
[00:35:14.953]Okay, so to do this, we started
[00:35:17.614]with our baseline simulation.
[00:35:19.993]So you see, there's actually
[00:35:21.193]64 hydrographs here, annual hydrographs.
[00:35:24.123]These are the months of the year.
[00:35:26.099]This is the water year, starts in October.
[00:35:28.254]And you can see 2011 here as the big peak.
[00:35:30.761]So the first part we said well, was it the initial
[00:35:33.552]moisture conditions that were the key?
[00:35:35.714]So we set up experiment number one.
[00:35:38.334]What we did was we shuffled the
[00:35:39.555]initial conditions in the model, the soil moisture
[00:35:41.737]and the snow pack for the 64 historic years.
[00:35:45.977]And then give the model each of those initial conditions
[00:35:48.619]the 2011 forcing, the 2011 precipitation.
[00:35:52.238]Okay, and what that looks like is basically
[00:35:54.777]you take this initial condition in 2011,
[00:35:57.454]and apply the, sorry the forcing from 2011,
[00:36:01.198]you apply that to the initial antecedent moisture
[00:36:04.355]from 1950, 1951, and all the way through.
[00:36:08.435]Conversely, the second experiment we took
[00:36:11.161]the meteorological forcing, so the climate
[00:36:14.334]from each of those 64 years,
[00:36:16.773]and applied it to the 2011 initial conditions.
[00:36:20.514]So this would tell us, you know,
[00:36:22.537]is it the 2011 initial condition that's really special,
[00:36:26.675]and if you ran it with any other climate,
[00:36:29.049]you would still get a flood event.
[00:36:30.734]So this is how we tried to diagnose this.
[00:36:34.835]And one thing to take note of is that 2011 was one
[00:36:38.691]of very few years that reached this unit
[00:36:41.440]of 10 million acre feet, shown here as 10,000 KAF,
[00:36:46.734]these are thousand-acre feet.
[00:36:48.171]So just keep that number in mind,
[00:36:49.475]that 10,000 is generally considered a large flood event.
[00:36:54.073]Alright, so the first experiment,
[00:36:55.753]where we took the 2011 climate and applied it to each
[00:36:58.493]of the initial conditions from the different years,
[00:37:02.414]right away we see oh wow, pretty much all
[00:37:04.951]of these years produced a peak flow around 10,000 KAF.
[00:37:10.398]So this suggested to us right away
[00:37:12.034]that 2011 forcing produces many years
[00:37:15.003]with what would be considered flood flow.
[00:37:18.494]And if you look at this a little bit more carefully,
[00:37:22.451]here I'm plotting the stream flow from those years,
[00:37:25.193]versus the antecedent moisture.
[00:37:27.643]You get these really high flow years here.
[00:37:31.333]The bigger symbols are those years
[00:37:32.697]that are actually known historical flood years.
[00:37:35.653]And those little symbols were not flood years.
[00:37:38.654]And what this suggests, so here's 2011,
[00:37:40.798]that actually there were other years
[00:37:42.857]with much wetter initial conditions,
[00:37:45.038]and had the 2011 meteorology actually
[00:37:47.403]fallen on those sort of wet conditions,
[00:37:50.834]we would have had much worse flood events.
[00:37:52.857]So the 2011 flood event actually could have been
[00:37:55.155]considerably worse, at least according to the model.
[00:37:59.835]Conversely, if we do the opposite experiment,
[00:38:02.499]where we just use the initial conditions from 2011,
[00:38:05.173]and the 64 years of different meteorology,
[00:38:08.475]you see that very few years, and this actually
[00:38:11.118]looks a lot more like the baseline,
[00:38:13.977]very few years produced these flood flows,
[00:38:16.355]and that actually the 2011 initial conditions
[00:38:18.718]were probably inadequate to drive flood events,
[00:38:22.515]because very few years have these large floods.
[00:38:27.459]Okay so, returning to our questions.
[00:38:30.579]Rather than the antecedent moisture,
[00:38:32.318]it turns out that actually, it was the precipitation
[00:38:35.235]and the forcings that were the key key driving factor,
[00:38:37.475]which was I think quite heartening
[00:38:38.979]for people managing the reservoirs,
[00:38:42.411]who had to release water and flood certain areas.
