Water Resources Podcast – Jürgen Kusche
[00:00:22] Bridget Scanlon: I'm pleased to welcome Jürgen Kusche to the podcast. Jürgen is a Professor of Geodesy at the University of Bonn in Germany, and also heads up a working group with about 15 people who focus on applications of satellite data to improve our understanding of Earth and Ocean processes. Jürgen is also affiliated with the Institute of Geodesy and Geoinformation, the Faculty of Agriculture, and the Detect Climate Research Center.
It's really nice that he's affiliated with Faculty of Agriculture because water is strongly linked to agriculture production. Today we are going to focus on his research related to the water cycle, using GRACE satellites. Emphasizing extreme events, especially droughts and floods, and understanding climate and land use drivers on water storage.
So thank you so much Jürgen for joining me today. I really appreciate it.
[00:01:18] Jurgen Kusche: Thank you Bridget, and I'm really glad to be here.
[00:01:22] Bridget Scanlon: Maybe you can start with providing some background to the listeners on how GRACE satellites track water storage changes globally. Although, we've had other podcasts focusing on this, I think it's always good to review it and provide some background to the listeners.
[00:01:37] Jurgen Kusche: I'm glad to provide a basic understanding how this works. We have a lot of satellite missions that are able to track, soil moisture, but that's typically looking at really at the surface. And the new thing about the GRACE mission is that it makes use of gravity.
Whenever we have a redistribution of water at the surface of the earth. And that could be a loss of ground water, it could be water deficit due to drought, but it could be also the flooding of a big area. That means we have an accumulation of mass and it adds to the gravity.
Gravity is the 9.8 meter per second square. But it's not constant. When we go to the top of a mountain, it's a little bit less. But if we have a drought in the area around the mountain, that it'll be even a little bit more or less. Droughts and floodings change the attraction of the gravity field.
And that means that a satellite in orbit, it might be 400 kilometer of altitude, but it will be a little bit less attracted by the earth. It's very small. It's very tiny, but that changes the movement of satellites. And with GRACE, the idea is that we have two satellites which chase each other, so they fly very close to each other, very nearly in the same orbit.
But when they pass a mountain, the first one is attracted by the gravitational attraction of the mountain, and it'll move a little bit faster. So there is a change in the distance between the two satellites. And then the second gets closer to the mountain, and then it catches up. Then we have the distance between the two satellites, gets shorter again. So this is a signature in the distance between two satellites.
Let's assume we have a drought in the area. So it means that the attraction of the mountain will be different the next time when the satellites pass. The next times this evolution of the distance between the two satellites will be a little bit different.
And why are we looking at the distance between two satellites? Because this is something that is, let's say comparably, easy to measure. We have a measurement of the distance with GRACE. We did this with the microwave instruments. It's similar to what surveyors use when a new motorway is built. We count basically the microwave wavelengths between the two satellites, and because these wavelengths in microwave band are very short, this gives us a very precise measurement. This is how the GRACE satellites work. In the meantime, we have GRACE Follow-On mission since 2018, and we have a laser instrument on board.
So it's a laser interferometer, it's a technology demonstrator. And the good thing about the laser is, as you know, laser works in the optical band. The wavelengths are even shorter as compared to a microwave. We call this a link. The wave lengths are even shorter. Assume we can measure the wavelength to a fraction of say 1% because the wavelength is shorter, the accuracy is higher with the laser, and that means that with the GRACE follow-on with the laser instrument, we are probably factor of 10 or 20 better as compared to GRACE.
So this is really great.
[00:04:49] Bridget Scanlon: Thanks for explaining that. I was just talking to Clark Wilson the other day, at the University of Texas, who was involved in the early development of the GRACE satellites. And he emphasized how important the linkage between NASA and DLR in Germany was to making the original GRACE mission a reality.
Because the funding from NASA was insufficient and the German Space Agency was really very valuable in making it happen. So, we are extremely grateful for that collaboration and it's really nice to see it evolve over time and continue.
You were talking about gravitational attraction varying depending on whether it's a dry period or a wet period. And we can also monitor glacier melting or ice sheet mass loss and all of these different things with the gravity. Unlike soil moisture, then the gravity data extend deep into the subsurface so you can get it from the atmosphere to deep in the subsurface. Unlike most satellite data, which are restricted to the top few centimeters of the land surface, GRACE data extend into the deep subsurface.
