[00:00:00] Bridget Scanlon: Welcome to the Water Resources Podcast. I am Bridget Scanlon. In this podcast, we discuss water challenges with leading experts, including topics on extreme climate events, overexploitation, and potential solutions towards more sustainable management. Today, I would like to welcome Petra Döll to the Water Resources Podcast.
Thank you so much for joining me today, Petra. You're welcome. Petra is a professor of hydrology at Goethe University in Frankfurt. And prior to working at the University of Frankfurt, she worked for many years also at the University of Kassel, both in Germany. Her research focuses on global hydrologic modeling, particularly the WaterGAP model, and also transdisciplinary knowledge integration and participatory research methods. And today we will cover both of those topics, global modeling and stakeholder engagement approaches and methodologies. She has contributed extensively to the IPCC reports and received numerous awards, including the Henry Darcy Medal from the European Geophysical Union in 2019 for her work in global freshwater system modeling.
So we met many years ago, Petra, when you attended a conference in Nebraska on the High Plains where we were both attending the Dougherty Water for Food conference in Nebraska. And I always appreciated how you try to bring in process understanding and learning from different regions into your global modeling.
We're hearing a lot these days about very hot conditions in and around the Mediterranean, parts of Europe, and drought, and maybe you can describe a little bit about what's going on in Germany these days with the drought conditions.
[00:01:56] Petra Döll: So in Germany, one could say that we had now five years of, let's say summer drought, but also winter drought, but in the summer drought is more visible and has a more stronger impact, particularly on our forests, but also partially on water supply.
I mean, Germany. It's not like the Western United States. I mean, it's a humid place. So we have been used to having always plenty of water and the problem were more floods and water quality, but not so much drought. So when, with our also global model, we started to work on drought. This really, I was one of the few people in Germany being interested in drought because I was looking outside of Germany.
But now, I mean, yes, we had in 2003, this very strong central European drought. But then afterwards, everything was back to normal. And then it started in 2018 with a very strong summer drought. So starting July. And I would say that in Germany, the situation has not really recovered. I mean, we had a little better year in 2021.
And what is visible, for example, is that now in the news, you always have the newest maps. From the German drought monitor, it's a soil drought monitor, which, runs every day with precipitation from 2000 stations in Germany and then runs at the set. This is a research organization, not a university, and they run their model every day and produce, so to say, drought in the uppermost 25 centimeters and down to two meters.
And that has gained very high public attention. Now, and everybody, again, now starting this month, July, in the newspapers we have, we cannot go on like that. How can you water your garden without watering so much? And so it's really all of a sudden, it has become a topic in Germany for Germany.
[00:03:53] Bridget Scanlon: Right and not so long ago, you had extreme flooding in parts of Germany with fatalities in Northwest Germany. And so I think it just emphasizes the increasing extremes that we're experiencing floods and droughts. And you mentioned that Germany is a humid country. And so you're not accustomed to droughts.
And then maybe our days in the past when we looked at droughts in West Texas and stuff, they're always used to droughts. They never it much water, but for humid regions that are accustomed to having a lot of water, it's maybe more, more challenging. Are you seeing any impacts on energy production or cooling for thermoelectric power?
[00:04:36] Petra Döll: Yeah, in the extreme, particularly in 2003, that was a big thing because there, there were still more power plants with once-through cooling instead of cooling towers. Now it's, I think it could still be a little of a problem, but I mean, those power plants don't need so much water anymore because they're all cooling towers now.
So it's more the transport on the Rhine and the other rivers that are for shipping. And it is more really the forest and of course also agriculture.
[00:05:08] Bridget Scanlon: So you're very well known internationally for your global hydrologic model, the WaterGAP model. And last year you celebrated the 25th anniversary of the model, which is great. Maybe you can describe a little bit how that work began and how you became involved in it and maybe a little bit about how it works.
[00:05:28] Petra Döll: Yeah, so WaterGAP is a global model for on the one hand water resources. So how much water is provided by, so to say precipitation minus evapotranspiration. But then always from the beginning, it was also a model of human water use. So we, from the very beginning, it was there to say something about water scarcity or water stress, and particularly under the impact of not only global climate change, but in general global change. So that was the idea that came from Professor Joseph Alcamo. Professor Joseph Alcamo is somebody who worked before at IIASA (Intl. Institute for Applied Systems Analysis), this very International Systems Analyst Institute in Vienna, and later he was heading the development of the IMAGE model (Integrated Model to Assess the Global Environment), which is an integrated assessment model. And then he got a professorship at the University of Kassel and was looking for a, what is called Wissenschaftlicher Assistant, which is sort of an assistant professor, but you have a boss, let's say, so it's a postdoc position for six years or seven years. And he was looking very broadly to do something at the global scale, so maybe atmospheric modeling or oceans or, oh, he said water.
