[00:00:22] Bridget Scanlon: I am delighted to welcome Huilin Gao to the podcast. Huilin is a Professor in the Department of Civil and Environmental Engineering at Texas A&M University, our rival and Huilin Huilin has received many honors and most recently became a fellow of the American Meteorological Society in 2025.
Her research covers a wide range of topics, but mostly hydroclimatology and focuses on reservoir storage and particularly evaporation at regional to global scales using remote sensing tools and modeling analysis. Thank you so much, Huilin for joining me today.
[00:01:08] Huilin Gao: Thank you, Bridget. Thank you very much for having me.
[00:01:12] Bridget Scanlon: So Huilin, we're both here in Texas and it's an interesting state and relies heavily on surface water, I guess in 2023, the Texas Water Development Board indicated that about 40% of the water use in the state was from surface water. So that was about 6 million acre feet relative to total water use of 15 million acre feet.
So much of this water is derived from reservoirs. Texas only has one natural lake Caddo Lake. And so many of these reservoirs were built after the 1950's drought. So from about mid 60's to mid 70's was a big, push on reservoir construction. And if you look at water use in the state, Austin, depending on the Highland Lakes, Dallas Fort Worth, lots of reservoirs in Houston, even.
So the total reservoir storage capacity in this state is 31 and a half million acre feet, and that's about twice the water use in this data. If we take 2023 data, 15 million acre feet is how much we use. So we are in fairly good shape at the moment with the reservoirs, right? It's about 70, 75%, almost full.
So Huilin can you describe, the spatial variability, temporal variability in reservoir storage, and give some background to the listeners.
[00:02:34] Huilin Gao: Okay. So, Texas, there's another fun fact, Bridget you mentioned there's only one natural lake. The other fact is that Texas is a state with the largest number of reservoirs. According to the national dam inventories. There are over 7,000 dams and the conservation storage capacity you just mentioned, it's mainly from this 189 large reservoirs with the smallest, still larger than 5,000 acre feet. So, if you look at the distribution of the reservoirs and reservoir storage, this is clearly related to the weather and climate in Texas. We all know that we have a very large precipitation gradient from the West to the East.
East can have like about 14 hundred millimeter rainfall, and West is like 300, 400 millimeter. So you see that most of the reservoirs are clustered to the east portion of Texas and a bit further now compared to the south part. And, as you said that right now we, the reservoirs, thank God, is in great condition, most of them.
And this, in addition to this storage variation is primarily driven by the seasonality. Like Texas has two rainy seasons right now. We are in the spring and then there's fall. And during summer is the drought season where lots of storage lost through evaporation. And there's also large inter-annual variations such like the 2011 drought.
And then 2015 16, 17, we had continuous years of flooding events. So large variations back to what it's like now. Is that, among the good news, there is this low storage line, if you start from the far north of Texas. And it's like a curve all the way drawing down to Corpus Christi, those reservoirs alongside land. The actual storage actually quite low.
[00:04:49] Bridget Scanlon: Oh, right. Yeah. So yeah, Corpus Christi was in the news recently. I mean, you mentioned the 2011 drought and I think it was looking up, you know, the reservoir storage went from about 30, 31 million acre feet, or it was maybe 80 or 90% beforehand the drought, but then dropped to like below 20 million acre feet.
And for listeners, 1 million acre feet is about 1.2 cubic kilometers, so they're pretty similar, those units. And so that 2011 drought, you know, it stayed low. 2011 was the most extreme year, but then those reservoirs didn't fully recover until about 2015, which was the wettest year on record in Texas, but slightly greater than 1941.
So, Corpus Christi has been in the news recently with their reservoir storage issues, and there was a nice NASA website showing the large decline in storage recently in those reservoirs. Can you describe that a little bit? Huilin
[00:05:46] Huilin Gao: Yeah. How about this? I can compare the 2011 drought with the current drought going in Corpus Christi. As you mentioned that the 2011 drought is so significant and is actually represent the most severe single year drought ever happened in Texas. If you look at, the annual rainfall across Texas, and for 2011, it's about half compared to climatology and I actually came to Texas A&M in 2012. When I first got here, all I heard is about 2011 drought. It was, so much impact. And I heard about, people talking about 2011 drought in Chile 2015, as you mentioned, that there's serious flooding that ended the drought.
