 Good morning, everyone, and a warm welcome to all those who are present here in the Armstrong Atrium, and to all those who are viewing this through our live stream channel. To this, our fifth Purdue Engineering Distinguished Lecture of this academic year. My name is Arvind Raman, I'm the Executive Associate Dean here in the College of Engineering. Now this series began in 2018, really as a way to bring some of the world-renowned thought leaders in their respective disciplines to Purdue Engineering to engage in meaningful conversations and thought-provoking discussions with their students and faculty on the grand challenges of the time in the discipline and also the opportunities. And our speakers, when they come here to the visit for about one and a half days, they engage in both a lecture, which is what we're going to have right now, as well as in a panel that's going to happen tomorrow. So please make note of the panel event tomorrow as well, if you can attend it. And in the process of doing so, you know, they are able to really bring our entire community here in Purdue Engineering who's interested and engaged in that particular area across disciplines to really have meaningful conversations on the topic. Today's lecture by Professor Anna Barros is co-hosted by the Loyal School of Civil Engineering. So I'd like to call upon Rao Escovindaraju, the Bowen head of the Loyal School of Civil Engineering to introduce her speaker today. All right. Thank you, Arvind. And again, good morning to all of you. It's my distinct pleasure to introduce Dr. Anna Barros. She's the Donal Begar Village Chair of Engineering and Department Head of Civil and Environmental Engineering at the University of Illinois, the Granger College of Engineering. Her primary research interests are in hydrology, which also happens to be my area. She also does hydro-metrology, environmental physics, is interested in water cycle processes in regions of complex terrain, remote sensing of the environment, predictability and risk assessment of extreme events. All very, very relevant topics. She has served on multiple national committees over the years, including Space Studies Board of the National Research Council, Water Science and Technology Board, Board of Atmospheric Sciences and Climate, and the U.S. National Committee for the International Hydrology Program of UNESCO. She was a senior fellow at the Energy and Climate Partnership of the Americas and a founding member of the ASC Committee on Climate Change and Adaptation. Dr. Barros is the past Chair of Atmospheric and Hydro-Sepheric Sciences at AAAS and President-Elect of the Hydrology Section of AGU. She was the Chief Editor of the Journal of Hydro-Metrology for five years, member of the Editorial Board of AGU Advances. She is a fellow of AGU, AMS, ASC and AAAS, senior member of IEEE and a member of the National Academy of Engineering. Anna. Good morning, everyone. I can see that my first slide is here. Thank you so much for the very kind introduction. It's a real pleasure to be here today, especially because we emphasize so much interdisciplinary and multidisciplinary, and that's really what my career has been about, working at the interface of disciplines. I work a lot in mountainous regions, mountains, some tall, some small, and I also do a lot of remote sensing work. The focus of the talk is going to be on processes, on precipitation processes in mountainous regions, and what we have learned through observations from space, observations on the ground, and modeling to help us understand really freshwater resources, which is what precipitation is all about. Before I start, just as an introduction, I wanted to show the classic textbook definition of autographic precipitation. What happens is when you have an obstacle like a mountain, and you have moist air that comes against the mountain, the air is lifted, is forced up, and as it goes up, it gets to cooler temperatures and eventually condensation starts. Of course, in most of these drawings, in books, in textbooks, and this one is from Encyclopedia Britannica, you don't talk about aerosols, for example, but actually you cannot form stable clouds without, of course, having aerosols. Then on the other side of the mountain, as the air comes down on the other side, of course, it warms up, it expands, and so usually, not so usually actually, but in the textbook representation, there is a rain shadow on the lee side of the mountain because all the moisture evaporates again if it was still there, if it had not been removed. That's basically the textbook definition of autographic precipitation. Then next, I wanted to introduce to you the basis for the colors of water. Actually, if we talked with some faculty earlier, and we think always about gray water, right, or brown water, and so on from the point of view of water quality, but the colors of water that I want to talk about are related to policy. In 1985, Malin Falkenmark, who was working at the UN, really came up with this framework. It was a paper published in 1985, which actually has very, very few citations despite having had huge impact. In this paper, she formulated the problem of water resources as having two components. One component being blue water, which is the water that flows in rivers that's available from aquifers. The other one would be green water, which is the water that is needed and must be available for vegetation. That was the idea of the consumptive water used by vegetation. One of the citations that I like to use from her work is that a land use decision is a water use decision. I think if we move forward now and we think about all the work that we have done in the last 20 years, and some of that done by faculty here at Purdue, focusing on land-atmosphere interactions, we actually know now that vegetation plays a very important role in the formation of precipitation itself. I think about that as adding the white water or the clear water to the cycle. That citation actually needs to be changed or improved in some way. It's not about just use. It's that