And when I saw this in the position and the job description, I thought, oh, okay, let's apply. And I think because I came, I got the job. That's why then I thought that the best choice is to develop a global hydrologic model, because in this integrated assessment model of RIVM (Dutch National Institute of Public Health and the Environment) in the Netherlands, this IMAGE model, water was not in there.
And it was very clear that, I mean, if you want to see how the global, let's say, emissions and climate and everything and land cover is developing, I mean, you need to have water on the continents in the system. But that was not the goal, to integrate it into IMAGE, but just to have a standalone water resources and water use model, and I started working there in 1996, and it was really possible to do that because of the connection that Joe had to the former colleagues, for example, regarding climate input. Now, at that time, just having global scale information on precipitation and temperature. I mean, that was not around, but also in the UK, but I mean, that was the connection. I mean, it was his idea and also his knowledge about how to do global scale work was the initial beginning, so to say, but then because I'm a water person, I mean, I brought in the, let's say the hydrological knowledge and also modeling because I came from groundwater modeling, then I did modeling in the unsaturated zone. And so, I mean, I sort of say it came then to the global scale to bring that all together.
But also what I found really interesting was also the water use modeling. So, for example, we also then talked to a water economist and he actually joined me on a trip to Brazil. So I learned a lot and a water economy now.
[00:08:35] Bridget Scanlon: Right. So you mentioned that WaterGAP was developed partially to address water scarcity concerns. And it basically, I guess, would you describe it as a water budget model? You look at the inputs and the outputs and then evaluate change in storage. A lot of the other types of global models at that time maybe were developed by climate scientists and they were used using them for the lower boundary condition for their climate models and they were maybe more physically based but they didn't have that human water use component and so that was a unique aspect of WaterGAP that was bringing in the water use so that was critical.
[00:09:14] Petra Döll: Definitely, for example, when we then thought, okay, irrigation outside of Germany, at least, this is a major water user. I thought, okay, we go to FAO because FAO, Food and Agriculture Organization of the UN is in charge of irrigation because that's when you look at their websites and so on. They should have information on irrigation worldwide.
Well, turns out, well, they had some country values for that estimates of country values, If you look from a hydrological perspective, this is just what we didn't want to see. We did no longer want to see like, like one water scarcity value. So water use over water resources for China or for the U. S. I mean, that doesn't make any sense.
So our goal was in the beginning to go for river basins or drainage basins and to make assessments at that scale. And so FAO didn't have that, those maps of irrigated areas available. And also, I mean, when you looked at the literature, there were some publications where people have drawn some polygons on maps and said, well, this is the irrigated area.
So, I mean, the water use part was really not well covered at all at that time. And I think we were very lucky that we had actually, at that time, a student, not a PhD student, but really a master's student, Stefan Siebert, who is also now a professor, that he was working with me as a student assistant. And then we decided to go to FAO and ask them if we could, let's say, in cooperation with them, of course, work on such a map of irrigated areas or areas equipped for irrigation.
And they agreed and Stefan stayed there and went to their, so to say, in their cellars and went through all the folders they had from their experts from the different countries who once in a while sent in reports on new irrigation projects. And that's how Stefan in the end, I mean, together with me, but it was mainly his work, how we came up with this first map of areas equipped for irrigation, which then allow it to do then now what modelers do.
Now you think, okay, now I have the area. Well, what approach can I take to estimate the irrigation water demand, which is, I mean, there's many approaches and we just took always simple ones, or, I mean, in that case, tried out a more complex one and compared it to the simple one and then decided for the simple one.
But only with these geographic data, so to say. Now I'm in the geography department. I wasn't before. Without these geographic data, you can't go anywhere. And I think that also is something, but in the beginning of WaterGAP, we put a lot of effort in creating data sets that at that point in time were not available anywhere else.
[00:12:02] Bridget Scanlon: Right. Yeah, your work with Stefan, I mean, is well known internationally. And I mean, I think you guys were the first to point out that irrigated agriculture was responsible for about 70% of global water consumption and 90% of global water withdrawal. The other way around, the other way. Sorry, 70 percent of water withdrawal and 90 percent of water consumption.