So these are important distinctions then between the gravity data and other satellite data. This is a really exciting time for the Geodesy and GRACE because, you mentioned we had the original GRACE mission which lasted much longer than we anticipated.
So it launched in 2002 and then the first mission ended in 2017 and a gap of about almost a year, then until the GRACE follow on mission in 2018. That was really nice, that depended on the solar cycle and things like that allowed the first mission to extend for such a long time period.
Now, we're working on the GRACE Continuity mission in that's a projected to launch in December, 2028. And again, a strong collaboration between NASA and DLR. There's a lot of similarities then between GRACE continuity mission, (GRACE-C), and GRACE follow-on.
And you mentioned the laser ranging instrument. In the GRACE follow-on, you had both the microwave and the laser ranging interferometry. But now the new GRACE-C mission will just have the laser ranging interferometry. And so that should improve the accuracy of the measurements. And The European Space Agency (ESA) is also talking about new missions, Next Generation Gravity Mission (NGGM), and MAGIC, which would combine GRACE-C and the European Next Generation Gravity Mission, NGGM.
Maybe you can describe then, how having at least two pairs of satellites would improve the measurements and what that might look like, Jürgen.
[00:07:28] Jurgen Kusche: Yeah, I'll be glad too.
Bridget, as you mentioned, we will have the GRACE-C mission in late 2028 if everything works out. And if we are lucky, we may have even GRACE Follow-On continuing into this time. Which would be good, because then we can, let's say, calibrate our measurements. We didn't have the chance when we had the GRACE mission because there was a nearly a one-year gap.
It would be really nice to look at the same glacier, and look at the same drought, and really compare the measurement systems because they're totally independent of each other. This is the beauty of that.
But you're right, the ESA is planning now for a second mission, which will be called Next Generation Gravity Mission. This is to be launched in 2032, and this means that the GRACE-C NASA mission will be still flying.
When they fly together, this will be called MAGIC, beautiful MAGIC. Yeah. And this will be really like a new generation of measurements. And the beauty of about having two missions at the same time is that we will be able to get much more out of just, having half the error level. There was a lot of collaboration about it and the ESA mission they will not fly exactly like the GRACE-C mission. The idea is that the ESA mission flies in a non-polar orbit.
The satellites fly around the earth, the GRACE Follow-On, GRACE-C, they nearly fly over the poles, and because all the orbits converge at the poles, there's a lot of data at the poles. And the gaps between the orbits at the equator are much larger.
Now, the idea is with the European mission: We don't fly over the poles, we have enough data from GRACE-C, but then when we fly in an inclined orbit, we have more data in the tropical regions. So let's say between plus minus about 70 degree or so. That means there's much more data in the tropical areas.
We can align the orbits to have a much better temporal coverage. Because of these much more data, much better temporal coverage, we will be able to compute gravity fields at a much faster rate. With GRACE and GRACE-C we were able to derive gravity fields at the monthly rate, every data product is a mean monthly gravity field.
We see the change from one month to the next month, but it's difficult to see the effect of for example flash flooding. And there are certain effects that happen much faster. It's difficult. It also leads to a problem in the processing of the data, which we call aliasing. Because it's challenging to computationally remove everything that happens on these very short timescales. With the NGGM- so with the MAGIC mission- we will be able to derive operationally three or maybe five-day data products.
We will be able to map the water storages, the droughts, everything. The effect of floods at a much shorter timescale. And of course, at a higher resolution also because, simply because we have more data. So these are really exciting times we have to say. The overlap period between the two missions, that may be four years, five years, six years, nobody knows exactly because we cannot predict the solar cycle so well.
But this will be an experiment that shows us how good we can be with two missions at the same time.
[00:10:53] Bridget Scanlon: Yeah, I think that's going to be fantastic.
And for the listener's, MAGIC is an acronym that stands for Mass Change and Geosciences International Constellation. And that is GRACE C, GRACE Continuity, and the European Space Agency, ESA's, NGGM, Next Generation Gravity Mission together.
You mentioned having both missions then and then the different pathways. So the GRACE-C would be a polar orbit and the next generation gravity mission may be an inclination of about 70 degrees. So you'll get improved data then, and in the tropics near the equator up to plus or minus 70 degrees.
