 My name is Ravas Govindaraju. I'm the head of the School of Civil Engineering. I also would like to welcome you all to this distinguished lecture series from the College of Engineering. We have a very accomplished speaker with us today. A couple of announcements I would like for you to hold your questions till the end so that we can pass the mic around to you when you're asking questions so that it gets recorded. To help me introduce the speaker, we have with us Dr. Irvind Raman. She's a senior associate dean of the faculty and the Robert Adams Professor in Mechanical Engineering at Purdue University. He went to IIT Delhi for his undergraduate, was at Purdue for his master's, went to UC Berkeley for his PhD and joined us in 2000 as an assistant professor. Irvind's research interests are in nonlinear dynamics, vibrations, fluid structure interaction. He could be a civil engineer, you know, with that kind of research interests. I don't get to embarrass mechanical engineering faculty very often, but now that I do have the chance, let me tell you a little about some of his interests. So Irvind has essentially pioneered the use of cyber infrastructure in our community here for research, education through advanced simulation tools, online classes, which are used by thousands around the world and led the College of Engineering in strategic initiatives for global engagement in Latin America. Irvind is also an ASME fellow recipient of the Gustus Larson Memorial Award for the ASME Keeley Fellowship from Wadham College, Oxford. He's a College of Engineering Outstanding Young Investigator and also an NSF Career Award winner. Irvind. I think we should keep our claps for our distinguished lecturer today, but it is really a great honor here to welcome our lecturer, Professor David Maidment. A few words about the Purdue Engineering Distinguished Lecture Series. This lecture series started this semester and aims to bring the brightest, you know, the best, the leading faculty around the world to come to Purdue Engineering and engage in a discussion with our larger community here on grand challenges as well as grand opportunities in the specific area of engineering. And today we are really proud and privileged to have Professor David Maidment join us today. Professor Maidment is the Hussein Al-Harthi Centennial Professor of Civil Engineering at the University of Texas at Austin where he has been since 1981. In 2016 he was elected into the National Academy of Engineering for the application of geographical information systems to hydrological processes. He's the recipient of many, many awards and I'm going to read out just a few here. The Mike Howard Lectureship of the Texas Floodplain Management Association. He was named the Geospatial Scientist of the Year by Geospatial Media. He received the Ray Lindsley Award from the American Institute of Hydrology, the Vente Che Award from the American Society of Civil Engineers and several others. He even has the American Water Resource Association name and award in his name. Without further ado, let's give Professor David Maidment a big hand. Well thank you and I really appreciate the opportunity to be here today and to talk with you here at Purdue about something that I've been involved with for the past several years which is the development of a national water model of the United States. And I've put a map here of the rivers of the nation that are simulated in this model and also of Indiana and I'll show a little bit as I go in what this means for Indiana and even for what this means for a typical no county. So I want to start with a little bit of a philosophical introduction here and this was a book which I worked on for four years, quite a long time ago, a handbook of hydrology which covers all subjects in the field and I asked myself at the end of this effort how was the knowledge in this book worked out and I came to the conclusion that the knowledge in the book was worked out by three methods by deduction, by experiment and by observation and by deduction I mean by reasoning from an existing base of knowledge. So for example given a certain proposition and a reasoning process I reach a conclusion, that's the classical path of mathematics. By experiment I mean the reduction of nature down to a small microcosm and its replication in a laboratory and by observation I mean the direct observation of the natural environment. I don't think there are 23 things, 23 ways by which new knowledge is discovered I think there are three ways and this is what they are and in fact by combinations of them. So by deduction to me the greatest example of deduction is Isaac Newton and I know you've all got the principia by your bedside table at night so you can study the works of Isaac Newton but this is my copy of the principia, the wonderful translation made by the University of California Press and in 1686 in the introduction of the principia Newton said the ancients divide the mechanics into two parts that the rational which proceeds accurately by demonstration and the practical it occurs that mechanics are so distinguished from geometry that what is perfectly accurate is called geometrical and what is less so is called mechanical yet the errors do not come from the art but from those who practice the art. So in this sense rational mechanics will be the science of motion resulting from any forces whatsoever and of the forces required to produce any motions accurately proposed and demonstrated. So this is what Newton wrote in 1686 in the introduction to the principia Mathematica and his contribution was he established mechanics on a precise basis and he used geometrical arguments to do that. So this is a very sharp edges, right? This is what Isaac Newton was about. To me the greatest of the experimenters is Louis Pasteur. Louis Pasteur was a biologist, a French biologist and he simplified the whole world down to a laboratory. He was the first person who demonstrated that microorganisms in the human body cause disease. He also was the first person who developed vaccines so he developed the technique of vaccination initially for anthrax and for rabies. This was a foundation of scientific medicine and for hydrology it's hard to conduct laboratory experiments on a small scale but the Darcy, Henry Darcy who developed Darcy's law for groundwater flow is one of the people who did that. But for me, anyway, Pasteur is an example of the experimental method. And for me one of the greatest observers of the environment was Charles Darwin. This is my copy of his book, The Origin of the Species and he said, When on board HMS Beagle as a naturalist I was much struck with certain facts in the distribution of the inhabitants of South America and in the geological relations of the present to the