And so, so that is extremely important because the elephant in the room then is irrigation. And that links water use and water scarcity issues to food production globally. And so that was extremely important to have the spatial coverage and then understand the importance of irrigated agriculture. So some people talk about, turning off the faucet when they brush their teeth and things like that to conserve water. But I mean, maybe what they're eating might be much more important considering water footprints and stuff. So it's great to have this global picture then with these global models and the UN relies heavily on these data, looking at, water issues globally. I think another aspect of WaterGAP that may be a little bit unique is that you calibrate it, minimally calibrated global WaterGAP to streamflow data. Maybe you can describe that a little bit, et cetera.
[00:13:22] Petra Döll: Yeah. So as I said, I mean, the major focus in the beginning was to come up with water stress indicators, more or less water demand or water use divided by water resources. So then when you see how, then how do you compute water resources? I mean, there's strongly a function of input data, of precipitation input data, and of course, at a global scale, at least in places outside of, let's say, Central Europe and United States, let's say, the number of gauging stations is very low for each, let's say, half degree grid scale, what grid cell, what is the spatial resolution of WaterGAP, and in the very beginning, for example, the global precipitation climatology center, they didn't want to come up with half degree mapping datasets, because they said, no, from a meteorologist’s point of view, we do not have enough gauging stations. I have to give good self to say anything robust. Okay. So very obviously, I mean, in some areas of the globe, you didn't have the slightest idea how much precipitation was falling, like in the upper Amazon or in the Himalayas.
Now, when you think about all these big rivers that come out of the Himalayas. So what do you know about precipitation in Himalayas? So I think it was there and then in the beginning, I always thought, Oh, only precipitation is the uncertain thing. And then when, for the first time we could get our hands on a second data set on radiation information, So that we need to compute low radiation that, or shortwave radiation, that we need to compute net radiation for potential evapotranspiration.
When we, when I got my hands on a different data set from the ones that we have always used, I saw what is it, what a huge impact it is, the uncertainty of the radiation data. No, so everything that happens in terrestrial hydrology is so much dependent on input of climate values. And then of course we have all these other things that we don't know what happened on earth and what the parameters and so on.
And it was very clear that whatever we would compute can be very much of even long-term averages. So I think that's why from the very beginning we did this very coarse based on specific calibration of the model in a way that we just force the model to simulate now it's plus minus 10% of the observed long term annual mean streamflow.
We just forced it so that at least from a ballpark value you can say something about water stress because when you look at other models that are not calibrated, particularly in more semi-arid areas where they have water use and so on. I mean, they can be off by a factor of whatever, five or so, or I mean, maybe that's a lot, but a factor of two, and that still a factor of two makes, I mean, a lot of a difference than when you try to compute a water stress indicator.
And that's why we went for this very simple calibration target, or also process. And we didn't want to do the very normal way hydrologists calibrate. They have a model, they have daily streamflow time series, and then they calibrate 8 or 15 or whatever parameters. Because what we learned from Keith Bevin's equifinality idea that also if we, given that the input, the climate observations or even our model, but climate observations are so uncertain. And then you put all the uncertainty in the parameters. And then anyway, we have complex model that we always felt this is not the right way to go. Right. Just do it and go away.
However, now we have started, a long time ago already, we have a large German research project with geodesists also, where we now do try to calibrate more than say one or two parameters in WaterGAP and try to calibrate against monthly time series. Because we want to calibrate at the same time, not only against stream flow, but also against total water storage anomaly. And then also we started with no cover extent. Right. And you have just other multivariable calibration approaches.
[00:17:41] Bridget Scanlon: Right. So the early approaches to water scarcity index, like Falkenmark and those early works, you mentioned, you have the supply and demand and then to estimate scarcity divide by the resource. And so what do you consider the resource? So Falkenmark's early work considered surface water only. and WaterGAP considers surface water and ground water. And so can you describe a little bit how you evaluate these resources?
[00:18:14] Petra Döll: Okay, so I told you that I come from, I'm a geologist to come from the groundwater.
So in global hydrologic modeling, I always had a special interest in groundwater. Like around the year 2005, I think we wanted to come up with reasonable estimate of groundwater recharge so that we can say something about. Renewable groundwater resources as the long-term average of groundwater recharge.