That would be very valuable.
You also mentioned the improved accuracy with the laser ranging interferometry. And I think, when we talked before, you mentioned maybe 10 to 20 times higher accuracy, with that relative to the previous microwave system for monitoring the distance between the satellites.
You indicated that maybe you could get increased temporal resolution. You guys generally provide the data for monthly timescales because when you aggregate over a month, it gives you maybe plus or minus two centimeters on certainty or that's the goal.
It takes about five days, to complete a cycle of the satellites orbiting the earth to get a full global coverage. And Univ. Texas Center for Space Research (Himanshu Save) did a five day solution for us, and last couple of years to look at flooding. The one monthly solution there is also, in addition to, having a long time period and masking high extreme events, you also have a long latency.
The data provided like 40 to 60 days after the events, after a latency of that time. With all of this, you would allow to increase the temporal resolution and reduce the latency. Then hopefully help to address extreme events like flooding, particularly droughts, maybe have a longer duration.
But flooding in particular.
[00:12:56] Jurgen Kusche: Exactly. Exactly. These are two good points that you mentioned. There are already five daily solutions and people also tried three daily, or daily solutions, or, rolling averages. But there is a tradeoff.
This is what we found out over the years.
So, If you focus in a shorter time period compared to the one month, it means that we may have a global coverage, but we don't have as much data as we would have if we would wait one month, collect all the data. That leads to the problem that- let's say at the large scale, looking at the continental averages -it's still possible to derive for three-five day solution, it's possible to derive a good, measurement. But if you are really interested in smaller hydrological basins, like the Rhine Basin in Germany, then it becomes very challenging. This is the smaller the areas, which we are interested in, we need to wait longer.
It also means if you are only interested in the trends, comparing to climate models for example, this is something that we can do at a good high resolution. But currently, if you're interested really in very high temporal resolution, then we have to be satisfied with, the very big basins like the Amazon, the Congo Basin, the continental averages.
With more data from having two missions, we can improve. But there will be always a tradeoff. But with more data and in particular, both orbits will be aligned to each other, optimized so that we are pretty confident we can do much smaller basins as compared to now with the higher temporal resolution.
[00:14:31] Bridget Scanlon: And what people typically quote is the current resolution with the GRACE Follow-On is a basin of about 300 kilometers or a hundred thousand square kilometer basin area for maybe the two centimeter uncertainty. David Wiese, discussed the overlap between GRACE Follow-On and GRACE-C. You could reduce that to about a base area of about 40 to 50 thousand square kilometers with the same uncertainty.
It's important to consider the tradeoff between space and time. If you want high temporal resolution, then the uncertainty is large and you need to increase the area that you're looking at. And if you want high spatial resolution, you need to aggregate over a longer time period.
You have described this trade off in your papers, and also David Wiese has described it in his. Brian Loomis and his group have been looking at high spatial resolution and aggregating over much longer time years beyond the monthly timescale, to get down to maybe 10, 20 square kilometers that sort of thing.
So, there's tradeoffs in those things, and that's really important to understand.
[00:15:41] Jurgen Kusche: Yes, that's true. It depends on the application. You can tailor the way how you analyze the data to your application if you're interested in the longer timescales. Like you mentioned, the droughts, or let's say the depletion of water resources, then you can really get this higher spatial resolution.
If you're interested in the short time scales, you will have to live with a little bit worse spatialresolution. As I said, with two missions at the same time, this will be even better than today for the monthly solutions, but it'll not be the same. The only way out is if you go back to the original along track data, this is a very challenging approach.
But what we typically do is we compute a model of the gravity field every 30 days or maybe every five days. But then we try to compute a complete model for the world.
If you are interested in a specific hydrological basin, in a specific drought event, or flooding event, you can try to look at the data really as it is collected along the satellite, and then the next time the satellite flies over this region.
You can do a dedicated sort of local inversion and some people have shown that we'll be able to look at these kind of events with an even increased resolution. The price that you have to pay is you need to know where you have to look at.
You cannot just look at all the data. You need to know, where you have to look at. You can exploit the original data much better because you're not interested in mapping the whole world.
[00:17:13] Bridget Scanlon: That's very interesting.