past inhabitants of that context. These facts seem to me to throw some light on the origin of the species. That mystery of mysteries as it has been called by one of our greatest philosophers. On my return home in 1837 it occurred to me that something might be perhaps made out of this question by patiently accumulating and reflecting on all sorts of facts. That was my emphasis here which could possibly have any bearing on it. I can here give only one, the general conclusions at which I have arrived with a few facts and illustrations but which I hope in most cases will suffice. And this book was published in November of 1859. It recently had its 150th anniversary and it's been called the most accessible book of great scientific imagination ever written and it's still in print. And notice the difference in philosophy between Newton and Darwin. Newton's trying to make everything with precise edges. Darwin's comfortable with ambiguity. He knows that you'll never resolve this, right? So you've got one world here of mechanics, of prediction, of deduction. You have another world of observation, of uncertainty, of ambiguity. And it's in the bringing together of those two worlds I think that in hydrology we have to contend. And this is a simulation that's been made by the National Water Model. So this is a modern form of deduction. Given rain we calculate flow across the whole nation. What you're seeing here is a simulation of flow on the three hour time steps for a three month period during May, June and July of 2015 there was part of the spin up for the National Water Model and the colors here represent the magnitude of the flow in the major river systems of the nation, of the continental US. And the blue is smaller and you can see as time passes the effect of storms passing across the country and the rivers of the nation, sort of like the bloodstream of the river system that's flowing along. I first saw this, the president had a water summit in 2016 and we saw this at the White House actually when the National Water Model was unveiled. But this is really an amazing accomplishment which has been achieved by the National Weather Service and so water is now continuously being simulated and forecast throughout the nation using this National Water Model. And I want to tell a little bit of the story about how this came about and then what some of the implications of its existence might be. So the opportunity was established when this center was created on the Tuscaloosa campus of the University of Alabama. This is the National Water Center and it was established by the National Weather Service and you may say, why Tuscaloosa? Well, thank you Senator Shelby, that's his hometown. And so you know, 25 million or whatever, 50 million fell out of the sky and they said, okay, let's do it. And so this, but it was sort of an interesting thing because having a National Water Center then created the opportunity to assess hydrology in a new way for the nation. So there was an opportunity here to sort of say, here's a blank canvas, how can we paint this in? And then thinking about that, there are really four water problems that need to be solved. One is too much water which is largely what we focused on, flooding. Another is too little water, not enough water, it's droughts. Another is dirty water, it's polluted and then environmental issues, ecological integrity and so on. So in the larger scheme of things, the goal is to address all of these questions and the short term, flooding is the focus. And that analysis is now conducted at River Forecast Centers. The National Weather Service has 12 of these in the continental United States and these different colors that you see here represent the regions of the nation that are forecast by different river forecast centers. I think here you're in the Ohio River Forecast Center region, which is the blue one and you can see large river basins here that are forecast with a hydrology model based on average basin characteristics. And what is now happening is that these the activity of these centers is all being nationalized at the National Center. So just like there's a National Hurricane Center that deals with hurricanes wherever they occur, the National Water Center will be the center for doing flood prediction across the nation once. All of this is centralized and functioning from a single system. Now the computation is not done in Alabama. The computation that's done part of the weather forecasting system in Washington but the management of the model is done in Alabama. So this is the National Water Center. There was a conference there in May of 2014. This was the people that were at that meeting. I was one of the speakers at this meeting and you can see this it's a charge of mahal, this is a place, it's a huge building and they didn't have hardly any staff there and it's only slowly building up and so I thought well you know you could regard that as a problem or you could regard that as an opportunity. So that evening actually when the picture was taken I proposed to the director of the National Water Center how about we create a new flood data modelling, forecasting, inundation mapping system for the nation, atmosphere to the oceans, coast to coast, near real time, high spatial resolution in one year. Just bring in the academic community and just crash it out. You know I add in mine hackathons and sweaty bodies and you know all this kind of student activity and I said at the end of this message if you know if you think this is too crazy it's okay you know just this don't just put it in the garage can I won't worry you know but that's not what happened they put this message up on the big screen in there and a couple of days later they said okay we'll go for it and so what has happened since then is the National Weather Service Innovators Program which is a partnership between the National Weather Service and the academic community and this money passes through the National Science Foundation to the consortium of universities for the advancement of hydrologic science and it includes a summer institute for graduate students and 105 graduate students from 49 universities have participated in summer institutes at the National Water Centre and a number of them are here in the room so hey if you've been to the summer institute raise your hands we've got some graduates here yeah okay Ryan yeah so okay so Dr. Merwade has been very he's been participating in the National Water Centre summer institutes and Ryan was a student when he was at Auburn University so yeah so your students have been participating in this experience and this is really casting forth a new vision for hydrology for the nation of being able to think about something at the continental scale so you know this is