So groundwater recharge from soil, just that. We are still using an, what we call, heuristic approach, where we say, okay, we compute, let's say, total runoff, and then we know, as hydrologists, we know if the relief is steep, if the soil is not coarse, but it's a clay, And also, of course, if you have a permafrost and so on.
So from these spatial characteristics, physical geographic characteristics, we just model, so to say, a fraction of the total runoff to go into the groundwater and the rest go quickly to fast surface and subsurface flow. And then also what we then saw was, okay. Then, if you do it like that, you would overestimate groundwater recharge in areas with really heavy rain.
A fall like in India, now if you have just a fraction of that temporary constant fraction that would work. So we say, okay, we assume a maximum infiltration rate, so to say. And then I had the chance to get my hands on a few values of chloride mass balance, groundwater recharge estimates, and then we sort of say tuned our model a little bit to that. And so with that, then we can, we know how much more or less we know, we estimate how much groundwater recharge is. And then in our model, like in all other models, I would say almost all global hydrology models. It's just that then we have a base flow out of the system into surface water bodies.
So lakes, wetlands, and rivers, which is just a linear outflow. So the groundwater volume times an outflow coefficient. So that's how we connect.
[00:20:17] Bridget Scanlon:. Right. And since you mentioned earlier, irrigation is the elephant in the room, it's responsible for most of the water use, then how do you partition that water use then to surface water and groundwater? I would think that's a very important part of WaterGAP
[00:20:34] Petra Döll: That is based again on the work of Stefan Siebert, who looked at, so to say, again, data from FAO to say, to estimate for, I think, 16, 000 spatial units worldwide, what is the fraction of groundwater supply versus surface water supply? So we base that totally on his work.
[00:20:56] Bridget Scanlon: If surface water is available, then you would assume that the irrigation is sourced by surface water? No.
[00:21:04] Petra Döll: We have really... Per those areas, the constant fractions are also something that does not change over time. I think this idea, other modelers have this idea that they see as long as surface water is there, then surface water is used. But I don't think that fits reality because, I mean, farmers have groundwater wells and they will use the wells. No, why would you buy a groundwater well and build a groundwater well with a lot of money and then not use it? So the situation, at least in many places, will be like that. I know in the Central Valley, United States, that's not the case. There, people are so well equipped, they can use either one or the other. But we decided against that but went for a constant fraction.
[00:21:46] Bridget Scanlon: How did you come up with the constant fraction?
[00:21:48] Petra Döll: That is based on when you look at the FAO, so to say FAO reports. FAO gets regular reports from local experts, country experts, who are paid to give a current situation of irrigation in Brazil or Argentina or whatever, and they provide such sort of information.
If those irrigation areas are, let's say, mainly from groundwater or mainly from irrigation.
[00:22:19] Bridget Scanlon: Right. I agree with you that most places they cannot afford to maintain both systems like surface water, groundwater, or we saw a transition in India from canal irrigation to groundwater. It's usually people mainly operate just one system and usually cannot afford to maintain both.
So when you model then when you model irrigation water use from groundwater and then you can see over time then you know there's a cone of depression that forms around these wells and then you can capture some surface water and Lenny Konikow suggests from his U. S. work and also globally that 85% of the groundwater that's pumped eventually comes from capture of surface water or reduced evapotranspiration.
And I think, the other, another global model suggests that when they run with and without capture, they get similar estimates of about 85% globally of the groundwater that's pumped is coming from capture. I was just wondering, do you see similar things with WaterGAP?
[00:23:25] Petra Döll: Yeah, we actually have never analyzed global numbers like that.
And I was wondering also, I mean, what do we do here? Because we do not have a model different from that has a capture zone or a cone of depression, because we do not have a gradient based groundwater model. So our groundwater is like a box. It just flows in and out. However, I think that, so what we cannot do is we cannot model if there is a change from groundwater losing to the surface water or back, and that is why we also work on a gradient based groundwater model like PCR GLOBWB has with Inge de Graaf, but it's really difficult to get it to run. So we don't currently, it doesn't work. We have a steady state version by Robert Reinecke, but really to couple with WaterGAP, we're still struggling.
Yeah, but anyway, our model, what happens in our model is, as I said, now we get into the box, we get the groundwater recharge, and then out of the box we take groundwater abstractions, net groundwater abstractions, so groundwater abstraction minus the return flow, and then, of course, that impacts the storage, and though if we have a lot of groundwater abstraction, the storage goes down, and though that's, thus does the outflow to the surface water go down.