actually land use and land cover have a huge impact on water resources as they are. It's really a more general principle that should govern us. This is especially important in civil engineering and when we look at regional planning and so on because of the big decisions that we make in terms of infrastructure systems and water systems and so on. Going back to talk about mountains. Another aspect of the presence of mountains that I wanted to emphasize is how they set these huge continental scale boundaries between wet and dry. Between the wet side and the rain shadow side. You can see that there along the western US and for example in the Himalayas. Another type of boundary that's also very important in tall mountains is where the tree line is. We have limitations in how much moisture you have at higher elevations and how much condensation you can actually can make happen at higher elevations. As above 4,000 meters or so we start having much lower vegetation because of temperature but also because of decreases in moisture availability. The tree line is another important boundary that is associated with big mountain ranges that we must consider. So looking at this issue of spatial organization and the important role of satellites and observations from space in helping us look at earth as a whole, right? As the entire planet. So we can look at very large scale features. This is a picture of the cloud cover from the international space station over the Andes and you can see on the, I'm not sure if I can point here but on the side of the Amazon you see all the clouds basically protruding out and then on the Pacific side you see the very dry rain shadow actually on the ocean side and that's also because that is a very cold current in the ocean along the coast of Peru which basically contributes to this dryness. But so if we look at mountains around the world what you'll see is that even though whether it's very hard to predict and clouds seem to be very hard to predict when we do something like a principal component analysis for example of cloud fields over one or two decades what we find is that the first principal component actually explains 70% of the variability of clouds over the Himalayas over the Andes Western over the Western Rockies and that's just because the topography is actually controlling that condensation. Now clearly what happens along this along with cloudiness is that precipitation is also associated with this and so you can see in this in these two figures here what I am showing are precipitation features identified from the dream satellite at night and day over the central Andes and what I want to point out if I can I don't know how to point with this actually I should have asked but I'd like you to look at that area of white by the by the bottom right of that of that plot and so this is interesting right because we're looking at 15 years of data and you can see that there are areas there where it was never detected precipitation was never detected over 15 years every time the satellite passed and that's actually where the tree line is so you can clearly see these features from the large-scale observations from satellites thank you so much I think that this will will help me yes maybe and so and so on during daytime because of solar of solar forcing we have a little bit more evidence of a very light a very light rainfall other but because of this of the spatial distribution of these precipitation features we can now go from these observations and actually start looking at what the landscape looks like right and look at at continental-scale landform evolution patterns and so what you see here in the first in the first plot is a distribution of those precipitation features I showed you before in this in that in that little yellow box in the Andes and you see two peaks right we see a very large peak at low elevations and that's the the convection in the Amazon and you see another peak at at higher elevations which is just before we heat the tree line right so we have another another increase in the in the number of features but what's really fascinating from this is that when you look at this landscape this is actually what it looks like and you can see from looking at those rivers the size of the boulders in the rivers right so this precipitation can move it effectively moves mountains and and we know that the Amazon for a basin is formed because of these materials that are coming from the from the Andes and so the synergy between the high elevations what happens in the in the mid elevations which are basically a transport region and then goes and and supports the Amazon basin so when we look now at this from the point of view of a river basin we can actually look at the distribution of of of these precipitation features according to the river networks and what you'll see is that we we can actually match each one of these of of these peaks and features in in this distribution to features in the landscape in the mountain itself and so it's this association between the spatial distribution of precipitation landform right at all scales and that can be explained by a dynamic processes hydrology in this case is is is really important so i'll show a lot of the Andes here today and so we have actually an observing network that goes from the low elevations in the Amazon basin up to above 4 000 meters in the in the in the Costa Nipata valley in in in Peru and we have these towers that are above canopy level so we can measure the precipitation without being affected by the vegetation you see here on the side this is a picture of the ridge where our where our rain gauges are and so what i'm showing here is a comparison between what's observed which is in blue from satellites and and those are different satellite products and what's observed at the gauges and the point of showing you this distribution i have two points here one is to focus on very light rainfall which is essential for the water cycle for many years we when we use rain gauges we miss a significant fraction of the light