the simulation that I showed you earlier and it is based on this geospatial data set called the NHDPlus so the NHDPlus NHDP stands for National Hydrography Data Set and that's the blue lines on maps the national elevation data set is the terrain elevation the watershed boundary data set is all the drainage areas of the country and the national land cover data set is the distinguishing characteristics of land cover each of those data sets took 10 years to develop roughly between 1995 and 2005 then it took 10 more years to put them together so that you have this area of land flows to this stream and this is the characteristics of that area of land and so on so by about 2015 basically there are a lot of very small catchment areas on average 3 square kilometers and a reach length of 2 kilometers where each little catchment just drains one stream so the land and water systems shake hands one to one like that then all the computations are done on top of that catchment system the first prototype for the national water model was built and run in Austin on this computer the Texas Advanced Computing System that tank is right outside my office 1.2 million gallons that cools the tank during the summer when it's too hot to breathe in Texas and so we used a small portion of the Texas Advanced Computing Center to do this and what we were doing was to take a weather forecasting model which is called the High Resolution Rapid Refresh Forecast or the HER so there's a weather computation then that feeds weather and precipitation into a land atmosphere calculation model which is called NOAA MP which is calculates the soil and water energy balance on grid squares then that's mapped over onto this river reach model which is here shown for the Mississippi River which is all we had at the time we started this and then that produces stream flow forecasts which produce the probabilistic flow forecast that you see at the top there so the pieces that were new in this process the weather model and the NOAA MP already existed but running them in real time and transferring the result over onto the land surface and routing it through the land surface that was the new part that was developed with the national water model we demonstrated that that could be done in 10 minutes for the whole continental US and I have to say you know I didn't know I didn't even believe that was possible when we started I mean I was the one who proposed it the whole country in 10 minutes I mean just that caused a few ripples when that happened and this is using a framework called WORF Hydro which comes from the National Center for Atmospheric Research and that's how all the dots get connected so I want to acknowledge the contribution of David Garchus here at NCAR because he's the one who built the framework that we used to do this and so what's happening is that we've got the what we call and hydrology in GIS the vector world which is the catchments and reservoirs and rivers and so on and that's how the water is routed through the horizontal part of the landscape and then over that is laid a grid initially it was three kilometre squares and now it's one kilometre squares and all the land atmosphere calculations are done on a grid and then a geospatial transformation is done of the runoff from the grid squares onto the catchments and then everything is routed through the horizontal flow system that's how this process works and in the so what's happened is that all of this is now being moved into the nation's water forecasting system so in Washington there's something called the WCOS which stands for Weather and Climate Operational Supercomputer System so there's a big thing doing weather and there's a smaller piece of it that now does water all the time and so there's four models that are run continuously the first is called the analysis model which is the best estimate of current conditions two days up to the present and the present the second is the short range forecast which is repeated hourly for 18 hours ahead then there's a medium range forecast which is three hours time steps for 10 days ahead then there's a long range forecast which is for 30 days ahead and there's five terabytes of information that's produced each day so you see a little note there that says Wabash River at West Lafayette, Indiana so this is the national water model rivers and streams in Indiana Indiana is divided into 21,501 reaches for the national water model calculations and this is Tippecanoe County so here in the county there are 472 separate river reaches that are simulated by the national water model and the water flow is updated every hour and one of these happens to be on the Wabash River I didn't realize until I started studying your campus that you've got a pretty big river that runs right by the campus here so on the Wabash River this is one of those little reaches so the reach ID here says 105,115 and so that's a particular section of the Wabash River right here in the campus area and that particular river reach has these forecasts that were made I made this slide a couple of nights ago and this is the orange that you see on the left is you've had some high flows here you must have had some rain a few days previously so you can see these high flows and now the forecast is that over the next few days the Wabash River is going to drop down but it's still going to be 2,000 CFS that's a pretty big flow by itself it's going to be up but not as high as it was for a few days earlier so the orange is the analysis model of what was going on up to the time the forecast was made the red that you see here is a short range forecast for 18 hours ahead and then the purple is a medium range forecast which goes for 10 days ahead so this is being calculated everywhere all over the country and this is what the current forecast for the Wabash River now I want to turn to a different subject here now and that's Hurricane Harvey which has had a particular impact on me so this is Harvey it came ashore, Hurricane Harvey came ashore on the night of August 25th so if you follow the hurricane track as it came in from the Gulf the intensity of the hurricane intensified it went up from being a tropical storm which is the thing marked in green to various levels of hurricane and finally it was category 4 hurricane winds of 130 miles an hour when it reached the coast which was near Rockport, Texas which is well I used to have a condo at Port Aransas which got the roof ripped off so we had a catastrophe there it wrecked our coast with the high velocity winds however the hurricane went and land and then it turned around and went back out to the Gulf again and even while it was still out in the Gulf you can see the rain bands over Houston so Houston is 165 miles to the northeast of Rockport