So, we do have also a strong impact. that, I mean, the outflow, the groundwater discharge to the surface water, but it decreases as we increase the net extraction from the ground. So we also simulate that, but not in the sense that there is a cone of depression. And also what we cannot simulate is that if evapotranspiration changes, because we have nothing like capillary rise or things like that.
So that is not covered. But actually we had, I mean, I had a paper together with Claudia Herbert on groundwater stress indicators in 2019 and there we proposed three new indicators of groundwater stress and one of them was what is the outflow to surface water bodies like rivers in a naturalized system without groundwater withdrawals and with groundwater withdrawals as a measure for groundwater stress or rather the stress that groundwater abstractions put on the river ecosystems.
So it is modeling but not so physically or cone of depression.
[00:25:50] Bridget Scanlon: Right. So, so that's really nice to point that out. I mean, we need, models to develop projections of the future, but also to test the different hypotheses and to evaluate different processes. And then to see when we synthesize and integrate all of the different inputs and outputs, does it match reality?
And I think that's why modeling is essential for all of those reasons. But also, you mentioned, when you simulate inputs and outputs, then you get change in storage. And you have been working with geodesists in comparing WaterGAP with GRACE satellite data on terrestrial or total water storage, which is from the land surface to the Moho. And so you work closely. And of course, the German space agency developed NASA with. I mean, developed GRACE with NASA. And so that's a great collaboration between the two countries. So from early on, you've been doing these comparisons with GRACE and using GRACE in different ways. Maybe you can describe that a little bit, Petra.
[00:26:54] Petra Döll: Yeah. The only thing that I might correct, it's not the European or German space agency, but it's the GFZ (GeoForschungsZentrum) German Research Centre for Geosciences that is handling GRACE, this organization, so to say. And, but anyway, it was for me, when I learned about GRACE and what it can measure, it was really a big surprise for me because I had never thought that gravity would be meaningful as a geologist.
I knew gravity as a static gravity field, tells you know what happens in the Himalayas where you have the crust and so on. But then all of a sudden, I mean, the geologists were able to measure the monthly changing field. So how the field changes from month to month and over the continent. I mean, I think I never thought about it before.
That is mostly related to the change in water storage. And then for a global modeler, I mean, that was wonderful because I mean, if they have such a big footprint, like 200, 000 square kilometers, or maybe 100, 000 square kilometers, I mean, then everybody always complains about that we have half degree grid scales, 2,500 square kilometers.
That's so big, particularly when you look in Europe, but then when you compare it to the GRACE footprint, I mean, we have a higher resolution and that's why then also that was very interesting for those geodesists in the beginning to get our computation of total water storage anomaly. For example, to check dealiasing, because we have daily changing total water storage, but then they only have monthly things, and then they can, I couldn't do that, but they can evaluate if those daily changes have an impact on their, let's say, signal or their processing.
So I think that was the first contact, so more like, they needed our data to get their processing right or just assess how, what they get from uncertainty. But then very early, we worked with the geodesists who were able to do the processing of these GRACE data. And I think actually since 2005, we worked with them.
And this project that I told you where we try to calibrate WaterGAP, that's also with geodesists who know about GRACE. And so GRACE is really great for global hydrologic modeling, but it's also a difficult data set and observation in inverted commas. And I think it has been a struggle, but I mean, I really like interdisciplinary work and I get a lot of fun out of understanding what other people say because everybody has no own way of putting things and geodesists are very good in math, so they have all these spherical harmonic, blah, blah, blah, that for me is hard to understand and, but it's fun to do that.
But up to now, I would say that GRACE results are very difficult to use because leakage problem, for example, when you ask geodesists about leakage, they throw up the hand. They say, okay, we don't know really what it is and how to estimate it. Not because we are in the situation now where we would like to know, okay, if we take such a piece of land, such a part of a basin, and then we add up all the, I mean, geodesists provide, of course, at a half degree resolution or for mass content.
And so they provide something, but then you look at different solutions and they look quite different. And then none of them really includes the leakage error. When you look at people who, when you talk to geodesists who use spherical harmonics, they say, mass can't forget about it. That's just fake. And then we like, but tell me I'm the hydrologist.
So what should I do? And I can tell you that up to now, I have not talked to any geodesist who told me. Use this to calibrate your model to really know. So, I think it gives you, of course, very interesting insight, but to say, okay, this is the truth, or even this is the truth with the uncertainty bounds they provide.