rainfall and so and and as you can see this is something that also we have a difficulty in seeing from we using ground-based radars and this is true also from space so we're missing a lot of the rainfall that actually explains the resilience of these of these ecosystems and then when we look at at the extreme rainfall what you'll see okay so the order that was a little bit different you'll see that up to 1500 meters we have actually observed as much rain as you would get from a hurricane like Katrina in New Orleans in about six hours so you can get rainfall intensities in these mountains that are on the on the order of magnitude of a hurricane impact and those are also not well predicted by of course or or observed by satellites or in fact by most by most of the of the ground observing systems so we were really interested in this very extreme event right because we looked at all other types of evidence we could find and we could not really explain this how does this happen and so after looking at this what we discovered was that all of these extreme events over over the history of the of the precipitation records that we could find in this region were associated with these cold air intrusions which are basically cold fronts that come from South America all the way up into into the Amazon and these and these and these events at the at the at the point where the air basically from south and the air from north along the end is meet that's where these extreme events form so they're extremely localized and you can get these very very heavy precipitation events and you can see that from looking for example design model simulations that show that the high the highest vertical wind velocities which are needed to form those really deep clouds that will produce that very heavy rainfall happen exactly at the intersection of the two air masses and so they happen in very very specific places okay so in this case we started with ground observations we said look at all this rainfall how could this possibly happen so then we did modeling to examine what was actually going on and so we identified processes after having done this we went back and looked at climatology from satellite again and and in this case from the team satellite mission and what we found is that when we look at those precipitation features that I was showing you before that seem just randomly distributed in that landscape actually they're very different when they are associated with these cold air intrusions and in fact we looked at the diurnal cycle of extreme rainfall everywhere along the Andes and we found that this was associated with cold air intrusions as well and so this is you know this is fascinating that we can do this because we're looking at what happens at low elevations we're looking at extreme events but I want to emphasize that before we did this study when we thought about cold air intrusions most of the papers were all about the effect it had on crops and because of the frost and how you know farmers would lose all their all their harvest because of these of these cold air intrusions and the link between cold air intrusions and extreme rainfall was actually had never been made before and we could only do this because we were there to observe it on the ground right and if we hadn't seen it we would not have believed it pretty much and then we had the model to help us build a hypothesis and we had the satellite data to to help us go back and really provide the historical context for for what we were seeing here and so this is an example of the kind of research that we can that we could do over the last 20 years that was not possible previously now focusing on this region this is really interesting is to look at the distribution of of of land cover along the Andes and and and specifically focusing on vegetation this is actually you know a historical historical drawing there from the Andes from Humboldt's trips in in South America right and it's amazing nothing is that to scale but but the science is is really there and and and quite and quite accurate and so I want to focus now on that picture that I showed you in the beginning where the tree line shows very clearly and and I went to go back and and look at the at the role of vegetation actually in in in helping not during the extreme events but all the other precipitation that happens in this region of the world is actually aided by vegetation and without the forests in the Amazon we could not we could not you know possibly have the vegetation that we are observing so there's a very a very close link between the cloud line and and the tree line in this region and so these are all very pretty pictures that I wanted to show you just because I had very pretty pictures and that's why I love the mountains and we love being up there but this is a depiction of our of our of our towers on this envelope monitoring network along the elevation and mapping also the the vegetation that goes with it and so you can see that at the lower elevations we have basically tropical forest right and as we go up with elevation we're getting into into the cloud forest at higher elevations so now I wanted to show you a little bit of of history of what things looked like many many years ago in terms of of the tree distribution in this region of the world and this is from paleo records and using our colleagues in in in in in other departments looking at at at this kind of analysis so what you see there on the right is the the the distribution of trees in the pre-inca era and what you see on the left is what it looks like the same region but after after the after the incas and after the the you know the the colonization and so on and so the first thing to go right as as agriculture expanded we're at we're the trees at at high elevations and so now what is going on though is with all the deforestation happening at low elevations against the the hill slopes of the Andes what we have is the pressure in terms of the vegetation is actually happening from the bottom up so we have this region of the