If we focus in on extreme events and flooding, some of your papers described the flooding that occurred in the Ahr Valley in July, 2021. That was extreme high intensity flooding, I think about 190 people died during that.
In your paper you described that the flood hazard maps were outdated and people did not have a memory of flooding because the previous large floodings occurred in 1910 and 1804. There was no memory of that. Communication was an issue. And then of course It happened at night- And it reminds me of the flooding that we had here in Texas in July this year. Happening overnight, happening during a holiday and all of these things. It was interesting to read, some general aspects about the flooding. A lot of people that died were in basements and stuff, and the pressure of the flood water couldn't even get out and all of these different things.
Your work then, you talk about how you could possibly use the GRACE data. And If it passed over that region, what you could detect and how it could help. And also, we always focus on forecasting floods and predicting them and that sort of thing. But also the recovery after the flood and what happens to the storage I think is also an important aspect.
And if we have some detailed GRACE data, we can see how that evolved. And my colleague looked at that with hurricanes and stuff, and not just looking at the beginning of the flooding, but also the recovery afterwards. And with the higher temporal resolution, you can get a lot more detail.
So maybe you can describe your work there.
[00:18:52] Jurgen Kusche: I should first describe a little bit this flooding event because this is geographically very close. The area is 30 kilometers south of Bonn. There are lots of employees of our university living in this place. Commuters. I also have relatives there.
It's very close and we haven't had this kind of flooding in Germany since the sixties. In the sixties (1962) there was a flooding at the North Sea coast in Hamburg, which there were a lot of fatalities, but we haven't had this sort of flooding in Germany since that time, and this clearly showed that we were not, that we were not really prepared for that.
It's an area within a very narrow valley. The people, they live from tourism, they make wine. Many people live very close to the river. The river is very small and typically, children can play in the water, it's a very narrow river, but then we had these extreme convective event.
So this was really a massive downpour of water over three days. Record breaking rainfall. And it was predicted by the weather services, but the valley is so narrow. The weather people, they cannot predict the occurrence of the rainfall within an accuracy of hundreds of meters. Yeah, they can predict the rainfall, but whether the flooding occurs in this valley or the next valley really depends on where exactly the rainfall is because it's not the plain. It's very narrow. So yeah this happened four years ago.
It was a devastating event. And then later on we started to look into whether we could learn something with the GRACE Follow On satellite data about this event. The first thing that we discovered was that the satellites didn't fly over this event. We did lot of sensitivity computations for our paper, and we are pretty sure if the satellites would've flying over the events they would've seen the flooding.
The flooding was big at the scale of Germany. There are bigger floodings in the world, but probably the satellites would've seen this one. The next interesting thing that we then started to understand is when we looked at precipitation and at different precipitation products, and they all predicted the convective event over about the region, but the amount of precipitation in the different forecasts was very different. We have a very good radar network for the meteorological forecast of Germany. We have high resolution weather predictions by the weather service, we have the ECMWF products. So we have a lot of products, we looked at all of them and the amount of precipitation that was predicted, sometimes there were huge differences for particular hours, like 100%. Yeah, so huge differences. So it means that if you use a hydrological model to understand what happens, and a lot of colleagues here use hydrological models, the hydrological models depend on the precipitation. They depend on the input data. And if the input data are so widely different then the hydrological simulations will be different. And we looked at what could we have seen with the GRACE satellites and we discovered that basically, because this is all a movement of mass, that's water vapor and it's accumulating in a certain region, then it turns into liquid water. When the water vapor starts to condense still in the cloud, then somehow the precipitation starts. That's all the movement of water. It's a history of movement and water, and at some time the precipitation hits the surface, then the flooding develops.
And the satellites would have seen this accumulation, even before the precipitation event occurs, simply because the water mass converges in the atmosphere. Part of that is typically removed in the GRACE data product because people remove what is called the aliasing product because we are normally not interested in the atmospheric mass.