But 20 11 drought brought down the storage capacity significantly. It did not hurt too badly to the reservoirs, to the east side, but places like where you are, Austin, like Lake Travis and Lake Buchanan like storage are record low, it took five years to recover. And in short 2011, drought is a statewide drought. I still remember that you published a paper in 2011 using the Grace satellite to look at the terrestrial water storage, and it's red everywhere. So yeah, that's the 2011 drought.
And let's look at this ongoing drought in Corpus Christi. So Corpus Christi has these two reservoirs, Lake Corpus Christi and Choke Canyon reservoir. It's down to the end of the curve. In fact, if we go up a little bit right now San Antonio River Basin also suffering drought and they have a lake, Lake Medina also with the storage in single digits. So let's talk about Corpus Christi Reservoirs. So, a few years ago, Lake Corpus Christi, like about five years ago the reservoirs almost full, like close to 90%.
And then there's multiple years of drought. And we know that when we talk about drought it's always started with the meteorological drought by lack of precipitation. Then there's, agriculture drought triggered by this and the shortage of soil moisture, and then go downstream. There's the hydrological drought in this case it's a reservoir-based drought.
So with this shortage of precipitation, the soil moisture's record low, and then even there's some rainfall. It's not enough to generate around most of the rainfall like infiltration, increase some moisture. And also, goes, through evapotranspiration And so the inflows to these reservoirs became record low.
And we also know that with drought there's oftentimes high temperature almost all the times and heat waves. So these will increase the reservoir evaporation. So if you look at this reservoir water budget, it is really against, the storage. And another thing I want to mention is that particularly what's happening in Corpus Christi is that the human activities, human factors play a role as well. Across Texas as you mentioned, like those large cities rely on surface water reservoirs and, so also, Texas has the largest population growth rate. Same with Corpus Christi. So increased domestic water use. And in the region there's a lot of industrial development over the years. So increased the demand.
At the same time, these reservoirs they have to maintain the environmental flow to make sure the flow going to the bay, can be good enough for the ecosystem service. So altogether this brings the drought very severe. In fact I think this current situation, they claim this is stage three water restrictions.
So they start to regulate the amount water to be used.
[00:10:19] Bridget Scanlon: Right. Right. And then, so you mentioned a couple of things, Huilin. So the 2011 drought, the worst one year drought, but then, there was some recovery in the fall of 2011 with the North Atlantic Oscillation, brought in some rain. But then the reservoir storage went back down, stayed low for until 2015.
And Mike Dettinger talks about drought busting floods, and it seems like sometimes you need a flood to end a drought because a normal rainfall will just get absorbed by the low soil moisture that builds up during the drought. So, and in 2015 then, we had the Wimberley floods and we had Halloween floods and things like that.
So, oftentimes the drought ends with the flood. So, it's the Corpus Christi reservoirs, the more recent drought is sometimes is not a statewide drought, but there are areas then that, are suffering quite a bit. And so, need to try to manage that. And thanks for explaining, the different types of drought, the meteorological drought, and then the agricultural from soil moisture and the hydrology drought.
And then some people talk, we sometimes in a permanent groundwater drought, if we are over exploiting it. So a lot of your work focuses on reservoir evaporation and these days people say, traditionally we've stored most of our water from wet periods to use during droughts in surface reservoirs.
And now they talk about storing water in the subsurface, in managed aquifer recharge, aquifer storage and recovery. and some of that is to try to minimize the evaporation losses and stuff. But maybe you can describe the importance of understanding reservoir evaporation and the different types of reservoirs that may be subjected to more losses, with larger surface areas and things like that.
[00:12:07] Huilin Gao: Okay, thanks, Bridget. Yes. So reservoir evaporation is very important for places like Texas and more generally for the Western US as well. This is because when you have a reservoir, picture like you have a swimming pool, the evaporation is affected by what's happening in the atmosphere.