[00:38:45.571]So this kind of suggested to them that maybe they couldn't
[00:38:48.291]have known, and maybe they made the right call.
[00:38:52.147]But this is a very contentious issue,
[00:38:53.966]and we repeated this experiment for other months,
[00:38:56.510]not just the start of the water year,
[00:38:57.966]and I could talk more about that later.
[00:39:01.705]Lastly, just to get a sense for this,
[00:39:03.646]the way that this is set up.
[00:39:06.363]If you look at the historical average precipitation signal
[00:39:12.483]through the years, it's this blue line.
[00:39:15.406]And you can see that it peaks in, you know, May and June.
[00:39:19.277]And if you took the trend, because there is actually
[00:39:21.535]a wetting trend, if you actually applied that to the data,
[00:39:24.255]which we did, I didn't show the results for it.
[00:39:26.117]You see you only get this small, this red curve here,
[00:39:29.194]when in fact 2011 was such an extraordinary year in terms
[00:39:32.927]of precipitation late in the season.
[00:39:36.618]So in May and June, making it
[00:39:39.354]a very difficult year to predict.
[00:39:42.735]A couple of other points.
[00:39:44.277]Some of the areas, so I told you
[00:39:47.135]that the whole watershed was very, was relatively
[00:39:50.154]insensitive to antecedent moisture conditions.
[00:39:52.997]However, the easternmost sub-basin,
[00:39:55.957]which is Gavin's Point above Sioux City,
[00:39:59.194]was actually very sensitive to the antecedent
[00:40:01.317]moisture conditions from that year.
[00:40:04.437]And it also turns out that that's the
[00:40:08.229]least regulated sub-basin, so that whatever
[00:40:11.477]flood waters were coming down, there is the least amount
[00:40:15.935]of reservoir capacity to hold those flood waters back.
[00:40:20.778]Furthermore, we found in our analysis
[00:40:22.975]that actually that region had uniquely
[00:40:26.993]high antecedent moisture at the start of 2011.
[00:40:30.391]So it was uniquely sensitive, plus it had
[00:40:33.150]this uniquely high antecedent moisture,
[00:40:35.333]kind of like a perfect storm there
[00:40:38.031]that led to these high flows.
[00:40:41.311]So the statistically significant areas
[00:40:43.450]are highlighted in gray, and you can see
[00:40:45.215]that that subregion is uniquely sensitive.
[00:40:49.535]The last thing I'll say about this in terms
[00:40:51.071]of the results from these experiments
[00:40:53.393]was that we looked at increasing the temperature,
[00:40:56.195]and increasing the top one percent
[00:40:58.437]most extreme precipitation days, and we see these
[00:41:01.061]very muted sensitivities, which was somewhat surprising.
[00:41:03.951]I think we realized that actually,
[00:41:05.375]the reason why changing the top one percent
[00:41:08.175]precipitation days didn't do that much
[00:41:11.007]is because there's probably on average
[00:41:12.915]about 100 rainy days per year in this climate.
[00:41:16.234]So the top one percent of data is only one day per year.
[00:41:19.233]So even if you increase that event by say,
[00:41:21.898]20 percent or 50 percent, it's still only one day.
[00:41:24.673]And we've talked about revisiting
[00:41:26.218]those experiments a little bit.
[00:41:28.874]Looking to the future, these are simulations
[00:41:32.517]from a large ensemble of climate models based out of NCAR,
[00:41:37.194]and they show precipitation, temperature, and runoff
[00:41:40.833]going out to the year 2100 for 40 model simulations.
[00:41:45.061]And while precipitation is increasing slightly,
[00:41:47.781]and temperature is increasing dramatically, the effect
[00:41:51.615]of that is, you know, perhaps a slight reduction in runoff.
[00:41:54.497]I think the big, there's a lot of uncertainty here,
[00:41:57.098]and perhaps the most important part
[00:41:58.938]of that is that this large variability
[00:42:00.958]that we see could portend, there's nothing to suggest
[00:42:04.282]that it wouldn't portend for the likelihood
[00:42:07.599]of continued drought and flood events
[00:42:10.019]like we've been seeing the past, which I think
[00:42:11.818]are the most serious risks that this region faces.
[00:42:18.819]So to conclude, you know, I talked a little bit