Normally, we are interested in the droughts or the floods or the glaciers. But part of that is also left in the atmospheric, de-aliasing products. So what we believe is that we can see this let's say accumulation of water for several of the big precipitation events. Not only the one over Germany. I have a PhD student who did some statistics and he found that there are actually like 1000 or so of these events that could be seen every year already now in the GRACE Follow-On data. Which doesn't mean that we can really study it, but it's visible in the data. And because of climate change, because the atmosphere is warming, you know, the water holding capacity of the atmosphere is also increasing with temperature. Also, he, we found that the number of these events big enough to be seen in GRACE Follow On is seems also to increase. And this is interesting. Of course, many of the very big convective storms occur over the ocean, far away from the land, but some of them also over the land.
But this is very interesting development.
[00:24:21] Bridget Scanlon: Yeah, I was fascinated when I read those papers and the same issue everywhere. I guess, the meteorological forecast, they can predict possibly how much rain would occur. But figuring out where it will fall is more uncertain for them. And so that has a huge impact then on the hydrology and where the flooding is going to occur.
And we had the same issue here in Texas. So you described the Ahr Valley being a very narrow, steep topography and responds very rapidly with lot of runoff, but predicting ahead of time if it's going to occur in the Ahr Valley or if it's going to occur in another valley is difficult for the meteorological forecasting.
But then the value of GRACE, them showing how much water mass in total is in the atmosphere is also a very valuable parameter that we really haven't looked at in the past. We generally remove the atmospheric signal and focus on the land water storage impacts, floods, droughts, or, melt water or whatever. And so, you indicated in your paper then if the water mass exceeds a certain mass, like 0.6 km3 then you can detect it with the GRACE with the laser ranging system.
And then you detect about a thousand of these globally, with many of them in the oceans. But this could be a very valuable data set in the future then for predicting floods and understanding, the linkage between atmospheric water movement and flooding. I think that will be huge contribution.
[00:25:58] Jurgen Kusche: Exactly. I don't think that we should really talk about predicting floods, but this would be helpful to improve the meteorological models. Basically the coupled models that describe the water cycle, including the atmosphere. Because the atmospheric convergence is of course very much depending also on the response of the land surface.
Evapotranspiration depends on the biosphere, the land surface. So I think, in the future, these data, the GRACE data, the MAGIC data they can be used to improve these coupled models. And in that sense also the forecasts, because the same sort of models are used at meteorological services for the forecast.
Once we improve these models step by step, we will also help them to improve their predictive skills.
[00:26:48] Bridget Scanlon: Also as I mentioned, we generally remove the atmospheric signals. You working on this aspect indicates that maybe there, we need to improve that atmospheric signal in order to correct the GRACE data for the land storage. Maybe you can talk a little bit about that Jürgen?
[00:27:05] Jurgen Kusche: Yeah, I mentioned in the beginning that the beauty of the GRACE and GRACE Follow On missions is that it observes all the mass changes. So it's not restricted to the soil mass and the top layer. It observes changes in the root zone, it observes changes in the aquifers,
it observes changes in the glaciers, the snow pack and so on. This is the beauty. At the same time, it's also, in a way, it's a curse because it observes also the atmospheric mass density, the dry air basically, which is not the signal of interest, and it's also the water vapor, which is highly reliable.
Which could be maybe in the future a signal of interest. But typically, let's say we are interested in droughts. We are not interested in the atmospheric dry air. The same happens also for the ocean. We know that the ocean reacts quite quickly to atmospheric pressure changes.
And very often we are not interested in the ocean, like you're interested in a river basin. But if you have a large signal in the ocean, that might somehow disturb what we try to observe. So what we do is we remove the signals that are created by the atmospheric gravitational attraction, or by the ocean gravitational attraction simply by the sloshing around of the ocean. We remove this from the data, the ranging data between the satellite, and then we analyze these data as if there would be no atmosphere, no ocean. This gives us then the water deficit during a drought. But of course, the models we use here cannot be perfect.
They cannot be perfect and therefore we make errors. And these tiny errors, some of them are averaged out when we collect all the data for one month, today, but some of them are not averaged out. We see that some of the atmospheric moisture convergence, in particular the cloud water and the atmosphere, that has a small effect.
And it's very tiny, but it can be seen if you remove it correctly then this helps us to isolate the other signals a little bit better.
[00:29:14] Bridget Scanlon: Of course, everything is an advantage and a disadvantage at the same time. So you mentioned, we get the terrestrial water storage, which is the atmosphere to the moho but oftentimes we're interested in a component of that. But I think we are recognizing and acknowledging the value of terrestrial water storage.