What drives it is a vapor pressure deficit. So the drier is the air, the more likely for you to have larger evaporation. And also the energy part is that the warmer is a temperature, then more likely have the evaporation. And as you mentioned earlier, there was no reservoirs, literally in Texas until after the 1950s drought. These reservoirs are constructed, most of them, for water supply purposes, and when they are most useful are during the summer, like during the drought season, and then that's when the reservoir evaporation is the largest. I can throw a number here for instance, in Texas, on average, the annual gross evaporation loss is about 2.4 million acre feet.
And how much is that? Earlier you mentioned that the total water use, includes surface water and the groundwater combined every year is about 15 million acre feet. So this is equivalent to 50% of this water use loss through the evaporation. And this reservoir evaporation is most significant during the drought years or drought seasons.
When the worst reservoirs are managed, they're looked at, what comes in, what leaves the reservoir, right? Evaporation is this invisible loss. And then there's also precipitation. So, we use this term a lot is net evaporation, is the evaporation minus the precipitation. So in good years, wet years, then, even if you lose a lot of water through evaporation, you gain a lot from precipitation, so you are good. But during drought years is the opposite. That you have a shortage of precipitation and you have extra amount of evaporation loss. And going back to the reservoir storage, we talk about, distribution of the reservoirs. They are driven by the precipitation gradient But if you look at the reservoir evaporation rate map it goes the opposite way. It is the places where doesn't get enough rainfall means like they have arid, semi-arid conditions. They have the largest evaporation rate. This cause the thing is that I always compare like reservoir storage, like your savings, like when you don't have lots of savings, your expenditure, you know, rate is very high.
And you mentioned that talk about different types in general, in addition to the rate, there's also the surface area. So what we really care about is the reservoir evaporation volume. This is one thing. So larger reservoirs tend to lose more. And then, different functions of the reservoir functions oftentimes is associated with the shape of the reservoirs.
For instance, like for hydropower generation, there are lots of these reservoirs called run on the river. In general, they're fairly full because you need that elevation potential to generate hydro power and those type of reservoirs, they usually don't have too much like evaporation issue, there's evaporation, but it's not a concern.
But this water supply reservoirs, like in Texas, the topography is fairly flat. The reservoir is relatively shallower. These are the places that we need to be extra careful with the reservoir evaporation. Actually, the agencies are very, cautious about how to account for reservoir evaporation
[00:16:13] Bridget Scanlon: Right. And Huilin I like your bank account analogy. You know, because I think you know, you really need to look at all the inputs and the outputs and, to understand the budget and precipitation levels or rainfall levels and evaporation then how much water you are diverting for use and things.
And you mentioned the precip and temperature gradients in Texas. I mean, I really think of Texas as a template for the US because like the 98th or hundredth meridian goes through the center of the state. And we kind of like in analogy for the entire US, the Eastern US is humid and the western US is semi-arid.
And so, we are a good field model for the us, I feel. So traditionally, you know, we've relied on pan evaporation to estimate evaporation and did that for decades. Can you describe that a little bit just to give context to the more advanced approaches that you've been using recently?
[00:17:12] Huilin Gao: Okay, a class A evaporation pan is commonly used by National Weather Service not just for measuring like representing reservoir evaporation, but also for evapotranspiration in general. So what's a class A evaporation pan looks like I always call it, it is like a large size, bird bath.
So, it has four feet diameter, 10 inches in depth and it is installed on a wooden platform. On the ground, it is metal containers. The wood prevents the heat transfer between the pan and the ground. So typically how you take the measurements is like a rain gauge. You start with the same water level every day and you check and you factor in if there's rainfall or not.
And the advantage of evaporation pan is that it's fairly simple, easy to install and inexpensive. And again, deck is of record is very helpful and there are limitations. Picture this, how we can use this evaporation pan to represent the large reservoirs or lakes right typically practice the use some empirical coefficient. When you say coefficient generally means they are fixed. Right. But we know that every reservoir is different. They have different asymmetry, they have different sizes and different depths is very important because that affect the heat storage. But the evaporation pan does not represent the heat storage effect, meaning that you put an evaporation pan back to that even next to the swimming pool, you know that next to each other, they don't have the same evaporation.
Right? So the location also matters a lot. And even this instrument is relatively cheap. We don't have many of those. In Texas, both the NOAA has a number of like about 18 and Texas Water Development Board installed another 64 evaporation pans. But we all know Texas is large.