but the hurricane impacted Houston and southeast Texas despite the fact that the center of the hurricane never passed over that area so the hurricane was spinning it was bringing in big rain bands of water off the Gulf and it was just deluging southeast Texas without ever covering the area that was most impacted by its effect so I've marked on the bottom here 5 days so the hurricane came about 10 o'clock at night on the 25th of August and it continued raining for 5 days it was kind of like a nois flood happened in that time Hurricane Harvey was the worst storm ever recorded in the history of the nation so on the vertical axis of this chart is the amount of rain the depth of rain that fell on the horizontal axis is the area and from 1,000 miles to 50,000 square miles the dots that you see here represent the worst storms that have ever occurred in the history of the nation and so there's the second most the third most, the fourth most, the fifth most Harvey is number one in all cases and if you take the average between the blue dots and the orange line that's 11 inches of rain now 11 inches of rain is a catastrophe all by itself this is 11 inches of rain on top of the worst storms that have ever happened in the history of the nation, that's what Harvey was this is, I mean off the charts doesn't even record this fact this is completely unprecedented level of event in the country for 3 days Harvey was 5 inches more than previous storms for 2 days it was right at the top of the previous storms so I was 10 days at our state operation center in Austin helping the Texas Division of Emergency Management do flood emergency response for Hurricane Harvey and I have to say it was the 10 most intense days of my life you just cannot imagine the pressure that's involved with dealing with the catastrophe of this size there were hundreds of aircraft and helicopters there were thousands of trucks and boats there were tens of thousands of personnel our entire National Guard was mobilized 16,000 troops, there were military people there there was urban search and rescue from all over the country was there the way that the emergency response is organized cities are within counties and counties are within disaster districts which you see these colored areas here the disaster districts are within these green coordinating areas and then at the top is the state operation center so all the resources coming in from around the country are coordinated and in response to requests coming up from the local areas through the disaster districts at the state operation center and Chief Ked is the one who makes those allocation decisions this is what the state operation center looks like when it's running there's about 120 people in there the first thing that happens is that there's a weather briefing and the screens that you see here are some of the information that was being provided at the weather briefing purple is bad any time you see purple that's bad and that's the flooding line so every day they had flooding is bad and the different little red things that you see at the top there are the different functions so one of them is finance for example I mean you're spending just millions of dollars you have to keep track of that because you're going to pay it afterwards there's logistics, there's operations there's deployment and so on communications so every day there's a briefing, a weather briefing and then a review of around all the disaster districts and then a review around all the functions of the state operation center and then that's how the coordination is done we went on Tuesday they said last night there was 26 inches of rain on Beaumont it's now Venice the water is just instantaneously you've got feet of water by that point there was an all-out assault and all the resources are there the air was so thick over Beaumont with helicopters they couldn't get any more in, there were 85 helicopters up at one time, pulling people out of the water like you see here and this is really dangerous work you can see the operator on the hitch there the operator goes down, cinches the victim up to himself and then they haul them up again, one at a time they've got to come up through the trees, if somebody's in a tree you've got to pull them up through a tree this is very very dangerous work and then into the helicopters then the helicopters went to the high points where you see the Chinooks here, the bigger helicopters and the big helicopters shipped them out of the region to Galveston and other places now this was a an enormous effort the Air Force was involved the Navy was involved and it's to be said that this is in the Harvey, 80 people died and every death is a tragedy and I don't want to minimize that but in Katrina, 1,800 people died and it was because of situations like this that a better response happened during the flood one night, in the state operations said I'm called Dark they're not professors or something and I'm just called Dark I'm the kind of data guy and I walk past the office of another guy, he's called Country Country is the youth director of urban search and rescue he had 142 aircraft and helicopters Country says they fly, they fly you get a aircraft, you don't get an aircraft and his face was just absolutely fixed and he called me into his office and he just said Doc, I need data it was one of those no fluff moments where you just cut right through all the talk and so I suggest we need a request to the president of the university so the Taxis Division of Emergency Management sent a letter to the president of the University of Texas and said look, we need help here we need expedited water information that can help with Harvey so one of the things that we've been working on with the National Water Center and with the help of researchers, actually some of the people in the room here have worked on this is a method for flood inundation mapping called the height above nearest drainage or hand and what that means is that if you imagine a point on the landscape where there's a house that if you say okay the depth of the water above the stream at that location is called the height above nearest drainage if the depth of the water in the stream is below that then you're okay and if it's above that then you're not okay and we've computed this map of relative elevation or hand for the whole continental United States as part of the work that we've done with the National Water Center and the Summer Institutes actually this work was done at the Super Computer Center at Illinois the Roger Super Computer Center at Illinois another thing that we've done is to collect all the address points in our state so you may not realize this but every building in Indiana to which you can make a 911 call has a dot on the roof