And I know you were talking about removing the atmosphere and the ocean, but terrestrial water storage includes all of the components, surface, reservoir, snow, soil moisture, and groundwater. And I think that's a valuable data set. And I think the climate observing system now regards as an essential climate variable.
And so that's really nice because oftentimes in the past, hydrologists would immediately jump to where there's groundwater or which component. But just looking at the total water storage itself I think is very valuable for looking at droughts and floods and things before we try to determine which component it is by subtracting other components and introducing more uncertainty. The GRACE data is oftentimes used for looking at groundwater, and in Germany you have used it. There's the G3P program that you use there. And also, you've experienced intense droughts in 2018, 2019. And then, most of the time, I come from Ireland originally, and people think water scarcity? It never stops raining
[00:30:39] Jurgen Kusche: Yeah.
[00:30:39] Bridget Scanlon: How can you have water scarcity? But it's the management. So Europe has experienced intense droughts in many recent years. And so in Germany you were looking at the impact of those droughts on groundwater storage.
And maybe you can describe that a little bit. And I think Andreas Guntner was involved in some of that work.
[00:31:03] Jurgen Kusche: Yeah, the issue of the water resources in Germany, this is something that also very much came to our mind here. As you mentioned, we had these very strong droughts, over the recent past, there were some strong droughts also in the beginning of the 2000 years, so people got concerned.
We don't have so much of irrigation here currently in Germany, but we need, of course, we need water for industry purposes for. Also , we have the Rhine, the big river in Germany. This is a sort of motorway of the rivers.
And it's a big transport system. So we have a lot of ships and, in particular in the very recent years since Germany does not use gas anymore from Russia. We have a number of power plants, which are still working with coal, and all this is transported along the river.
And during the big droughts, the water level in the river fell so low that the shipping was really affected. So the big vessels couldn't use the load as usually. They could transport one third of the load, which was a very big restriction during that time.
So really we experienced problems with the water resources during the drought during the last couple of years. It means that people ask about what, what's going on. And of course, politicians and water management ask about what's going on.
We have very good drought monitoring and let's say, soil moisture monitoring systems in Germany with sensors, but they cannot look really, at the deep layers in the soil. And of course we also have a monitoring of the aquifers, of the groundwater aquifers.
But of course people looked also at the GRACE data because the GRACE data tells us everything. And so people came up with numbers that were astonishingly big. So basically that would mean that over the last 20 years, which is the time since which we have these gravity measurements, the full amount of the Lake Constance which is the biggest lake in Germany, would've been gone. Not from the lake, but as a volume over Germany.
Which is a big number, of course used here just to visualize the amount. But we looked at the GRACE data and it's still uncertain, because Germany is, on the global scale, it's not a big country. We often believe it's an important country, but it's not big. So we found out if you look at different GRACE data products, there are some differences.
And also 20 years sounds like a lot, but on a climate time scale it's not very long. So we found out that if you add one or two years, in particular because we had these strong events at the beginning of the end of the time series, so if you add one or two years or three years to the time series, the trends changes.
And that introduces some uncertainty. And we should be, let's say, at least cautious by interpolating these trends into the future. So yes, together with Andreas Guntner and a few others, we looked at the trends in Germany and in fact, yes, Germany is losing water. The volume of water storage is not a constant, but it is losing water.
I wouldn't say it's an alarming rate, but Germany is losing water. In Europe there are other regions like Spain, particular regions around the Mediterranean that experienced much bigger loss of water during the last 20 years or so.
[00:34:32] Bridget Scanlon: So, yeah, I think that was very interesting and points out some of the estimates of the decline, so you mentioned Lake Constance and I think that's about capacity, about 48 cubic kilometers, and over the 20 years. So average some of the estimates were about two, 2.4 cubic kilometers per year from GRACE data declines in storage. And then over the 20 years, then that's about equivalent to Lake Constance and for people in the US 1 million acre feet is oftentimes the unit they use, and that's equal to 1.2 cubic kilometers. So the units are similar.
But we oftentimes just average and develop trends, but we sometimes ignore the ups and downs in between, and then the time period that you select, depending on what's happening at the beginning and the end of the time period, you might, that could really impact the trend that you estimate.