These uncertainty bounds they provide, they're not the real uncertainties at all. Right. So that they will say, but then how to come up with the real uncertainty bounds of that measurement is there, I would say. And I think too much, like, I mean, like, for example, trends, for example, GRACE people are people with interest in GRACE.
They really often interested in trends. But we also saw in our work with geodesists, how dependent trends are on how you do the filtering, the leakage correction, everything. Trends are so sensitive to the slightest change. I mean, not only in the time series that you use anyway, but also in the way how these things are processed.
Right. And yeah, I think we have to be very careful with being, believing in that sense, those values per se.
[00:31:39] Bridget Scanlon: Right. Well, I think you raised some important points. GRACE, I mean, as Jay Famiglietti said, GRACE gets the Big Picture. And so you're looking at a basin like the Amazon that's six million square kilometers. I think that's the strength of GRACE. And then we have several different processing centers in Germany and the US and other regions, and each of them are using different processing approaches. And then when they get similar results. So we did a comparison several years ago, comparing the output from the global models with GRACE’s data and so on. A lot of variability between the two and then within the models. So I think it's important to, and I know you mentioned earlier, you're looking at snow data, you're looking at GRACE, and you're looking at all different types of data to try to constrain your models. And I think the Intersectoral Impact Model Intercomparison project (ISIMIP) that now compares a lot of different models, that's excellent.
And then now, all of these models calculate the terrestrial total water storage to compare directly with GRACE. And so I think that's very good also. So I think the more we use as many different data sources as we can and do more comparisons, then I think it will increase our confidence and maybe help us estimate uncertainties more.
But I think when people talk about global contribution to sea level rise or things like that, I think GRACE is a stronger tool for that sort of thing than models because it is a big picture.
[00:33:13] Petra Döll: So I totally agree, but maybe I can tell you about the project we had that was on, it was called Closing the Sea Level Budget, financed by the European Space Agency, and led by a geodesist, Martin Horvath from Dresden. And there the goal was that on the one hand, people measure the sea level and so the steric component from the heat and then to close, so to say, the sea level budget in the sense that where does the, what GRACE see over the ocean, where does it come from?
From Antarctica, from Greenland melting, from the glaciers on the land and from the land border, which the rest what we model. For me, very surprising was that when we add up our 60, 000 grid cells where we put in the climate, we have all these information on the soil and so on. And from that we get, and of course water use, and from that we get total water storage anomaly averaged over all land areas except Antarctica and Greenland.
In this, paper by Cáceres et al. (2020), we see that amazingly I was prepared for everything. I thought, well, can be just everything, but that it fits so nicely. So I think it's more like, it depends a lot of your climate input and it just must be an averaging out of errors. But it's really amazing that we have almost no phase shift.
So really every year goes down. We also see 2016 really got, we went down and now we see we go up again. I was so surprised to see that you bring in like this one big, I mean not big, of course GRACE is also, everything, but we have so many steps in between. You model what happens in this soil in the ground, but then the lags, blah, blah, blah. And then also when you average everything up, It amazingly fits and shows trends and the seasonality is fine because, I mean, I thought the seasonality can be totally off. Why should it be the right one? But it is.
[00:35:07] Bridget Scanlon: Right. I think when we did, when we compared seasonality, we found that the amplitude from GRACE was higher, but,
[00:35:13] Petra Döll: Yeah, it is.
But in, in sometimes it seems lower, but when you average over the whole land area, I mean, in your paper, you saw it perfectly that we do not do a good job at the basin scale, but globally it all averages out to very beautiful fit from GRACE with our model.
[00:35:32] Bridget Scanlon: Right. We saw the opposite effect with GRACE and the models, but that was earlier on, so I'm sure they have improved with more of these model intercomparisons and stuff.
So I would like to shift a little bit to your sociohydrology aspects and your participatory methodologies to integrate knowledge and then communicate with stakeholders and try to parlay what you are developing then to the stakeholders to help them with decision making in policies. If you can describe some of that, Petra.
[00:36:04] Petra Döll: So actually I entered that field of research when I had a project in northeastern Brazil. On how impacts of climate change in northeastern Brazil could affect agricultural production, water use, water resources, but also migration. It was a large project of, let's say, 10 German universities and Brazilian universities.
And we came up with two, let's say, integrated models at two different scales for two federal states in the dry northeast of Brazil, and then also for one small county, let's say. And then we put in all our scientific knowledge, including a little bit also of social knowledge and migration and economics and agriculture and so on.