metaphorically speaking and so if you make an emergency call that call gets routed to a dispatcher the dispatcher doesn't see you they just see a dot and they say police go there, EMS goes there fire goes there and 55,000 of these dots in Texas as a whole there are 9.2 million so we went around all emergency communications districts and we connected all these emergency points because that's where people live so if you want to know where flood is impacting and how many people are impacted then having these address points is a critical piece of information actually so during Harvey what we were doing is that we had a system that was automatically converting flow into depth into inundation using the hand method and then inundation into impact by counting the number of address points that were flooded so we take the discharge forecast from the National Water Model we transform the discharge into depth by using a rating curve that's calculated from the hydraulics from the hand method we create an inundation map from the water depth and then we assess the impact on people and property and so the importance of this is that from a hydrologic perspective, from a water perspective okay we've done our thing when we've calculated the flow but from the emergency response perspective it's how many people are impacted and where? We've got all these resources flowing in do we send the helicopters there? do we send the clearance vehicles there? How do you make those decisions in a major crisis? So we had this calculator running and this is a calculation that was made on Friday the 25th of August at 1500 hours and it was at 3pm in the afternoon now bear in mind the hurricane did not get ashore until 10 o'clock that night so this is 7 hours before the hurricane got ashore this was the estimate of what the damage would be from Harvey from flooding Region 2 has got here 238,000 addresses flooded Region 2 is Houston, Beaumont, Port Arthur South East Texas and that's what happened Region 6 which is the next one to Region 2 22,000 addresses flooded and so in looking back on the experience that we had in Harvey I think this is a pretty important data point what this said is that it was possible with this computational machinery to be able to estimate what the impact of the hurricane was going to be before it even got to the coast now this is only one data point and we have to sort of go through a number of these things and figure out whether that's a viable method in the future but assuming that it is through what that means for public safety and for whether or not to order evacuations and so on then that's an important piece of information I think I don't know if you knew this but the largest evacuation in the nation's history just happened in Irma 6 million people evacuated from Florida they thought Irma was going to go through Florida and the governor called for an evacuation of the state 6 million people left Florida that's the largest evacuation ever that's ever been ordered in the nation's history but you know you don't order those kind of evacuations trivially in about 4 or 5 years ago after Katrina Texas ordered an evacuation of Houston 4 million people they tried to get 4 million people out of Houston and that was a disaster more people died in evacuation than died in the hurricane so you know evacuation is a very critical thing you know how to make that decision especially a big decision for a huge city like Houston one of the other things that we did with the address point data was to estimate the impact of flooding on major river systems and in the flooded areas so the picture on the left shows the inundation of address points on the natures river which is near Beaumont and then on the picture on the right is an inundation area actually in Beaumont itself the red squares that you see here are the US national grid this is what's used for the deployment of air resources in rescue so when the helicopters go out they have a military system and they say you go there and they go there on the US national grid system so the dots there are the address points the blue is the inundation and lots of people contributed to this process and just to give you a sense of the pressure involved in this one night our faculty team that was working on this we had a telecom at nine o'clock at night and just before that we got the mapping in for five rivers the Colorado Brassus Trinity and Nature and Sabine and we did this overlay here and about one o'clock in the morning we concluded that one hundred and seventy thousand people were going to be flooded on the Brassus River and fifty thousand on the Colorado River and we communicated that to Chief about two o'clock in the morning I mean you can imagine this is not small numbers of people here this is a huge huge impact that's going on so this is our new Stampede system Stampede 2 and so what we were doing after we got the call from the TDEM the Texas Division of Emergency Management was creating these flood inundation mapping for local scale for use in urban search and rescue and this is my colleague Harry Evans he's Chief of Staff of the Austin Fire Department and he's been working with us for three or four years so he's our public safety representative and we do the science and he does the connection with public safety but one of the things I've really learned out of this process is the enormous economy of scale of high performance computing and I just didn't know I mean I'm a hydrologist I've been doing my laptop computing for you know decades I didn't know you could pull off this kind of stuff but it's just really amazing there's an enormous economy of scale I just heard when I was at the National Water Centre they say they can do the inundation mapping of the country with the hand method in six seconds and that kind of thing is really critical in a time it really matters now where are we going to how can we look forward what does this kind of extension of the Newton world let's call it that call for well the first thing is we need more Darwin we need more observation you can't suddenly go from forecasting 6,000 basins to forecasting 2.7 million reaches you need some more measurement to support that so if you think about that it turns out that there's 27,000 Taxas Department of Transportation bridges whose flow is forecast by the National Water Model well the US Geological Survey has gauges and that's considered the gold standard in hydrology but what about all these bridges how about we do something about them so it turns out that these 27,000 bridges have 15,700 reaches of the National Water Model so we've got 15,700 entities here that we're calculating and 27,000 structures that are sitting on them and what we did during the Harvey was we're working with this commercial