So if you divide this number 82, divide the surface area of Texas by this 82, what we get is that we have one evaporation pan every 3,280 square miles. What does that mean? That means almost exactly you have one evaporation pan in 10 sizes of the city of Austin. So very, very little.
And you know that in Austin there's the Highland Lakes. There are so many of them and they are different, And so what. The data are used because they are not evenly distributed. Texas Water Development Board generated this gridded reservoir evaporation data. It's one degree by one degree. One grid cell is equivalent to, again, 12 sizes of Austin. So this introduces some issues and the data are monthly. And if you look at the observations across the US the water body turn for the reservoirs from inflow, outflow, reservoir storage, and, evaporation data at daily times step we have less than 10%, even less than that.
[00:20:56] Bridget Scanlon: Right, right. I really like it, the image that you have in some of your presentations of the evaporation pan and all these vultures standing on the edge, drinking out of it, which kind of amplifies your bird bath analogy.
[00:21:10] Huilin Gao: Yes. Yeah,
[00:21:12] Bridget Scanlon: Thinking that we can rely on that then, but I mean, we are always advancing and improving things I'm really impressed with what you have been doing, providing daily data at a much higher spatial resolution using a Penman equation.
And then I think your work then emphasizes the importance of heat storage in the reservoir and wind fetch on reservoir evaporation. So maybe you can tell us a little bit about your methodology to get to the daily data.
[00:21:40] Huilin Gao: Okay, thanks. We can start with evapotranspiration and evaporation in general, the same. There are many methods out there. We talk about measurements and there's modeling approaches. Some modeling approach is based on radiation, some based on temperature, some more simplified, some more complicated.
And Penman’s equation is one of the most commonly used and trust approach for calculating open water evaporation. The advantages I would consider one is the physics it represents. Two is where the data come from. First of all, when I talk about open water evaporation, I'm not talking about reservoir, just open water in general. So pan evaporation has two components. One is the energy component, the other one is aerodynamic component. So we call it combination method. So using these two components together, it factors in all these meteorology variables we talk about, the radiation, like net radiation, the temperature, the wind speed, relative humidity, all those things.
And this affects the evaporation rate, and we know that. And so that's the physics it represents. And then the benefits, when I talk about those meteorological data, they are handily available. All meteorological stations collect measurements for those, and we have very good re-analysis data to provide those gridded data sets.
And what we contributed is that if you look at this Penman equation, we have two parts, right? To start with energy part. If you look at it, there's the net radiation minus the heat storage change if you look at the original Penman equation. But when people use it, most time they ignore the heat storage term because oftentimes it's small.
However, it's not small for reservoirs, especially the large reservoirs. So how we did this is that, we factor in, we calculate the heat storage change term. In the paper we have about 12 equations there, I'll just be brief. So the heat storage change term becomes a function of the reservoir average depth as well as its water column temperature. Water column temperature we estimated using an equilibrium temperature approach. What's the benefit of considering heat storage? It's that, picture this back to the swimming pool. When it's warm, you go to the swimming pool. It's really cold. Because the water absorbs the energy to heat up. So there is a time lag between the air temperature and the water temperature.
So in the spring season and the summer the lake absorbs heat, and then in the fall it releases heat. So if you picture this compared to the pan evaporation, you see that the peak time will be different. And also we know that the deeper water temperature is lower than like shallow water. So the peak value would be different. So we modified this equation by adding this heat storage term. Back to it and carefully, using physical based literature and to quantify the, heat storage term.
And then there's the wind fetch. The wind fetch is basically when wind blows over the water and we try to use the, relative humidity from like gridded data, like over land, then you know that that relative humidity will be different when it blows over the lake.
So we introduce fetch length which factors in the size of the reservoir and relevant to the direction of which direction the wind blows so that we have this aerodynamic component modified as well. So that's what we did. And when we have this algorithm, we are able to use it, calculate daily evaporation rate for individual reservoirs with the reservoir area and depth factored in.
And when we have the result, we did compare it with Eddy Covariance observations. Those are the flux towers, considered the most direct accurate measurement. Of course, those measurements are very expensive. So they are generally used for supporting like research, like providing data to gain physical insights and validation.