[00:42:21.941]about large-scale hydrometeorology,
[00:42:24.015]and there are still open questions
[00:42:25.701]on how to downscale and do other things related to that.
[00:42:30.138]In terms of the regional extremes,
[00:42:32.698]the extreme runoff I showed is primarily produced
[00:42:35.477]by extreme precipitation and antecedent soil moisture.
[00:42:39.018]In the Missouri, precipitation has been the largest factor,
[00:42:41.679]which poses a unique challenge for predictability.
[00:42:47.137]Finally, if you're interested to learn more
[00:42:49.313]about the National Climate Assessment,
[00:42:51.439]there are resources online, a website and Twitter.
[00:42:55.621]And if any of you are attending the EGU fall meeting,
[00:42:58.778]there's actually going to be a booth all week
[00:43:02.259]that has information, and then there are
[00:43:04.421]individual sessions, including a town hall,
[00:43:07.477]that you can check out.
[00:43:09.738]So with that, I thank you for your attention.
[00:43:12.375](audience applauding)
[00:43:15.754]Ben, nice talk. Thank you.
[00:43:19.854]I got the last part of your talk,
[00:43:24.383]and the experiment you designed, and sample type
[00:43:28.527]of approach, I wonder if one of the reasons
[00:43:32.991]on top of those that you listed that didn't allow you
[00:43:39.530]to see well why we couldn't see this event coming.
[00:43:47.194]Are we looking into the wrong place, the wrong causes?
[00:43:51.690]And I'm gonna be more explicit.
[00:43:53.514]Are we looking at the wrong place,
[00:43:55.514]and say should we be looking frequency
[00:43:59.329]and intensity of events, and how these are propagated
[00:44:04.127]into the rest of the components
[00:44:06.010]of the water cycle, including the storage?
[00:44:10.527]And then how does increasing storage,
[00:44:15.327]or if we can look into joint probabilities,
[00:44:18.387]or something that will include intensity, frequency,
[00:44:23.306]as well as the storage turns into a single thing,
[00:44:26.469]and then propagate the whole group of variables
[00:44:30.510]and state variables into the simulation or the forecast
[00:44:35.850]or the historical forecast that you put together?
[00:44:39.770]Yeah, I think I understand what you're asking is,
[00:44:42.170]you know, were we only looking at,
[00:44:44.577]maybe we were missing some of the key variables
[00:44:46.511]in terms of the intensity and frequency.
[00:44:48.657]And I think that that is probably true.
[00:44:50.191]We only had a short time interval,
[00:44:53.148]like a small budget on this, so we thought we'll try
[00:44:56.090]to cover what we think are the most relevant variables.
[00:44:59.487]But certainly, in terms of the impacts,
[00:45:02.447]kind of the status of the reservoir system
[00:45:06.129]in terms of capacity was important.
[00:45:09.610]Now I think there was a drought
[00:45:12.530]back in 2006 or somewhere around there,
[00:45:15.646]so it's a very difficult system to manage
[00:45:19.267]in terms of the infrastructure.
[00:45:21.150]Because on one hand, you want to guard against drought,
[00:45:24.430]but also leave enough storage for floodwater,
[00:45:28.350]so that's a challenge.
[00:45:29.966]I would be very interested, actually,
[00:45:31.310]to delve a little bit deeper into the statistics,
[00:45:34.830]as you suggest, because I think there might be
[00:45:37.326]some useful information there in terms
[00:45:41.049]of the likelihood of these things.
[00:45:42.809]So that's a really good point.
[00:45:46.567]Any other questions?
[00:45:57.467]Hello, I am Aziz, actually I am a PhD,
[00:45:59.230]a research scholar.
[00:46:03.034]So speak louder.
[00:46:05.544]Okay, presentation is very nice,
[00:46:09.582]and I am very happy to attend this.
[00:46:13.482]Okay, okay, sorry.
[00:46:16.403]The question is like how you isolate the temperature
[00:46:20.746]for the VIC simulation, actually,
[00:46:24.860]in the hydrological modeling prospect?
[00:46:27.678]Like, you took like the mean of the temperature
[00:46:32.083]when you are isolating the temperature?
[00:46:35.822]Right, okay so, the question is
[00:46:37.244]how did you isolate the role of,