world is actually under stress historical stress in terms of decreasing of the vegetation covered of the forests from high down earlier on and now presently going up from the amazon up up up the slopes and so what does that mean right in terms of of implications for what the resources in this in this region of the world so that's what we wanted to actually check and and and there has been you know many of you have probably heard about the red program where there's incentives uh to uh to local uh communities excuse me to preserve their forests and ecosystems and so on and there's a lot of work and huge investment made into into that work and so along with the with a number of colleagues that's a very you know that paper that is cited there has lots of authors because it was really a lot of work we did an inventory of the changes in vegetation since since these measures were put in place and in fact what you see is that at higher elevations there has been conservation of the forest but in the lower elevations that's not happening and that's what is you know basically along the uh the the foothills of of the Andes against um the amazon so for us then the the the scientific question and I'm sorry about about how this was working out was to ask if if this keeps going right what is the implication of deforestation along the foothills of the Andes on precipitation in uh in the Andes themselves and so what we did was a bunch of studies looking at sensitivity analysis between the control which is having forest and and the the another scenario which is a deforestation scenario and so what you see is that for all kinds of relevant uh weather systems weather types and so on the the vegetation plays a very very important role in producing medium you know moderate rainfall rates so we have a strong decrease in rainfall at for all cases in the elevations that are orographically active in terms of precipitation so we're talking about the region between the low elevations in the in the amazon and say about 3000 meters but you know the peak of the activity in terms of precipitation is at about 1500 meters and so what you see there is is this very strong decrease in the moderate rainfall rates and low rainfall rates and so this this is linking deforestation in the amazon right to drought and and uh and decreased freshwater resources at high elevations in the Andes so we we tested um this also through through more general modeling and and so what you see is the squeezing of that uh orographic freshwater harvesting zone and what I wanted to show here is that as a result of of of deforestation the precipitation generally goes downslope and becomes lighter rainfall so you saw the negative sign there was actually an increasing light rainfall now imagine and I'll show you later what are the plans for damn building along the Andes right there's a whole room basically the equivalent of the great wall of china but along along the the high elevations of the Andes but if there's no rain at high elevations those dams are going to be basically doing nothing right because you can't collect water and you can't produce electricity from this and so it's this connection between deforestation in the in the in along the foothills and precipitation at high elevations that's not immediately clear right it's it's something that we we can only talk about because we've we've done these studies looking at at at these at these problems from from different perspectives and of course if it's not raining at the high elevations either we don't have erosion and so the material fluxes into the amazon basin will be severely decreased right and that's what maintains the amazon functioning because of course you know the rivers in the amazon are carrying sediments to to the ocean all the time right and so you need the disbalance of of of materials coming in and and going out so it's it's really not something very simple you know it when we talk about deforestation in the amazon it's not just about carbon capture and and and and so on it's it's about much more than than that especially in terms of the of the water cycle so one thing one of my students lead with these studies where instead of cutting the forest we actually focused on on the flux by which forest really mostly impact precipitation and the atmosphere which is through evapotranspiration so he did some studies where we extracted all the evapotranspiration produced by trees along specific specific bands or elevation lines you know basically every time stepping the model which was like every 10 seconds or so so this was a major a major labor of love but this was really important because as a result of this we could actually associate evapotranspiration with moist instability in the atmosphere and that means convective activity and convective storms right and in along the entire and his and his eastern and his range and also very importantly associated with this is that you you can relate to this moist instability to basically the moisture convergence along the the autographically active region of of the Andes and so there's a significant decrease because we have less instability in the atmosphere the absolute pool of the of of moisture is is really decreased and so what we can see is that you probably have heard about how trees and how recycling of moisture is so important in the Amazon which is which is absolutely true but in terms of mountains the role of vegetation is not to produce mass to help with more precipitation is to actually produce instability in the atmosphere that is necessary to pull the moisture up up slope so that's a whole a whole different a whole different perspective and so that's why we can have a cloud forest at at 3000 meters and 4000 meters the only reason why we can do that is because we can pull that moisture and form and form those clouds so so effectively these mountain uh uh forests are pumping the level moisture through this process of of of instability of moist instability and and of course the impact of of just you know cutting the forest then because