company called Kisters they have a big data system and we're just hearing about cyber infrastructure well this is a commercial big data and they call it the PET in Germany they're ingesting all the flows from the National Water Model they're converting flows to depths using the hand method and then they're putting out forecasts of flow and water level at each of these forecast reaches and this was a portal that we opened up for the Taxas Department of Transportation during Harvey so all the green dots here are the bridges that were in the impact zone about 8,000 bridges in the Harvey impact zone and it took three hours to do that now if you could have said to me we could have had a forecast portal for 8,000 bridges there's no way that could happen but it happened and all the other dots you see there are US Geological Survey gaging stations in different parts of our state so in thinking about this one of the things that we have been testing in the Summer Institute is the idea that you can measure stream flow using radar so the measurement that you see here is on the Kahaba River near Centerville, Alabama and some of our students know the Kahaba River near Centerville, Alabama spent some time there and so what's going on is that the level of the water is being hit by a radar beam the velocity of the water is being measured by another beam which is at an angle and it's using a Doppler system and then another measurement acoustic Doppler current profile is used to store the or to calculate the flow across the whole cross section and by using the reference velocity at the surface and the flow across the cross section the whole things put together and so the philosophy is let's put an instrument that just attaches to a bridge and use an indirect measurement of the flow water surface elevation and discharge and when I started thinking about this I thought hey wait a minute if I thought about this as a hydrologist I have water sheds I have outlets I do all that stuff but if I think of this as a transportation engineer yeah if we were to do that we would think about major highways so this is the bottom line that you see here is interstate highway 10 above that is 20 then we have 35 and 45 we've got 30 so if you think about the highway system we have the highways go across this way across the nation and they go down if you think about this in a sort of a scientific sense hey that's a transit model we could instrument the highway system the interstate highway system and be sort of like packet lines and so yeah we managed to persuade tax dot that yeah we should do this so they've started embracing the idea that when you build a bridge you put a water sensor on it and so our first experiment is going to be on interstate highway 10 this runs all across the south part of Texas San Antonio on the left and Beaumont on the right Houston in the middle these green areas you see here are districts that the tax dot has for describing its operations its transect of 20 gauges across I-10 and hopefully if this experiment is successful we'll do more of this and while we were doing the this planning this project I asked tax dot well they asked tax dot asked for what would be a build out you know if we did the whole state and so we did the calculation I said well if that would be one gauge and 10 on the 15,000 meters there would be 1500 gauges and the cost of that is about a million dollars well you know 50 millions not too much for tax dot they build highways one span on one bridge is 2 million you know 2 lanes each way that's 8 million yeah 6 of those you've got a measurement system and it turns out that we're starting to get a bundle of money back from the federal government I think it might have been voted on today actually that it's coming back to Texas and so 5 billion is coming to our Texas general land office 500 million is going to go into a research program and so okay maybe the 50 million is going to happen so it's not out of reach at all it's something like this could happen and actually quite seen that we could actually instrument our state and if we did that we started off with 500 USGS gauges we'd have 1500 of these radar things that would be 4 times the points of measurement that we have now so let me finish then by saying that I tried to create at the beginning a sort of a philosophical introduction here that says that there's deduction or prediction as one form of gain of knowledge another is observation and I think those two things go together and we've done a big step forward in prediction here we need a big step forward in observation to sort of match that up and it's an amazing thing water is now like weather forecast in real time for local streams all over the country I guess those of us in hydrology just never really thought that was possible but it is I mentioned earlier that high performance computing has an enormous scale I mean this is incredible what can be achieved and this is a continental scale view of hydrology so thinking from the top down and not from the bottom up the hurricane Harvey was just way beyond any historical storm and there was an earlier storm in Houston tropical storm Allison but they said it was off the charts I mean I've been teaching charts for 40 years I mean this is what I've been doing as an engineering professor it really bothers me off the charts what the heck is that all about and now we come in with Harries twice as big again that's way off the charts to me that's a failure of engineering frankly and we need to fix that I mean this whole off the chart thing and when you consider the public safety implications of that it's just completely nuts and we should be thinking about stochastic storm stimulation and bringing these storms inland and not just stopping them at the coast you know there's a serious deficiency of engineering design in this whole area I think that needs to get fixed and that densified prediction requires densified observation if you go from the left at the bottom there to the coal continent of the United States and when you're talking about the flow at the local portion of the Wabash river here and Lafayette Indiana you need more measurement than we've got now to support that thank you any questions if you have a question please raise your hand and Ganesh will bring the mic to you very good this national water model that you have can it forecast it can also use to know what will happen say how far the flood will take place if I have this many inches of rainfall etc with a given land use now after Harvey Houston is figuring out what to do with what will allow building etc can the model be used to find what areas to stay away from under certain because you have to really as I was telling you this morning the land use has to be brought in and