And then we tried to do scenarios of the future and all of that was in a way we were, I was wondering, we put so much work on what we think the reality is from the perspective of us researchers. When you go as a European or a German to Brazil, you notice that really, you do not know what is in people's heads, who then later make the decisions.
No, because they have just a different way of looking at things, how to keep data, what is important and so on. So and there, but on the other hand in Brazil, they had a very good regulation of water resources that they had these basin committees already 20 years ago, where then they decided at the end of the wet season, how much water we will give to what user.
So they had these basin committees. And at that time I thought, well, now that we have come up with these integrated models, maybe we should think about what people have in their heads. So, why put so much work on what we as scientists think is the reality of the system, but we never look at what people really think, what perspective they have, which can be very different from what is in our scientific or my scientific head.
So, well, unfortunately we didn't get that funded, but then later when I then came to Frankfurt, I had the opportunity to work with social ecological people. And didn't want to do hydrology there, but what I started was that based also on the work of a computer scientist from the Netherlands, who had what is called a dynamic actor network modeling thing, that what we did was we started before bringing people together in workshops.
Ask them what is their problem perspective, so what is your goal, what are the factors that impact the goals, what activities of what actors could impact the factors, what are external impacts, and everything you could do at a semi quantitative scale of one to seven, so to say. But actually the semi quantitative thing was not the important one, the important is that from each stakeholder, so representative, let's say, from an NGO, from the company who produces a chemical, from the water administration.
You get, for the problem field of interest, you get such a problem perspective of the stakeholder. And that helps a lot. We have made, I mean, we always do that now. That helps to understand what people know, but also what they do not know. And what, how they relate things. So what is their systems perspective and what are their goals, of course, also. And then in these participatory processes, let's say in the first workshop, people can then, or these stakeholder representatives can present themselves with this map, let's say, or with this, graph. We call them perception graphs. So that to avoid that after you had three, four or five workshops after two years, then one stakeholder says to the other, if I had known that you did not know that, or you thought that, then we could have talked right from the beginning.
No. So that was the beginning of that all that to say, okay. If we want to have a positive impact on, let's say, the development of sustainable solutions, then we have to take it as seriously, what is in people's minds, those who have them to make the decisions or implement some strategies, as it is to come up with a good systems perspective of reality.
And I mean, the natural sciences, but also the social reality, so to say. So these are two different things, like the, what we think scientific view, but then. Bring that together with a stakeholder view. And that was the beginning of finding methods, other methods to bring together stakeholder knowledge of practitioners, non academic practitioners who are in the field, so to say, and scientists who are far away from the specific problem, but have a general knowledge and a systems idea and have the time because they are financed in a project.
[00:40:53] Bridget Scanlon: Oftentimes, of course, most of the time, we neglect the social aspects, but I mean, many times when we're looking at issues like drought or things like that, maybe some of the solutions may not be technical, it may be more social solutions. And so when you have this more comprehensive approach, then incorporating the stakeholders much more integrally into the process, then you're more likely to come up with solutions. And then a co-design, I guess, is one of the terms that they use these days. But if you bring them in early on, then they're part of the entire process. And so they're much more likely to take the results from those analysis and apply them. Maybe you can describe a little bit, I know you've done quite a bit of work on risk management and the hazards and things like that.
Maybe you can describe a little bit on one example of doing this type of thing and some of the methodologies that you felt were successful.
[00:41:51] Petra Döll: Yeah. So when you use the term risk, I think this is a very nice term that was more or less coined in the fifth IPCC assessment report or a little bit before as a very good way to bring together the knowledge of social scientists and the knowledge of the physical scientists, because the hazard is mostly on the side of the physical scientists.
Now, let's say drought, less water than normal, but then the risk comes only if people are vulnerable or to the degree that people are vulnerable. When you talk about risk and risk management, I think for a natural scientist or hydrologist like us, it's a very good idea if you can work together with, let's say, social scientists or sociologists.
And I had the chance in 2016 to do a sabbatical at NCAR in Boulder where I could work with a sociologist there, one of the very few, next to thousands, no, but, of climatologists and hydrologists, with Patricia Romero-Lankao, who also, like me before, was working in the IPCC assessment report, and she's a sociologist, and her focus is on vulnerability.
So, we work together there in my sabbatical to come up with a roadmap, so to say, how to set up participatory risk management processes. Taking into account uncertainty. So, that's also something I find very important, particularly with climate change, of course, but not only to really include uncertainty in decision making in a very, I mean, upfront way.