you have less convective activities about 50% decrease in precipitation which is really a sobering value right if we think about that and and think this is not so crazy because these regions are actually very thin the mountains are very steep so getting rid of vegetation over a very significant area is very easy in this in this region and I don't mean easy in a in a good way of course um so we did other studies also looking at the effect of recycling in the mountains versus versus the Amazon and what's the impact of cutting just forest anywhere in the Amazon basin on the on the Andes itself and so in the in the Amazon it has impact on the recycling right and so it decreases recycling of uh of water and so and and so you have less rainfall but in the mountains actually the forestation of the Amazon leads to a switch in the diurnal cycle of rainfall and you can imagine how that is is would would affect ecosystems for example and and the the montane forests and so on it's not so much that in the model of course you know these are model results that the amount of rain changes but that when it rains completely changes right and so the type of of vegetation and trees that would would leave in this kind of of environment would be completely different so I promise that I would show you the the distribution of dams planned for this region of the world and you can understand why this matters a lot right because we're talking about very very thin regions of the world with very steep slopes where a lot is happening that's where a lot of the activity is going on these are truly hot spots for the water cycle and so the implications in terms of of of societal you know projects dams economic development and so on in these regions are truly are truly huge for us in civil engineering when we think about water resources it forces us to think about things in a different way right it's thinking about continental scale and everybody's upstream and everybody's downstream right and so we cannot really think about you know so this I think we were very fortunate to to do this work in this in this age when we think about the earth system as a whole right and we are all connected and and that really bears true in these in these data and in in this analysis right it's not just some you know some nice concept that we we talk about and so I know I don't have a lot of time but I wanted to show you how land use right and land cover actually changes microclimates or climate in the landscape and so these are results from a climatology of low level clouds done using Modi's data so you know a more than a decade of data over the southern Appalachian Mountains and the the scale here the color scale is the number of days in the season in the specific season during which when when cloud cover was detected and so you can see as you would expect this is the area of the Great Smoky Mountains where there's always lots of clouds right it is a beautiful area and you can see that in the summer we have lots of clouds over the ridges in in daytime and that is true also in the spring although although in in lower in lower frequency but in the summer basically every day every other day depending on where you are you always have the the ridge embedded in in in cloud but see those very dark blue dots everywhere so when we first looked at this we didn't know what it was because we were just using satellite data and we're plotting them we had not plotted them on the top of a DEM or anything like that and then when we started looking at them on the DEM then we quickly discovered that they're not just associated with the topography because our hypothesis was oh you know maybe this is associated with some valley ridge you know type of process but rather actually associated with the TVA so the Tennessee Valley Authority as you know has developed or has built since since the 1930s it was the first large-scale system of DEMs in the world and it was both an economic and and a social policy sort of of a project in this region and you would not have expected that it would have impacted in this way I'm sure at the time nobody was thinking about the the impact on climate that you would have by putting all of those DEMs and so basically you can see that the clouds are forming all along the margins and that's where the rain is happening and so on so we have the the the lake breeze is actually controlling low-level clouds and and rainfall in in this region of the world and of course this is very interesting also because in the 1920s and 1910s and so on this region was heavily used for agriculture there was lots of of grazing you know cattle and so on and so it was only after the late 1920s and 1930s that you know these practices were abandoned and the National Park was formed and so today we have lots of vegetation in this region and beautiful trees again and so on and the question is what is what is the relationship here right so we actually have an experiment of recovery of landscape recovery and and the question of interpreting that is is very interesting so we've done some studies in this region I don't have a lot of time but what we have found is that in this region you actually cannot tell the difference between high elevations and low elevations in terms of a rainfall amount so when we have very large-scale systems you know basically like the not as the purple box but as the pink the pink ellipse when you have a large-scale storm system you know they come through these mountains are not tall enough to really create a difference anywhere and so you just have lots of rainfall everywhere but when we have locally controlled processes and that's basically the the valley reach you know type of processes and where those low-level clouds are forming then what we find is that actually trains more in the valleys than it trains at high elevation so it's actually a reverse a reverse effect and what we found in this region that was very interesting these are data from a large field campaign that was funded by NASA in this region and this was part of the ground