[00:46:39.886]or just the temperature from the precipitation.
[00:46:42.126]Well one of the complicating factors is that due to
[00:46:46.163]the energy balance, when you have low precipitation,
[00:46:49.242]you will necessarily have higher temperatures,
[00:46:51.943]because of the way that the solar energy is partitioned.
[00:46:55.486]So we did a separate run that accounted for that.
[00:46:58.563]But in the simplest way, we had the historical observations
[00:47:02.082]of temperature, and we simply used
[00:47:04.947]the historical observations of temperature while
[00:47:06.718]setting the precipitation to a daily climatology.
[00:47:10.606]Then we did another run, which said
[00:47:12.424]that if the temperatures were so high,
[00:47:14.546]the precipitation would necessarily
[00:47:16.686]need to be a little bit higher than the climatology
[00:47:18.706]because of that energy consideration.
[00:47:21.688]And we found that there was a difference,
[00:47:23.464]but it wasn't a huge difference,
[00:47:25.891]when we looked at the results on the soil moisture.
[00:47:29.048]The inputs were a little bit different,
[00:47:31.032]but the results on the soil moisture were not,
[00:47:33.928]they were not qualitatively that different.
[00:47:38.408]The second question is the sensitivity
[00:47:40.936]of the antecedent moisture content
[00:47:43.747]in the simulation of hydrological model.
[00:47:46.104]Actually, I just wanted to know how you did
[00:47:48.408]that particular thing, like the sensitivity.
[00:47:51.662]Because in most of the paper I saw
[00:47:53.166]antecedent moisture content is one
[00:47:55.624]of the most important factors for the flood.
[00:47:57.646]Because ultimately, the precipitations,
[00:48:01.427]like the initial condition is the most important thing.
[00:48:06.483]Okay, so the question is how did
[00:48:07.784]we specify the antecedent moisture.
[00:48:10.744]So I showed you this hydrograph from 64 years of the model,
[00:48:15.406]the red line and the black line.
[00:48:16.990]And so, from that simulation we saved
[00:48:19.166]the state of the model.
[00:48:21.765]So the soil moisture and the snow pack
[00:48:23.667]on October 1st of every year.
[00:48:27.162]Then we, so the initial condition
[00:48:30.920]that we're working with was October 1st.
[00:48:33.166]And obviously, then we ran that October 1st condition
[00:48:36.846]with the different climates from the different years.
[00:48:38.922]Now obviously, October 1st is like nine months
[00:48:42.362]or 10 months away from when the flood happens.
[00:48:44.840]So it shouldn't be surprising that that doesn't
[00:48:48.126]tell you a whole lot about the flood event,
[00:48:51.148]because a lot can change between
[00:48:52.924]October one and June, when the flood happens.
[00:48:56.088]So we also ran another experiment,
[00:48:58.306]where we initialized it on March 1st.
[00:49:00.990]We got more sensitivity, but we really actually didn't get
[00:49:04.248]as much as we thought, and that's because, yeah.
[00:49:09.964]What's unique about 2011 is the really anomalous
[00:49:13.464]precipitation actually happened in April and May,
[00:49:18.581]and so the antecedent moisture hadn't
[00:49:19.619]registered yet from this precipitation event.
[00:49:23.077]So you would have actually had to initialize your model in
[00:49:25.464]April or May, because that's when the big anomaly happened,
[00:49:28.622]which made it much more difficult
[00:49:30.984]to predict than say the 1993 flood event,
[00:49:35.480]which was not as sort of as big of a spike like this.
[00:49:40.862]Does that make sense?
[00:49:42.583]Yeah yeah, thanks.
[00:49:47.904]Alright, got time for one final question, Dave?
[00:49:52.837]I'm curious about the temperature
[00:49:54.357]projections for the Northern Great Plains.
[00:49:57.157]Correct me if I'm wrong, but it seems like most
[00:49:59.045]of the warming experienced so far
[00:50:01.696]has fallen in the non-growing season.
[00:50:05.099]And so it's been getting warmer, but it hasn't been getting
[00:50:07.675]hotter with summer extremes, except 2012 was certainly hot.
[00:50:13.833]But this pattern of greater warming
[00:50:16.219]in the non-growing season versus the growing season,