And very important not to project that on for everything So important to use a lot of different analysis of GRACE data to, get some idea of the uncertainty and also, Germany is next to the Alps, and so there's glacier melting there. And so that can leak into your signal in Germany.
And so that could amplify the declines that you're seeing. So I know in your analysis you took all of these things into consideration, and so really the decline was much less than what the news media were reporting from limited data analysis. So, very important to be careful about these things and then try to understand and put things in context.
[00:36:07] Jurgen Kusche: Yes. And as I mentioned, we try to, let's say, improve the simulation models nowadays. And we use techniques like data assimilation where we really try to integrate the simulation of the water cycle and in the soil also in the atmosphere and the ground water. We try to integrate these simulations with data from the GRACE follow on mission.
It means that we really need to be cautious to, not, for instance, to mix up these signals from the Alps, from the melting glaciers that are not represented in a hydrological model, and not to introduce these signals through the GRACE data by not doing the right corrections into a modeling framework.
We want to improve the hydrological model, so we really need to be sure that the data that we use for that is really clean from everything that's not hydrological, i.e. redistribution of water that can be matched to what the model simulates.
[00:37:08] Bridget Scanlon: We've talked a lot about climate forcing impacting water storage. But some of your work also is looking at large scale and seeing how GRACE could help you figure out where globally regions are wetting or drying, by looking at precipitation minus evapotranspiration, deficit or surplus.
And I think this was work that was described in the paper with Jensen in 2024. Maybe you can describe this and some of the causes of this wetting and drying in different regions and relationship to land use and other forcing.
[00:37:43] Jurgen Kusche: Yeah. So then, what people typically call the wetting and the drying of the land surface is basically if there's more precipitation in the long term as compared to the evapotranspiration in the same region, then that typically means that there's an accumulation of water. It might be matched by the discharge through the rivers, but it may be not.
And if it's not, then it leads to an accumulation in the same region. On the other hand, if there is a deficit of precipitation, if you compare it to evapotranspiration , it appears as a drying trend.
So we tried to look at the models that people use for the CMIP experiments, Coupled Model Intercomparison Project. So this is basically the models that people use in the IPCC framework and we try to match this to what GRACE and GRACE Follow On and observe.
And we found that the IPCC models are not doing a great job in, let's say, simulating a drying and wetting as compared to GRACE. And we talked also to the meteorologists. And we believe that the reason for that is biases in the precipitation, in the simulation of the precipitation in the models, because these are models and they don't use precipitation as a radar satellite observes. They have to to simulate precipitation, which is a very tricky business. So we see these differences as compared to, let's say, the tools that GRACE provides. But on the other hand the IPCC type models are, they're still very, coarse in the spatial resolution.
And we know that there's, that are processes at the land surface that also lead to regional changes in the let's say in the regional climate, in the regional water cycle. Yes. We know that globally, it's the radiative forcing that leads to global warming. And that of course affects also the atmosphere and the circulation systems in the atmosphere.
And, the blocking effects and the heat extremes. But there is also a lot of land use change. For instance, take Europe, Europe is a continent where we have land use change since hundreds of years. We have also a lot of agriculture like as you also have in the US and agriculture means that there's technology evolving.
So the way how people use the land has changed over time. And it also means that the evaporative fluxes and the atmosphere change, and that the runoff changes. I'll give you one example. We had in the nineties the iron curtain came down and in the past on the east side of Europe, we had very large collective agricultural frameworks while on the western side of the Iron Curtain, we had much smaller farms, much more individual farmers, less organization, and that all changed a lot. And for instance, they used a lot of irrigation also in the past, in Eastern Germany, but also Eastern Europe.
But a lot of that broke down and was not used anymore. After the fall of the Iron Curtain, it took them a lot of reorganization time. So there was, this is like a large, let's say, land use experiment. And we also know that, for instance, in the northern countries, there's a regrowing of vegetation. So there's a lot of land use change ongoing.
People often think about deforestation in the Amazon, which is a very large effect, or regrowing forests for carbon sequestration, which we maybe cannot do because we don't have the space for all the trees, but there's, a lot of going on, let's say on the smaller scale. People use in agriculture, there's using more what's called agroforestry. Agroforestry means you have a lot of trees, rows of trees for many different purposes for increasing biodiversity, for instance. So there's a lot of change going on. This all affects also the fluxes across the land surface. We try to understand the contribution of these local effects, or for example when you start to irrigate land.