validation effort for for the new precipitation mission the global precipitation measurement mission and as it happens when you you plan field work you hope for rain and we didn't have that much rain for the rain field experiment that year but what was interesting was that we measured the contribution of fog at some locations at in those regions that are embedded by low-level clouds the contribution of of what we call you know fog or natural rain into into the system was twice as high as the actual precipitation and so the point here being that these clouds and and and we have low elevation cloud forests in this region and so the fog and these low-level clouds are actually essential for for the resilience of these ecosystems and and for the hydrology of this region and by the way a lot of this rain cannot be measured by rain gauges because they just don't have the resolution to be able to get that so when we calibrate hydrology models to make sure that we get the rainfall from the rain gauge is sort of right we're probably missing sometimes in this region about 50 percent of of the of the water of the fresh water input into the system and so and I think I don't know how much more time I have I have about five minutes so that's great because I will I will go on to show you what that looks like when you're there and looking at at at what being embedded looks like from our own pictures and network and it impacts the usual cycle and so on so this is when we started doing some some micro physics work and actually measuring the rainfall process from from aerosols through activation through through cloud formation and through rainfall and develop models that allow us to explain right so not only we have more rain at at lower elevations but I I you know please look at at at this data that showed the size the average you know mass weighted size of raindrops and so what you'll see is that the green dots are actually from a station at higher elevation at about 1500 meters and the blue dots are actually in the valleys nearby and so this is because of this see the feeder interaction process where we have a light rainfall coming in with with with stratiform systems and as this rainfall falls through this low-level clouds and fogs that are embedding our landscape basically the raindrops grow you know they collect smaller drops and they become much bigger and you can see that's twice the size right of the raindrops so our models weather models and climate models cannot reproduce these results because we don't have yet the ability to solve these micro physics in the models right but implications of these of course for what happens locally and I will focus for example just simply on I will I will go over this and focus on we'll not discuss this but focus on the impact of of having a more accurate representation of the local aerosols in in a model and and what you will see I will skip this too because this will take us much longer and but I wanted to show this in terms of the impact it has on the surface energy budget right so it's not just about that we have these larger drop sizes in the case of rainfall but but but that we have these multi-layer clouds forming in this landscape and as you can see for example at a certain time of day the difference at some locations in terms of the surface temperature is like being under the shade of a tree right in this case we are in the shade of the clouds and and they can be as high as six degrees or five degrees so this contributes to huge gradients right in in in temperature in this in this landscape so you can have one slope that has for example 20 is at 20 degrees C and another slope nearby that is at 15 or 13 degrees and that's just because of the cloud of the cloud cover effect and so this is another thing that we also must account for in our water resources and hydrology models right because we're not really accounting for this a spatial variability and I don't want to be it sounds like I'm being very tough on our hydrology models and I'm going to say and they work right we have been making decisions using those models for 50 or 70 years and for the most part it's still working well and why is that because engineers always do everything with a little bit of a give right and so we always have a safety factor it's not exactly the same concept as you would as it would be in structures in a beam or in foundations but it's very similar right the idea is um is the same now I just wanted to I will I will end here with this is an example of why so I talked a lot about trees about ecosystems about these interactions but I wanted to show you that in fact this is very useful for immediate um having better understanding of the spatial of you know detailed spatial variability of rainfall in these regions for example it's very important for landslides this area of the mountains is actually one of the most active landslide regions in the in the continental us and if we don't get the precipitation timing right and the right amounts we cannot predict when landslides are going to happen but if we have that we actually can look at the entire uh water cycle at at the watershed scale and predict with pretty good uh confidence based on the data that we have when a landslide will happen and when we'll have the highest uh subsurface flow uh rates at at specific places in the in the landscape so um so this is back where we started right and so we talked about deforestation and land atmosphere interactions and then we talked about you know uh air quality aerosols and precipitation and that central map there shows planned dams everywhere in the world so as you can see every mountain barrier that's not in the US because all the dams have already been built in some ways but everywhere else in the world there are huge plans for building more and more dams right and so after you built a dam it's there right it's a singularity in the landscape and so any changes in the redistribution of water and fluxes have implications at at continental scale and uh I'll terminate there thank you how do