[00:50:20.576]is that expected to continue, or do you think
[00:50:22.656]the growing seasonal warming is gonna catch up?
[00:50:26.877]Yeah, okay, that's a very good question.
[00:50:28.837]So is there kind of a seasonal component to the warming?
[00:50:34.108]There has been a seasonal component, is that gonna continue?
[00:50:37.170]So we actually didn't really look at that so much,
[00:50:40.210]other than that slide you saw with the increase
[00:50:43.730]in the days above 90 degrees Fahrenheit.
[00:50:46.508]Which, and I'll have to double check this,
[00:50:49.847]and this is actually an excellent,
[00:50:51.426]something that I can take and report back
[00:50:53.986]to the National Climate Assessment folks,
[00:50:56.004]so I appreciate that.
[00:50:57.410]It looked like there were fewer days exceeding 90 degrees,
[00:51:02.882]that seemed to be less extreme than the reduction
[00:51:06.327]in the really cold days, which kind of suggests
[00:51:09.650]that maybe this seasonal partitioning will continue,
[00:51:12.720]that you'll see less warming in the summer.
[00:51:15.484]But I haven't look at that explicitly,
[00:51:17.543]that's something very much worth looking at.
[00:51:20.720]So thanks for that question.
[00:51:23.186]Sorry I don't have a better answer.
[00:51:32.800]This summer, as you probably know,
[00:51:34.263]there has been a bit of drought in the region.
[00:51:36.802]Certainly, I think it's in South Dakota
[00:51:38.743]and certainly parts of Nebraska.
[00:51:40.242]I was just talking to a farmer over lunch
[00:51:41.884]who talked about very low yields, of beans in particular,
[00:51:45.724]due to drought in the key parts of the growing season.
[00:51:49.360]What really made you, what do you attribute
[00:51:51.778]to the fact that 2012 was over such a large area
[00:51:55.404]that we just don't see that kind
[00:51:57.324]of pattern when we look at spotty drought,
[00:52:00.306]which I think we could talk about for this year, probably?
[00:52:03.164]What really makes you see that as such a large region?
[00:52:06.514]Yeah, that's a good question.
[00:52:08.121]And I think that's actually more of like a large scale
[00:52:11.940]like atmospheric science, ocean atmosphere setup question.
[00:52:20.641]You know, I'm not an expert in that in terms of,
[00:52:24.264]I know that the connection between
[00:52:25.807]the drought and is not super strong,
[00:52:28.623]but there is some small signal there.
[00:52:30.847]And think that would be more of a place to look.
[00:52:34.047]And the difference in the sea surface temperatures,
[00:52:35.802]between the Atlantic and the Pacific,
[00:52:37.523]I know that was identified as something important
[00:52:39.663]in the Dust Bowl, which was another large drought,
[00:52:43.386]but I don't have, like I was trying to distinguish
[00:52:46.166]the work I did, it came after some of that large scale work.
[00:52:50.767]So there is that NOAA report that came out
[00:52:53.105]right after that may have some answers
[00:52:55.604]there that are gonna be relevant,
[00:52:58.186]because they looked specifically at that.
[00:53:00.106]So what were the drivers of this event?
[00:53:02.762]I just kind of look at the attributing,
[00:53:04.602]once the drying is there, then what can we say about it
[00:53:07.791]from a precipitation and temperatures point.
[00:53:12.490]And just to kinda add on to that,
[00:53:14.266]in 2011 there was significant
[00:53:16.548]drought in the Southern Great Plains,
[00:53:18.922]and then 2012 it kind of crept, or there was
[00:53:22.042]kind of a rapid onset into more of the central US.
[00:53:25.464]So there was something there in 2011 for the southern part.
[00:53:31.560]Good, well we're about out of time now, so thank you all
[00:53:35.220]for attending, and let's thank Ben once again.
[00:53:37.700](audience applauding)

The screen size you are trying to search captions on is too small!

You can always jump over to MediaHub and check it out there.

Comments

0 Comments

Ben Livneh - The Role of Climate on Hydrological Extremes in the Northern Great Plains Region

Description

Searchable Transcript

Comments icon comment

Related Channels

Comments