So embracing uncertainty. That's how we also. So, and the difficult thing is, so to say that people embrace it in a sense that particularly with climate change, people might say, Oh, if we say, Oh, we know so little and everything is so uncertain, well, then we don't do anything. No, that's of course, that's the balance.
You have to do it in the right way. So you have to get people used to, and stakeholders used to embracing uncertainty in the sense that they use information estimates on uncertainty in their decision making, so that is something that in two current projects, we work on how to translate multimodel ensemble output for stakeholders so that they can decide, depending on their how risk averse they are or not on what futures they would like to adapt to, because it's still often shown, particularly, of course, if you show maps, then the median of an ensemble is shown. And then, for example, in Germany, when you look at long term average, let's say, precipitation. Or even water sources, but let's say precipitation in Germany, it will not change much in the future.
People know in the meantime, yes, in the summer, less in the winter, no, but every, anyway, in the median, the models, the ensembles come up with more or less zero change. But then when you look at the range of models, no, then you will have some models who have 20% less and others have 20% more and who can say which is the right one?
I mean, with climate change, I think you cannot say that the median is a more robust estimate than any other model. So in general, with climate models, they say each model has the same likelihood. Nobody really can do, say, this is a better model that will have a more likely thing. So at least climatologists don't say that.
And I think for global hydrologic models, it's the same thing. When we then use climate input into hydrologic models, all hydrologic models, they have different, even if my hydrologic model does a better job for the current condition, it doesn't mean that they can do a better job for the translation of climate change into water.
So anyway, so the uncertainty in these future pathways, I think is important when we look at how to interact with stakeholders. And then the other thing is, as I said, is bringing in the perspectives of the stakeholders.
Now in a project on adaptation to climate change in the water resources in a biosphere reserve in Germany. I also have again the chance to work with sociologists. And this sociologist, she's not into vulnerability. That's not the case. But she is about why people transform or not. And of course, she uses also these methodologies that like focus groups and they do service for the public and so on. So methods that they can do well.
But also theoretically, for example, in that process, in that participatory process, we work from the social scientist side with a cultural theory of perspective. So distinguishing now, I don't know how it's called in English, like you distinguish satirists and hierarchies and egalitarians and libertarians or liberals.
I don't know. So these groups of people and the idea is to come up with strategies with measures that relate to all of them, not fatalist, that doesn't work, but the others, so we say that a multiple view strategy also, a harmonious multi view. So this is what we take from the sociologists in that project and try to include it then in our Bayesian networks.
I mean, that's like, I'm a modeler. So I always model. I like that. And instead of having long lines of text, I rather have a picture of graph with some boxes, that's like, ideally you can work in, let's say small teams. with social scientists for this work. But all the other projects I've done it just with my own group interest, trying to read on more, social ecological things and just, as I say, do it myself and write.
[00:47:38] Bridget Scanlon: Yeah, I think that's extremely important and I think we're becoming more and more aware of it these days. I've been doing some work on public water systems and vulnerability. So, I mean, if you have a water quality issue, that's, you need to have a water quality issue first. That's the hazard. It could be arsenic contamination or things like that. And then you have the vulnerability of the population. Can they deal with the problem? And then you have the exposure to that. So there are many different facets to it, but I think it's important to consider all of these together, then to evaluate how to address the problem and try to provide good quality water for people.
So this risk management is very important. And then communicating uncertainty is huge aspect. And I really enjoyed some of your papers on that issue. And as you mentioned, the mean may not change, but I think another aspect is adapting to extremes. You have seasonal droughts in the summer in Germany, and because you're a humid climate, you may not have much storage, or I know I come from Ireland and they have a humid climate, they're used to rain all the time, and if they have two months without rain, they're in deep doodoo, because they don't have the storage. So, these are all different aspects, so adapting to extremes, or the World Bank saying “too much, too little, and too polluted”, so we have to adapt to these extremes.
So our guest today is Petra Döll, she's a professor of hydrology at Goethe University in Frankfurt and she works in global modeling and also stakeholder engagement and the participatory processes to translate the science to decision making and policies. And we've talked about her WaterGAP model and then her approaches to working with stakeholders.
Thank you so much, Petra, for joining me today and sharing your thoughts on these issues.
[00:49:30] Petra Döll: Yeah, thank you very much for giving me the opportunity to share my views.