[00:26:45] Bridget Scanlon: Right, right. So I mean, you really have transformed our knowledge of evaporation from reservoirs in Texas. And you provide Texas Daily Reservoir evaporation data and only a lag of like two days to update it. So that's really impressive. Can you, and I will put the link to the highlights for this podcast. There's a lot of spatial variability as you were describing in reservoir evaporation even in normal years across the state. Maybe just emphasize that a little bit and what the app does, what data it includes and the time period it covers and things like that.
[00:27:23] Huilin Gao: Okay, so this is a monitor actually supported by TWDB and the Lower Colorado River Authority and the Army Corps of Engineers, the Fort Worth office together and in collaboration with my colleagues from Desert Research Institute, who helped to translate our algorithms into this API. So the website covers data for this 189 Texas reservoirs.
I mentioned earlier from 1980 to present with two days latency. And the variables we put on the website includes the gross evaporation rate and the net evaporation rate, which is gross minus the precipitation rate. We add participation rate as well. We provide the reservoir surface area and the reservoir evaporation volume.
And at current website, the reservoir surface area is from the observations provided by TWDB's website. So you will see some locations has like a constant value I can talk about later how we fix that. And so these are the data we provide. We have daily, monthly, and annually. You can easily plot them as you wish. And download the data, you can download the image, you can download the Excel spreadsheet. So that's how this website works.
And you ask me comment about the spatial variability. So there's very large spatial variability of this reservoir evaporation. As we mentioned earlier, we, in general, see the drier, the middle region has higher values than the east side. And we also mentioned that we brought in the heat storage and wind fetch as a result. Even for like Highland Lakes different locations, they have different reservoir evaporation values. Take an example like for instance In Austin area, you have Travis and you have, Buchanan and one's deeper one's shallower, then their evaporation rate will be different. The deeper one tends to have a relatively smaller rate, and then the shallow one has higher rate and peaks earlier in the season. So there's large seasonal variations as well.
And the largest variation is the inter-annual variability. When we, going back to, you know, during drought years is very significant. If we look at the net evaporation we see that on normal years, East Texas, the evaporation loss and the precipitation yearly balance out each other. But in the west side there's a larger evaporation about, let's say three feet per year in terms of the net evaporation.
But if we look at 2011 take Central Texas like where you for example, normal years is about two feet naturally evaporation, loss. And then during drought 2011, it was four feet. and remember earlier we said the storage was about 30% during 2011. And then when your storage cut to less than half and your evaporation rate gets doubled, we are talking about net evaporation.
So that's really concerning. Yeah.
[00:31:14] Bridget Scanlon: Right, so it's nice that you had this collaboration with the Texas Water Development Board and the Corps of Envineers and Lower Colorado River Authority, and I know the late Ron Anderson was very interested and in the research and everything, and a great proponent for the work.
So that was really nice. So you have people, stakeholders, then you're working directly with them who are going to use the data and you understand their needs and all of that sort of thing. So, that's great. And I guess Texas Water Development Board uses the data for planning and, LCRA operations and forecasting and the Corps of Engineers for many different purposes. Maybe you can describe that just a little bit.
[00:31:54] Huilin Gao: Okay. I can start with a Texas Water Development Board. Is a water planning agency. So, when they make state water plans, they run the model, which is called the Water Availability Model WAM, which developed by Ralph Wurbs professor who works here at Texas A&M. The WAM model has two primary inputs. First one is a naturalized flow. And naturalized flow means that you have the observed USGS flow and you take out all those withdrawals and get to the states. This is the river flow without any human interventions.
And then there's the next one is the reservoir evaporation.
So these are two major inputs for the WAM model.
And then, they use the WAM model particularly to run it. And an important variable output from the WAM model is called reservoir firm yield. So what is firm yield? It is the maximum amount of water that reservoir system can reliably supply during the record drought event. In this case, for Texas everything's compared to the 1950s drought. So that's the formula, firm yield. So this firm yield determines the water right and determines how the water is allocated. Going back again, when they make plans they need the model, they need input and reservoir evaporation is part of input.
