 Thank you, Brad. And thank you everybody for coming here. You know, I've been coming to this meeting since January 2009. I think Jaya had that meeting and then it was the time of the inaugural Presidential inauguration and he paused the meeting to stream President Obama's address. So I remember that first meeting and thank you so much to Greg to Brad and especially to Lynn McCready and the people at the CSG MS integration facility for inviting me here So my talk is on a little bit changing gears here. I'm not going to show you any equations I'm not going to show you any sediment I'm not going to show you anything which is directly with earthscape evolution But I will show you something which is a precursor of all of this and it's the global hydrological cycle extremes floods, droughts, landslides and permafrost So when you think about it hydrological extremes such as floods, droughts, landslides and permafrost are very important scientifically We all like the science, but they're very important. So slightly you see this in the newspaper almost all the time These are environmental extremes which can reshape the landscape right before our eyes We're not talking about millions of years or hundreds of thousands of years The question I deal with all the time is how can we harness satellite observations and publicly available models in order to quantify this and what is the spatial variability and the temporal repeat of such extremes What's the role of land use and land cover change and for the last question I'm trying to pose you an answer as well. What's the future and the future is fusion of data. So Start with a joke What did I tell the sediment? I think I gave it out yesterday Go with the flow So Seriously, when do you have more overland flow? You have more overland flow when we have a wet surface and low infiltration. This is basic hydrology So then coupling this with what I'm talking today You know floods occur when you have wet soils when you have low infiltration large overland flows and high sediment mobilization You have droughts. You have dry soils You hardly have any rain or infiltration. No flow and no sediment. No sediment supply Landslides is a sudden displacement of the soil and it can even cause secondary effects like blockage or flow Landslides can cause lots of second And then permafrost or is Suddenly becoming a very important fact in high mountain Asia parts of Alaska and parts of Siberia Where you have, you know, suddenly you have, you know, soil which can be eroded We're in a golden era We've had so many sensors so many things and we are Measuring many hydrological variables and many hydrological variables by many sensors or long periods of time Obviously the longest one is vegetation and surface temperature, but we also have rainfall soil moisture of a total water water level and the atmospheric variables And all of this fits very nicely to a global hydrological And this cartoon is also in the national academy's decadal survey And so you can see these different sensors Measuring different parts of the hydrological cycle And it's already self-explanatory there So if you look at this and create a nice movie You see that hydrological cycle makes sense. These are four quantities derived from four completely different sources The normalized difference vegetation index from the modus terra The liquid water thickness from the grace Uh precipitation from a final Eimer's run and then surface soil moisture from smos It's a 10-year period. That's why we didn't use map and you can see as it rains Especially in the middle of Africa the itcc goes up and down the vegetation goes up and down the liquid water content Increases with more blue and decreases with more red the soil moisture varies Appropriately and obviously a movie will not solve you all the problems, but it's a good way to start your story So i'm going to focus a bit on soil moisture before I start using it So we have downscaled soil moisture to one kilometer using modus This is a smaps oil moisture downscaled to one kilometer and most of these are published So the publications are listed below, but if you have any questions just reach out to me and i'll be happy to answer so you can see over here that the The one kilometer and the nine kilometer they show differences The higher spatial resolution of one kilometer is probably not as apparent on a global scale But you can see that sometimes the one kilometer has lack of data because of clouds because they're using surface temperature and vegetation index Now on the other hand if you come to the united states It's a little easier to see it because you can see the fact that you know There is more holes in areas if one kilometer The nine kilometer does look a little bit more painted over whereas it just looks like more blown on But if you come to an even smaller catchment and this is a Sacramento San Jose river basin as well as the Danube river basin You can't you can start seeing that the one kilometer starts showing you those pixelations which you don't see in the nine kilometers Now You don't have to take my word for this the global one kilometer soil moisture is validated for over 1500 locations and no other product has them So and it's like this is a public product available to everybody soon to be hosted on NSIDC website So you can see this the nine kilometer does well, but the one kilometer does better And we have shown this in our publication Now we have not stopped at one kilometer. There is a visible infrared imaging radiometer suite called veers on the eco stress mission and we have used this 400 meter Sensor to Downscale to soil moisture to 400 meters now at 400 meters. You're talking real business here Because you're talking about stuff where you can start mapping quarter sections of fields You can actually be there, you know in the field scale It's not yet at 10 meters, but it's good enough And you can see over here that it's very apparent nine kilometer or the san Pedro watershed versus the one kilometer versus the 400 meters and you can see over here the reason this is arranged june 18th june 21 and june 26 is to show this progressive dry down And you can see so nicely that the dry down is captured better by the 400 meters less better but good by One kilometer, but the nine kilometer completely misses the mark It doesn't show too much variability between june 21st to june 26th And statistical comparison with insidious sensor shows that the 400 meter Soil moisture actually performs better with comparison with sites So, you know higher spatial scale is better And this is an example of 401 kilometer and nine kilometer for the whole For the whole united states for conus And this is an example for the washington river basin from august 1 to 24th of 2019 And