because the model you have is basically static perhaps and some way it has to be made at least 5-10 years increment or I don't know what increment how the land use changes so that's a really good comment and so when I put this note here that says need to new years national water model in design mode that's kind of what I was getting at so traditionally in hydrology we've looked at the history of the rainfall that has occurred in the past and inferred design standards from that but if you think of design as design for the future so it's not just what we have built now it's where to build those are important questions if we build there what infrastructure do we need what I've seen in Texas is that much of the trouble that we've got is the result of decisions that were made many years in the past whose consequences were not understood at the time subdivisions were established in low lying areas whose risk was not known when they were established they've become part of larger cities and suddenly there are flood prone places and you have to buy them out and in Austin for example our citizens have already paid $100 million to buy out houses from low lying areas to the south of our city $100 million for a city of less than a million people that's a lot of money there's some federal money too but local money is being used for that so I agree and so this is really an important question here is to how you anticipate what development will happen and what will be the impact of the way that's organised right now you hear Houston as a place like there's no zoning I'm the chairman of a national academy we visited Houston about a month before Harvey it's the wild west of urban drainage here's the subdivision that has no pipes here's the subdivision that has some pipes here's the subdivision that has ditches here's the subdivision that has nothing at all it's like a patchwork quilt the future development will be and also if we have another subdivision and it gets put into this place what will be the effect on others of having that there there was unintended consequences for example sound barriers we put up a sound barrier for sound oh wait a minute that becomes a dam when they have a flood there are things like that that would just produce worse effects than what needed to be in Harvey so thank you Dr. Mademond for coming here so there are a lot of students in the room so as an engineering professor who started with basics and now we are doing super computing so what should we be training to our students to prepare them for all this that's another good question I think we should be paying more attention to computational science than we have in the past it's when we did this prototype for the National Water Model which wasn't even called the National Water Model at that time it was called the National Flood Interoperability Experiment but there was one routine in the model that we had there was about 100 lines long and that was a real bottleneck in the code and it took 10 hours to execute and somebody from the Texas Advanced Computing Center looked at it and said oh yeah you need to use hash tables and 10 hours became 3.5 seconds I'm not exaggerating that's exactly what happened 10 hours became 3.5 seconds that's how we succeeded in doing the whole country in 10 minutes now a computer scientist spotted that my student who wrote the code did not know and he was a hydrologist how would he know that so I think there's a greater need for integration of hydrologic science and computer science and computational infrastructure what that means I think there's a big leg up there that we could be climbing on that we probably aren't right now because that connection isn't being made too well I don't think I was wondering what your philosophical views are on going in the other direction of simplifying things recognising things are unpredictable stochasticity, non-linearity behaviour of complex systems that's a good question let me give you some reactions after the first summer institute in 2015 what we succeeded in doing then was to go from rainfall to flow but we didn't succeed in figuring out how to do the flooding part and so few weeks afterwards I started thinking about this and I suddenly realised wait a minute this is atmosphere to the oceans this is one huge network this is a network problem we have to solve traditionally what we do in hydrology is we solve this one reach at a time and we agonise over how high the water is and what the bridge is this is a national problem now if you're going to solve a national problem you have to simplify so the first thing I got was this is a network problem has to be solved as a network problem so we have to think about the links and the connection among them and the second thing that I realised was in hydrology we've always taken channel cross sections by cutting vertically there's no reason why we couldn't do this we can cut horizontally so in fact if we say ok we're going to now cut along the channel and at a certain level we know how far the water spreads out oh now we've got geometry and we have anundation map at the same time oh ok and that's what we're doing with the hand we're cutting the channels along this way instead of cutting them that way now by doing that and by just saying at this flow this level is going to be obtained it's a really simple model I hate it for that reason they say oh it's horrible it's much better this list flood there's lots of other methods that are better but the trick is that what can you do that's of acceptable accuracy that can be executed fast enough that you can achieve the result over the whole nation and that's what we're trying to do with the hand method we're trying to simplify the whole process now if we can do a better solution then that's fine but I've had in the past that I say any fool can be complicated it takes real insight to be simple if you get something simple then you can generalize it out to a lot of ways and for example I went to see the local urban search and rescue team operating in Austin the boat team that we have we have six boats in the city and I saw the commander his name is Chief Pomeroy he said he comes to the place where the floods happening he's the incident commander he puts a rock on the road where the water is now so we can tell whether the water is rising or falling and I thought rock on the road huh and then he's getting messages all the time from his bosses back at the Emergency Operations Center how big is the problem you're facing how many houses are flooded it's two o'clock in the morning it's raining it's dark I can't see but wait a minute rock on the road if I know how high the water is and I've got this hand map here's my flood map so I don't even need any model at all all I need is the rock