We already talk about the possible biases that those biases with the pan evaporation they have been using. So they have a strong interest of looking at this data and whether, we can improve the quantification of reservoir evaporation, for example over the last 40 years, according to our data and many other studies that evaporation rate has been increasing because, the warming temperature.
And then, across Texas, it increased from like, 48 inches per year to 52 inches per year. So they want to account for this nonstationarity into this management planning.
Then there's Lower Colorado River Authority. And yeah, the late Ron Anderson was a big supporter and interested in this daily reservoir evaporation.
And because among the Highland Lakes, the two largest ones are Lake Buchanan and Lake Travis, right? And so there's upstream and downstream. And downstream one is the deeper one, but downstream also has a larger contributing area is more vulnerable both to flooding. so how you account for the storage and balance the storage between the reservoirs as a system is very important. By having a better understanding, you know, input for the evaporation, they can try to avoid downstream flooding, but also at the same time maximize the system supply.
The USACE Army Corps, Dallas Fort Worth has their own hydrological model and one of the model input is the reservoir evaporation rate. So if they account that better, they have the same issue. All reservoir manager says that, in Texas, you struggle. You either have flooding or you have drought. Right? And so you manage them as a system. Earlier we talk about the storage capacity, the conservation storage capacity, right? And as a water manager, what ideal case is that you want to keep the water level right at top of the conservation pool so that you have plenty of water for supply, but you also have plenty of room for the flood control pool but it's easier to say than to do it. That's why , they want to, factor in the evaporation and they improve the inflow estimation. In fact many of the water reservoirs use the reservoir evaporation information to back calculate, to estimate the reservoir inflows as well. So that's some examples how these datasets can be used.
[00:36:27] Bridget Scanlon: Well, it's great that you're working directly with all of these different groups and I know Ron Anderson used to say, a one year drought is not too bad, but a multi-year drought becomes much more challenging. And then having more reliable information on the losses through evaporation then can help them hopefully manage the systems better.
And, of course, the Corps of Engineers is very interested in changing the rules and regulations, possibly the term forecast informed reservoir operations to try to better adapt to the increasing climate extremes that we experience, more floods and more droughts, to optimize how much we can store.
So, you have done studies across the entire US and also globally. So in Texas you had the bathymetry data and the area data from the Texas Water Development Board and then you were able to convert the changes in evaporation to volume using those data.
Maybe you can describe a little bit what you work is in the US and using reservoirs throughout the US.
[00:37:33] Huilin Gao: Okay, in the US we have, look at over 700 reservoirs and we generate similar data record and how we estimate the reservoir surface area is that we used the Landsat satellite image classification to help so that we can calculate the reservoir storage. And we cross the US, what we look at is that we found Texas does stand out as a hotspot along with the west US states like Arizona, California. And we also found clear significant trend across the US. And primarily the driver is the short wave radiation and changes over the past decades. And then you want me talk about global as well or
[00:38:28] Bridget Scanlon: Well, no, just describe a little bit I think you, your studies, you estimate maybe, across the US 34 cubic kilometers per year is an average reservoir evaporation. And that's, putting that in context. Then maybe that's similar 90% of the annual water use for public water supply systems and um, that sort of thing.
And you've done work in Colorado also, I guess maybe to validate your work lake Powell and Lake Mead, and of course everybody's looking at those these days because of the, mega droughts since about 2000. and you know, I forgot to mention, I would really like to put in a plug for the Texas Water Development Board and the Water Data for Texas and the reservoir storage data because you can get the recent data, you can get the long-term data and all of those things, but in many states or even from the Bureau of Reclamation or anything, you're looking for reservoir storage.
It's oftentimes very difficult to get both the recent data and the long-term records and everything. So, so that's a huge plus. And so your data also provides that for the US, right. You provide the reservoir storage.
[00:39:37] Huilin Gao: Yes. Yeah, actually our data not just with the evaporation. We are working on daily evaporation monitor for the Western US. So we have a similar website actually developed. We have not released it publicly, say release, but it is ongoing and it's two day latency.
Yes, as you mentioned that across the US like take Western US as example, all these states together, we look at, there's nearly 500 large reservoirs. Over 300 of them don't really have the complete data like elevation storage area. This really affects our capability of managing these reservoirs.