you can see over here. It's it's been dry anyway So it's not a huge dry down you just saw in san Pedro watershed But you can start seeing higher and higher spatial Resolutions at 400 meters things you don't pick up like this area you picked up Or which is not over here, for example So, you know higher spatial resolution actually helps us to solve a lot of problems Especially when you have questions about propagation of drought. It's important to go to higher spatial resolution So I said I'll talk about floods droughts landslides and permafrost But we have a lots of And many of these are all published over here. So this is a Hurricane florida's induced flooding in south carolina PD watershed 17 to 24 of september be using The gage network as well as two instruments one is the ua v sar flown by jpl 1.8 meters by 0.8 meter spatial resolution 16 kilometer swath elban and the sentinel one which is european seabed instrument with a 5 by 20 meter spatial resolution and a 250 kilometer swath and you can see that we have pretty good Coverage and again, we interspersed them because ua v sar did not fly on that aircraft every day And neither did sentinel overpass this place every day. So and and both of these are active sensors So and a pretty good spatial resolution and the proof of the pudding is in the comparison And you can see over here that for these gauges For different days we get very very good R-squared and low rmsc's and the best part is difference between September 18 and september 24th of 2018 You can see the the difference between the water surface depth or elevation It almost falls on a straight line r square of 0.99. I don't think machine learning could do better And this is another flooding hurricane harrowing in early september late august of 2017 in houston And you know houston is a very urbanized area as opposed to media watershed, which is pretty rural And you can see over here they pick up very nicely these elevations of these Flood depths are very sorry that this font is so small But the pink is greater than three meters which starts from zero to one meters But you can see this these r-squared and the scattered plots with the gauges Show pretty good results and again no massaging no model nothing just pulling the data making sure they're all georeference Now flash forward to a bigger scale and a different sensor modus over australia the wet time was late December 2010 and early january 2011 friday time was 2006. We're doing three measures here the land surface temperature Uh, the change in the land surface temperature from morning to evening is a measure of the Thermal inertia So the difference in the land surface temperature tells you if there's lots of water The land surface temperature will not change between morning to evening because you know water temperature doesn't change much during the day maybe one to two degrees and then the Api antecedent precipitation index more blue means more water and soil moisture from amsiri And this was also published And we actually used data from right this this building from the adjunct flood observatory and connected it to a few flood events in In december and january 2010 2011 and said that hey look We can see the lst decreasing the api increasing and the soil moisture increasing And we even demarcated the regions in the murray darling basin which had flooding so proof of concept And the last one was preparing a modus map at one kilometer or the whole of the lower maycon for 2003 to 2015 to demarcate the flooded areas and this was done with a lot of comparisons and with actives and passive sensors and published, you know, a couple of years ago Now drops There's a big ongoing drought right now the La Plata river basin big one You can see here that the ground water percentile is becoming redder and redder and there's also the The soil moisture percentile is also becoming redder and redder which means it's getting drier and drier And it's it's actually right now going on right now So what we have done is they're taking our snap one kilometer which I showed you and seen that just for the same month of august for 2015 to 21 You can see how it is drying up Spatially at one kilometer And this means that, you know, all those rivers over there. They're probably not going to be flowing not flowing means not This is lower maycon river basin for 2015 2016 They had a drought and a recovery and this shows very nicely the co variability of precipitation soil moisture total water storage for every month and this one is even more Charming so you can see over here the soil moisture keeps decreasing till about april and then starts increasing and it's reflected through the rainfall And the total water storage anomaly so you can see the total water storage normally keeps increasing This is the limpo river basin, which is now right now in a drought You can see over here and we have looked at the soil moisture the And the runoff and the soil moisture and the ets and all of that stuff from gldas as well as the p minus et P minus r and the total soil moisture anomaly all the way from 2002 So the one good thing we can do is we can put this in perspective and compare not just Just look at the flood or look at the drought and compare the two contrast is a very good You know instrument and you see over here This is the lower maycon river basin september 2011 nice wet rain january 2005 Nicely very very dry and that's reflected in the modus et anomaly Reflected in the gldas runoff anomaly and the grace water thickness see all different hydrological variables all coming from different sources But they tell a very consistent story which appeals to understanding the hydrological cycle and so is the case with mario darling basin where you see dry 2009 you just saw mario darling anywhere and then this is wet january 2010 and you can see over here the same things occur Lake victoria river basin and again. This is not published yet. So no pictures, please So you can see the same thing. This is wet october 2019 a dry november 2021 You can see the precipitation anomalies the runoff anomalies the et anomalies total water storage And the runoff again for the another period and then here is that nine and one kilometers And what you can see over here that the one kilometer shows you so much higher Spatial variability of the areas so for local planners for local People who want to do work in these areas local communities. You know, this is a powerful tool landslides So there's a lot of landslides all over the world and this is an inventory for 10 years in lower maycon river basin and Surprisingly, there's a lot of changes and each of these Areas we quantify the change ag to forest ag to urban forest to ag forest to urban, etc And what these then you actually the source