you turn this on it's gone any uh thank you professor boroughs any questions for professor boroughs there are a few minutes uh thank you for the talk um really enjoyed it uh towards the end you had mentioned uh if you had like better precipitation data it could help with um better modeling for landslides and things like that uh kind of like a general question what other data sources do you think could help like enable you to build better models for for the prediction you're trying to do uh like what sort of gaps are there that you think can be improved improved upon so that's a really very hard question right because the question you asked I don't have an answer for that I'll say everything we need everything right and so it's really interesting when you make a decision to go from a a more applications oriented use of models right to a more science oriented use of models one of exploration then we discover that all of these interactions right the interactions between clouds and and the surface energy budget and of course land cover which affects albedo and emissivity of that land cover right and then the type of rainfall this you know the importance of this of light rainfall right the very slow slow amount but that actually it's keeping everything functioning right and and so how do we all of those data are very very important right and it's one thing we've learned I think also with satellite data is that the best products are the ones where you start from the the raw measurement or the original measurement our nominal measurement and then in order to come up with with a retrieved quantity we use as much information as possible right as we do that even our retrievals get better so um yeah I don't have an answer for that but I can tell you that as a hydrologist I do think that clouds and rain have been underappreciated for the for a long long time because we have focused on the on just events right on rainfall runoff response big floods and so on but there's what keeps the system going is actually a whole different set of processes okay thanks uh conversation about data in some ways there aren't enough data but in other ways we have been observing climate for such a long time that there are lots of data quality of data is another matter but within the available data are there scientific groups looking at those data from the point of using modern methods of machine learning and artificial intelligence to generate data that are realistic rather than waiting for collecting many many more data for this urgent problem so in terms of of using AI to get that retrieval right we have actually used some and and that's helpful when we have observations right and when we understand the processes I think so we as our host we have been using say for example neural network since the you know the mid 90s when it was a bad thing to do right where everybody said oh that's a black box that's not something that you should do and so on and yet we were very successful right at different scales in terms of of prediction not very successful in getting funding at least on my side to keep the work going but but what we have now is we have a lot more data right and we have a lot more algorithms and access to tools and so on I I think you can't do it blindly meaning you can't just throw a bunch of data at some tools and expect that they that they will uncover you know miraculous things but but I think the informed user and the one who is always looking for a physical explanation you know behind what's found then that is a good a good use so we've shown for example using very simple we started using lots of data all the data that we had available actually from gpm to make prediction of the of this low level of of whether it was raining or not right because it's very difficult for satellites in mountainous regions to see below two kilometers in the atmosphere and I just showed you that all the activity that matters is happening below two kilometers so that sort of undercuts the you know the whole premise so we have shown that by using data really and using these tools to help us find predictability we can make huge improvements right going from for example false alarm rates that are very high to very low missed detection in increasing up to to 90 right and and that's because of this combination of the physics guided or physics based development of AI so I'm I think it would be great to have people in computer science for example and and physical scientists work together in these problems yes dr boros thank you for your presentation very interesting I'm professor jackal and environmental here in civil you indicated that the ground truth data that you were taking was very important as a matter of fact the light rainfall was extremely important but you were missing a lot of it are you suggesting to the instrumentation designers that we need some better light rainfall gauges to capture a lot of the data that you say we are missing so thank you for that question because if it was a different type of talk I would have gone into those details so I appreciate it very much so we actually started measuring this very light rainfall by using these diameters and these diameters are sensors that with some uncertainty allow us to detect individual raindrops of our our cloud drops even right we have an instrument that allows us to go into into the into the micron you know like a 15 20 micron level of drop size and so by doing that what we found was that we had these diameters alongside rain gauges and vertically pointing radars and we could see the signal in the radar the rain gauge could not see anything but the signal in the radar was was was um actually very strong following the the you know the cedar feeder effects in increasing the raindrop size we could perfectly detect them in the in the radar at low levels but the rain gauge would take half an hour to start detecting rain because it takes time for those drops to grow and yet our these diameters were already measuring drops much uh you know for half an hour or more and in