Earlier we talk about, like Dallas Fort Worth or, lower Colorado, reservoirs come as a system. It's not like you manage them individually. So that constrains the capability. So what my group has been doing over the last 10 plus years, ever since I joined here is to use satellite data observations to help fill in this gap.
What we do is that picture, this, you can do image classifications for when you have a satellite image you can extract. Nowadays, actually we've been using this machine learning algorithm a long time. We just didn't really call it, like we extracted the surface area, we develop algorithms to extract the area regardless. There's partial cloud cover, and then when we get this area, we combine it with the area elevation storage curve. That's something else we actively work on, is that we get radar or lidar data and combine that with the area so that we can develop the curves. Once we have the curves, we apply the area to the curves to get storage time series.
This is because those data are very sparse in time and also oftentimes limited to large reservoirs. So basically in remote sensing, there is this thing you can't get both high spatial resolution and temporal resolution together. So we want to leverage, and of course there are like more recent advancement that makes this relatively easier.
In short, through satellite remote sensing, we can gain a lot of information that we did not have the opportunity to collect In situ.
[00:42:22] Bridget Scanlon: And that's fantastic and it'll be great when you have this storage data out there for the US because I can remember asking the Bureau of Reclamation for Data on Lake Sakakawea and that's the third largest reservoir in the us. It took me months
[00:42:38] Huilin Gao: Yeah. even the large reservoirs, they have data. They are not, in one place to go. Yeah.
[00:42:46] Bridget Scanlon: Right, right. So you've also done analysis of reservoir evaporation at the global scale, and one of your papers, I think at 1.4 million Global Lakes. Maybe you can just describe that a little bit
[00:42:58] Huilin Gao: Yes. So with that work, it's in the line of the same idea actually, we calculated the evaporation rate and volume for each of these 1.4 million global lakes, this including lakes and reservoirs, from 1985 to 2018. And because we use this Penman-based method, this is actually the first time we use physical model to estimate the evaporative loss from the surface water.
What we found is that on average the evaporation loss is about 1500 cubic kilometers. If we put this in context, it is about three times of the Lake Erie storage capacity, so it is quite large. And then we also found increasing trend with about three cubic kilometers per year.
And so this increasing can be driven by different reasons, like increase of the evaporation rate, typically like in, not the high latitude. And in the high latitude, it's more like, reduced ice cover. because then you have more surface area exposed for evaporation or like increased surface area of some lakes.
Example, like the Tibetan Lakes they have been increasing. Another interesting finding we get from this work is that since we have both lakes and reservoirs, we compare them, we found that these manmade reservoirs, which only account for 5% of the global storage, actually contribute 16% of the total lake evaporation. And it also increases as a much faster pace than the natural lakes. And this actually, you know, really because, when it comes planning. We need to take good care of this. And the reservoir evaporation every year is about equivalent to 230 cubic square kilometer, and that's equivalent to about 20% of the global water use.
So that's quite a bit.
[00:45:19] Bridget Scanlon: Right. Right. And another aspect that I was very interested in your work is this integrated reservoir drought index. you know, And, we've seen the importance of reservoirs and, the impacts of drought. For example, in Cape Town Day Zero, they were all heavily reliant on six big lay reservoirs in that region.
And also Sao Paulo has been subjected to major droughts and they have another reservoir system that they're rely on heavily, not that much groundwater. So this drought index for reservoirs be extremely valuable for these large cities that are relying heavily on reservoirs.
Can you describe the drought index?
[00:46:01] Huilin Gao: Yes. So, for reservoir-based drought, a very common practice is to use the reservoir storage anomaly to develop drought index. Really, we have so many drought indexes there. And so that when we can see this drought condition compared to long-term climatology, how severe it is. And in this work, the reason we brought in the evaporation is because if you look at this, drought is also associated with, oftentimes lead to high evaporative losses, right? And the evaporation loss has a shorter persistency compared to the storage. Going back to the very beginning when we said like, there's a meteorological drought and triggers the agriculture drought, the hydrological drought, then there's a propagation time.