This is water that cools the land surface. We know that already this is extra water, which goes to the atmosphere. It evaporates, it's carried away by the current systems in the atmosphere and it may even end up in a different country. There were simulations that show that if you start irrigation, if you would do more irrigation in Spain, the water ends up in, in France or in other parts of Europe, so it can have large scale effects in the way of anthropogenic telecoupling. This is what we try to understand and this means also that the drying and the wetting patterns are changed, and we hope to see this with the GRACE data, with the MAGIC mission at the higher resolution, and be able to constrain the models to better understand these local effects.
[00:42:37] Bridget Scanlon: Yeah. I looked at land use change a long time in the past. When I started working in hydrology, I felt irrigation is the elephant in the room and we have a lot of irrigation in Texas. And I looked a lot at that. But also seeing that when you first start to cultivate an area, I think what you're talking about Jürgen, is the partitioning of water at the land surface and how it's impacted by the land surface.
At the land surface, then precipitation hits the land surface and then they can, some of it can evaporate, some of it can run off, and some of it can infiltrate. And depending what you're doing at the land surface then can impact that partitioning. And so when you cultivate land, oftentimes, we see in West Africa when you change native bush deep rooted vegetation to shallow rooted crops
with fallow periods, you increase the infiltration and the recharge. And you increase the water storage. So these sorts of things, and we've seen them in the US in the Southwest, in the early 1900's, and in Australia also. So, globally in a lot of regions when we changed the land use and of course in Europe it occurred earlier.
But it's a very interesting to think about what happened when the Iron Curtain fell, and you have those huge collective farms in the east and more irrigation and then smaller farms in the west, and what impact that has on the water cycle and I guess you are also using models then and constraining it with the GRACE data.
So that's integration of modeling and GRACE data I think is very powerful. And I think you've been doing a lot of data assimilation with the ParFlow model and GRACE data and WaterGAP and some of these other models to better understand. And you mentioned comparing with CMIP models and so maybe the, I think in your paper you indicate that in the Amazon the CMIP models are projecting wetting, but the GRACE satellites are suggesting drying, or did I get that the wrong way around?
But the Mediterranean region, seems to be consistently drying in the models and in the GRACE data.
[00:44:42] Jurgen Kusche: The Mediterranean region is nearly the only region on the world where the models themselves are very consistent with each other, but also with the GRACE satellites. But if you look at other regions, in West Africa or Central Africa is an example, the models are very off compared to themselves, but also very off compared to the GRACE data.
Yeah. As I said, we use these high resolution models. And you mentioned ParFlow, that's a physical groundwater model, which is typically not used in the CMIP context. So this also means that we hope by involving processes that typically people at the, big global CMIP context cannot include.
We hope to be better and to get closer to the reality. And by data assimilation, the idea is that, integrating the physics of the models with the observations we get the best out of everything. In a way that helps also to break down the total water storage estimate from the satellite missions into the different compartment in the statistically optimal way. It is not the silver bullet, it doesn't solve all the problems. In the end, it's a data set because it assimilates data, GRACA data, but also data from other satellites. And we also try to assimilate land surface temperature and other kinds of data sets. In the end, an assimilation is a sort of a data set that has been interpolated in an intelligent way by the models.
And this is what I compare to modelling and the different scenarios. Land use change and maybe future land use change, and future irrigation scenarios. This is where we are working on at the moment.
[00:46:18] Bridget Scanlon: Thank you so much, Jürgen. Our guest today is Jürgen Kusche. Jürgen is a Professor at the University of Bonn and working on applications of satellite data to the water cycle, and a very interesting discussion on how we can use satellite data, particularly GRACE data, for flooding and for droughts and globally for wetting and drying regions, identifying them and combining them with models to test the models and to improve the models so we have a better understanding of the water cycle. Really appreciate your work, Jürgen and really grateful that NASA and Germany collaborate closely on the satellite data. It wouldn't have been possible without that collaboration, so we are very grateful for that.
Thank you.
[00:47:04] Jurgen Kusche: It was a pleasure. Thanks for having me, Bridget.