many days when we would only have that fog or those very low level clouds we can detect the drops we can characterize the size distribution and the rain gauges it's like nothing happened right so those instruments are there and and they're getting better you know in the beginning some of them had biases and so on because they are difficult to um uh to build but but they are there and and the message is that yes after you start using those instruments you start noticing how much rainfall you were missing before any other questions for dog um professor boros i'm not in the area but um you know there's been a lot of talk recently about uh attribution science um you know so modeling of course helps potentially predict what changes you might expect in the ecosystem you're modeling but looking backwards and making observations and saying that this observation uh could not have happened without said human activity etc you know going backwards how is that playing into the kind of calculations you do and you know because i i understand those are being done at very large large scale simulations you know to look at um you know control case and versus non control case how's that playing out in your area so this is you know this is a very good question for example those experiments where we looked at the effect of the forestation along the foothills alone you know those were very realistic so we used our analysis from that paper in in global biology to to actually map the region where the forest is being is being eliminated right and and and uh and so we had that information if we don't have that sort of detailed information it's very difficult to to make the analysis now i should say that there's no regional or global climate model that can resolve these kinds of a very thin but critical features of the landscape right because we don't have the resolution to do that and so we need to to be at at below one kilometer resolution to do this actually right and so that is really the big challenge right um but i think we can because again i want to emphasize that i feel very privileged to have started my research career at the same time as the earth observing system a program at nasa was was being developed and so my generation i and my colleagues all benefited you know tremendous huge from having these huge data sets right and so we can do climatology at very large scales now we can do actually climate studies using satellite data which was unthinkable uh 25 or 30 years ago and so that has allowed us to do attribution in a much more uh how to say cause and effect you know sort of of way because you can go and look in detail and look at what's happening in that landscape and what's changing not that we're discovering anything i remember being an undergraduate student and working on environmental impact statements which was a big thing right actually the u.s was the the world leader in creating this this concept of the environmental impact statement for civil engineering projects right and um and yet we haven't really changed much in the law or anything like that in the in in these 50 years but i remember that having a professor saying well this dam is going to be built somewhere and there's going to be huge implications for agriculture because nobody's expecting that you're going to have so much more fog that these types of crops will not survive and you know he was a very smart man who could make these associations but it's not the most obvious thing for most of us right what we can do now we because we have the satellite data to actually demonstrate yes that these changes are happening and and and and so and see what kinds of implications they have i don't know that the models are yet there but but we have we're very fortunate to have 30 years of satellite data that allow us to go back and check some of our hypothesis and premises and nasa does not pay me to say these things any other questions for dr baros yes no one second excellent talk dr baro is very inspiring i have a question you have done your field work in many different areas of the world regions of the world what kinds of relationships have you found between the type of forest and the type of vegetation and the type of fog cloud covered you know precipitation so so what you see is is actually very similar in the Himalayas we had the field a field site in the Himalayas for five and a half years and uh and so and that's why actually i went to the end is because we couldn't keep that field site going so we we moved to another big mountain somewhere and and we're finding of course the way moisture comes in right is is very different in the Indian subcontinent versus how it comes in in the in the Andes um but but in terms of the distribution of vegetation with elevation the issues of changes in temperature and and changes in moisture availability they are very very similar um but um yes the the monsoon season in south america is not the monsoon season in in or in india for example right and so those are significant differences but in terms of the ecosystems and and the role they play is actually the same so i had a picture i don't know if in my in my last slide that is from from the Himalayas but looking down and as you look at the Ganges plain do you see this haze everywhere right and and and and that actually plays a very important role also in um so pollution from from you know from deli when when we have some of the big pollution events that actually goes up the mountains and some of that of those aerosols are trapped in the inner mountain region between the the the ridges and valleys and that completely changes the diurnal cycle of rainfall in in those locations so um yeah i could i could have shown that but which which is really fascinating so you see how the transport of course is controlled by large-scale atmospheric um circulation but then what happens locally is really controlled by the mountains well i'm going to thank you all very much we are running out of time thank you all for turning the fascinating talk and also those of you online