of the satellite the source and the number of landslides So we have a huge inventory of landslides So what we've done is a frequency ratio that is the number of landslides occurring in that category divided by the area of that category And what you see here is conversion of agriculture to forest is the biggest contributor All of these zeros says that doesn't matter And all of these positives are important agriculture no change show some landslide forest no change show some landslide But these are the ones which are the most bothersome and the largest And then we also use a regression model and the regression model is an established model and we put in The logistic regression coefficients and their p-values what this tells us is for each category Like land use land cover change coefficients for each of those Categories we see there is a place where there is a biggest contributor for example for a distance to road You can see that there are some places where it's very important at some places where it's not important So these tell you the order of magnitude importance of various factors in landslides permafrost This is a very simple map of the mean annual ground temperature from modus for a 14 year period And it's the mean annual ground temperature varies very widely over the last Over those 14 year periods not only that that it also shows you a very nice shift Because when you start talking about permafrost thought it's important to see that this m a lsd mean average Mean average Mean area land surface temperature. It's moving more to the right Which means that the periods are getting warmer most people's permafrost Zonation is based on climatology. This is based on actual satellite observations And you can see this is what we produced as a permafrost zonation index And which compares well with the past studies, but she shows significant differences. And again, this is being reviewed right now Now All I talked about so far is trying to say, oh, we can do this. We can do that But we have to not just monitor but also use this stuff together And synergistic usage is what is what is one of the most important concepts for my group You take six small watersheds in vietnam with rain gauges and stream gauges And feed one of these watersheds and we have all the results in the paper With a rain gauge driven simulation you calibrate it and validate it And then I merge version six, which is a nasa satellite precipitation product You calibrate it and validate it and lo and behold, you can see the answer so easily the Rain gauge driven simulations fail to capture the peaks of the simulated stream flow Whereas the rainfall from satellite does now the the key for that the message is That the rainfall from satellite cannot be accurate. We don't claim that but it has that spatial continuity Which is lacking in rain gauges, which are points and simply interpolating them is not the best way to go on And lastly soil moisture assimilation, you know, we have a lot of models. We have a lot of Satellite systems and tons of that, you know, you don't have to write another model to do something But you also have outputs on the model. This is the soil moisture output from the NOAA MP 3.6 and this is the soil moisture from the Cygnus and the SMAP sensors put together for a couple of couple of days here And what we have been doing is we have been using the Enhanced column ensemble Kalman filter to to merge the two and what we see and there's a very significant Finding is that there are some places where your estimates improve flux Estimates latent heat flux other other independent quantities Uh, uh, uh using just Cygnus data, uh, or just using SMAP, etc And this is where it's important because you are now feeding a model giving the error characteristics of the model and the observations and seeing What impact can you make for hydrological simulation So the credit goes to all the people who did the work. I mean, I'm just the messenger As I call myself. I'm the hydrological concierge and I don't know to program in python. So Somebody asked me, you know, python. I said, yes, it's a big snake in South America But I'm actually reminded of this great quote I used to walk past it for a few years when I was working at daughter It's difficult to say what is impossible For the dream of yesterday is the hope of today and reality of tomorrow Robert Goddard after whom the Goddard space flight center is named and the inventor of the rocket So final thoughts Observations and model simulations are two sides of the coin which help understand the problem. So you can't say either or Satellite and remotely sense observations can help convey spatial variability. That is of utmost importance for constraining or and or validation of models There is seldom no Alternative to institute measurement. So field hydrology field geology that always have a place Because you know, that's the most direct modification of any variable And hydrological extremes are environmental extremes which have a great impact for landscape evolution Work was funded by nasa and thank you very much Hi, oh, I didn't realize there was a bit of an echo in here. Okay That was a really cool talk. Um, I really like the perspective of using different types of satellite to show differences in hydrologic Regimes and so my question is more towards this is really cool for spatial analysis as we continue to Improve resolution within our satellites that we send up But i'm wondering in terms of temporal analysis. Are we able to use these new and improved? data sets to sort of quantify going into the future how for example landslides or hydrologic extremes may occur Yes, absolutely. The whole point of data simulation is to improve the status of the model right now You know, and there is a lot of other things which go into it and i'm just trying to You know, make a simple answer Absolutely and satellites are not going to stop going up into the sky in the future Actually, it's only going to get better. For example, I have a drone which was built right here in Boulder, Colorado Which gives you five meter soil moisture. So you may not be able to do a global One kilometer or five meter soil moisture, but you can do what I call as remote sensing or demand You know, you have it in a garage you fly it out You know, you get your soil moisture and then it's all you know Vet and you don't have to worry and then two days later starts drying down and fly it right then You don't have to do it and you can do that because I think Somebody told me that they do the same thing with their lidar and see erosion from their fields. Absolutely It's it's fabulous, but it's a local scale study. They're not talking anything about global hydrological cycle But that's where you know Think globally act locally