on the road and I can draw a map so a simple model could be used by a local first responder to get a map without even using a hydraulic model at all so I think simplicity is really important especially at this scale it's important because you have to do things really fast to get solutions at this scale Hello, I grew up in Houston my whole life through Allison I grew up here for Harvey I was up here at school but what political changes do you feel should be recommended to just the state of Texas because I do know they have different laws in terms of zoning and just how they're allowed to develop their land That's a really good question Texas being what it is I don't think the state is going to be telling local communities too much about how to conduct their business but well they actually do do that they told us we can't have plastic bags and some other things but anyway aside from that so let's just take Houston as a case study so so in Houston in the city of Houston all the internal drainage is controlled by the city of Houston and all the big channels are controlled by the Harris County flood control district which is one of the main ones is Braze Bayou just to give a sense of what's going on in Houston Braze Bayou is the channel that flows one of the main flow systems in the city to lower the level of Braze Bayou by one inch costs you $20 million so there's a project for $120 million just to lower Braze Bayou by six inches so Houston believes in engineering I'll tell you that they've got a so-called flood czar and he's an engineering he's an engineering solution the question of what should be done and where it should be done is something that I think should be better informed than it is now and so if we were to follow the path suggested by Dr. Sinhard then we should be planning out where is this development going to be for each new subdivision what will be the local drainage pattern in that subdivision and also what's going to be the effect of that coming downstream and I don't think that's probably taken account of now and even in Houston itself you've got Houston here and then you've got all the outlying suburbs you've got Katie, Sugarland and all the suburbs out to the west whose flow is now coming into Houston so Houston's got what they call the expanding floodplain problem so 20 years ago floodplain was like this and 10 years ago it was like this and now it's like this because the areas upstream are being developed and they're increasing the risk downstream that's what happened in the Myerland area that flooded really badly so I feel like what we're doing could be used to better inform people so that they can make both for development and for individual land purchase make more informed decisions in the future and I think that's probably how it's going to happen in our state I don't see the state starting to reign in local control too soon Hi thanks for a great talk so I want to go back to your comment on the of the chart data point and my question is I guess a classic question we know that large events like Harvey are more probable than we previously thought the heavy-tailed things but assuming we knew that for sure we had complete knowledge of the distribution and we knew what the worst case scenario may be what is the economics of it at what point do we decide this is too expensive to plan for this risk and what are your thoughts on that so yeah well they say Harvey was not completely unprecedented it was equivalent to the probable maximum precipitation which is the largest conceivable storm our state's updated our probable maximum precipitation estimates and Harvey's right at them it wasn't like it was inconceivable it just never happened before I don't I my sense is that often solutions are easily implemented actually what's happening in Houston is they're elevating houses I mean they're building houses up on a mound you drive up to your house cause it's on a mound not that hard to put it another if you're gonna do that anyway just make it a foot higher I mean for goodness sakes it may not even be that expensive if it's done ahead of time I think there are some solutions like that that could be implemented even for an anticipation of largely of extreme and improbable events probably if the landowners knew and had that choice to make for themselves that's the choice that they would make actually they would plan for something right out there even if the municipal people were not doing that more to the point also I think is we need to understand what the implications of this is when we were at the state operations center there was an enormous gap between what we knew and what we needed to know I mean I was there and people were coming up to me and I was saying should we stage out of Sealy should we stage out of Conroe should I stage out of San Antonio and fly people and just have air lifts the first couple of days it was just chaos and look what we needed was a water map across the whole landscape so that we could judge which areas were going to be flooded or were already flooded and which were had road access and so in retrospect we should have been able to do the hydrology and hydraulics to anticipate a water map across the landscape and see the situation as it really was and then we could have informed the decisions that were being made I mean they deployed a lot of resources to Katie a place in Katie then the water just came up it's now in Ireland and they're all flooded so all the rescuers had to be rescued which is crazy so I think we have to start thinking about stochastic storm simulation and the Harries that come on shore and thinking about beyond the range of what was being recorded at historical gauges and also thinking about design rainfall over an area we've always had depth, duration, frequency that's been a hydrology mantra forever but we've also got for depth, area, duration like I was showing we're not depth, area, duration and frequency make the whole thing into a four dimensional thing not a three dimensional thing and look over the last few years the perception is hardened into a belief that the sea level is rising there's not even questioned anymore and it wouldn't surprise me in a few years time that the idea that these historical storms are intensifying is also going to harden into a belief and we better start preparing for that possibility I think alright if there are no more questions thank you all for coming one more round of applause for our speaker David Mainman and I just want to let you know that if you have any further questions there is a small reception plan in Wood Commons and Hampton Hall immediately after the seminar so if you have a chance you can wander over there and have a chance to interact with him a little further thank you all very much