It takes time, right? So, but compared to the storage, the evaporation has a faster response. Much quicker. So when there's a drought onset, if we factor in the evaporation into this drought index, we possibly develop the capability for, early warning of reservoir based drought. It not necessarily work for all cases, but we did notice for some locations that putting in extra information from evaporation will improve our capability to, estimate reservoir based drought.
And also this kind of inspired that we also develop a global reservoir, water reservoir products using the MODIS and VIIRS satellites for NASA. And that's a project where we have simultaneous storage and evaporation. So we think, oh, why don't we use it?
[00:47:59] Bridget Scanlon: There's a lot of advances in remote sensing these days and satellites and everything. And, and as you mentioned, you've been using machine learning without calling it machine learning for a long time. So what are you most excited about with the recent developments and how is it helping you with your evaporation analysis?
[00:48:19] Huilin Gao: Okay, so, there are a few exciting things going on. I can start with the remote sensing first. We are really in a year better than ever because, when I first started this line of work about 15 years ago, we used MODIS, which has a resolution of 250 meters and in combination of radar elevation measurements, which has a footprint of several kilometers. So you can only capture the large lakes. Like with my very first publication I developed we call it global data set, but there are 30 large lakes 15 percent of the global storage. That's limited by the observation capability, right. And the, spatial temporal coverage.
But fast forward there's like more satellite, especially when the Landsat data became public available, made free by USGS as a first step, and then there's the Google Earth Engine, which made it even easier. You don't need to download the data. So, what it's like now is that for instance, there's European satellite, the Sentinel two.
So NASA has this product is called Harmonized Landsat Sentinel two, which brings us the 30 meter resolution sub weekly data so that we can estimate compared to monthly, right? We have much higher spatial and temporal resolution. And then on top of that, a lot of this development for estimating the area and for estimating the storage is the elevation data availability.
There's the SWOT mission is called surface water ocean topography mission. That gives these two dimensional elevation of the surface water across the globe so that we can get to the smaller reservoirs. There's still lots of room for us to improve the storage estimations for smaller reservoirs.
And then there's this cube sat like they bring to like three meter resolution images. And then, when you have this high resolution, the return period is long. But they are acute, they are cube sats. They launch hundreds of them so that you have daily measurements. So these are all very exciting stuff.
And then back to the evaporation. What does this mean? This means that we can better estimates the reservoir surface area, and we can get more dynamic elevation of this reservoirs. And we provide more comprehensive view of the water balance for these reservoirs. We know storage, we have the evaporation. And then, the SWOT satellite, a big part of it, is also estimated the river discharge. And specifically, a couple more things we're doing with the reservoir evaporation. Earlier you mentioned the forecast informed the reservoir operation. We are developing, at the early stage, developing a 28 day forecast capability for Texas reservoirs so that, this can complement those stream flow forecast. Actually, I'm fairly optimistic about forecasting skills with the reservoir evaporation because what matters most is during the drought seasons and then we don't need to worry too much about the precipitation and we have much better skills with the other meteorological variables. So that's, one thing.
And then I mentioned that we are expanding this to western US, hopefully across the corners and with the idea using the remote sensing data to add to not only the surface area, but also the reservoir storage so that we can help boost the data availability and hopefully lead to better management.
[00:52:30] Bridget Scanlon: Right. Well, that, really is exciting and with all of these new tools and data sets and everything we'll just be expecting more and more of you so you'll never be able to rest. So we can never get ahead of things. The more-
[00:52:44] Huilin Gao: Well maybe, who knows with the AI?
[00:52:47] Bridget Scanlon: Yeah,
[00:52:47] Huilin Gao: Yeah.
[00:52:49] Bridget Scanlon: You just have to turn it around faster and faster.
So our guest today is Huilin Gao. She's a Professor in the Department of Civil and Environmental Engineering at Texas A&M. Thank you so much for explaining reservoir evaporation and storage and its importance and drought and all of these things. Really appreciate it, Huilin.
[00:53:08] Huilin Gao: Yeah. Thank you so much, Bridget. This is a great opportunity to communicate.
[00:53:13] Bridget Scanlon: we will put highlights and script on the website and also links to papers and data sets and apps and everything so people can see where the data are available and all that sort of thing.
[00:53:25] Huilin Gao: Okay, great. Thank you.
