 Great. Thanks. Well, it's wonderful to be here this morning, and I really appreciate being asked to come and talk to you today about some of the work that we've been doing and as Balash said, he's been involved in this work, as has had many people. A few of them are listed here. I'm going to be talking about some global work that we've been doing, some global watershed modeling work. This work has really been looking at how watersheds and land use changes in watersheds are affecting the delivery of nutrients, nitrogen, phosphorus, carbon, and silica to coastal systems by rivers. But before I sort of go into what the modeling work is and some information about the spatial and temporal scales, I want to give you a little bit of background on the motivation for this. And the motivation is as mentioned that humans are really changing our landscape, both by the food production in watersheds, the energy production, the high population. I understand that the world population reached seven billion this week. And as a result of those activities and others, there's increased inputs of nitrogen, phosphorus, increased mobilization of carbon, silica, that are entering our rivers. And a portion of that is being transported downstream to our coastal systems. Once in the coastal system, those nutrients can lead to a large range of environmental problems, including algal blooms, some of which are toxic to ecosystems or to humans, as well as decreases in oxygen, resulting in hypoxic zones and anoxic zones in our coastal systems. So this work was really motivated by trying to understand how the changes in watersheds, including changes in hydrology, are changing nutrient inputs to coastal zones so that we can better understand where the source of the problems in our coastal zones are coming from. So now a little bit about the model, just very briefly. It is a global model. It's a global watershed model. And it's called news, global news, nutrient export from watersheds. And we can think of the model as sort of having three components, four components. There are nutrient sources in each of the watersheds. There are various hydrological and physical factors that modify the transfer of nutrients from the land to the river system. And then once in the river, there's in-river processing. The hydrological component of this is from STN 30, delineating over 5,000 watersheds globally. So just a little bit of detail about the components of the model. We have nutrient sources at half degree by half degree for each of those watersheds from both natural sources and from a range of anthropogenic sources. Those anthropogenic sources, many of them are related to agriculture, also to energy production. That's the atmospheric nitrogen deposition coming from NOx emissions from fossil fuel combustion, as well as sewage inputs. I'll say a little bit more about that in a minute, but the hydrological and physical factors then are responsible for moving a portion of these inputs in the water on the landscape into the river system. And then once in the river, there are a range of processes, biological, mainly, processes within the river that remove a portion of those elements, including removal in reservoirs behind dams. The output then of the model is nitrogen, phosphorus, carbon and silica, loading two coastal waters. While many of you probably aren't nutrient biogeochemists, maybe some of you are. So we're not only modeling the export of nitrogen, phosphorus, carbon and silica elements, but the different forms of those. So the dissolved inorganic forms, the dissolved organic and the particulate forms. Today, I'm going to talk a little bit about the dissolved inorganic, which think of it as ammonia and nitrate, very bioavailable, very reactive for biological systems. Dissolved inorganic phosphorus, phosphate, again very reactive, and the dissolved organic carbon. I'll talk a little bit about that as well. But the model is doing all of these elements and forms. Another thing to remember is that the output of the model, main output, is the export of these at the mouth of the river as it enters the coastal zone. So I'm not going to go into the model equations. I've just listed a very small portion of it to let you know that basically it's linear equations, relatively simple. Relating the inputs to the landscape, to inputs to the river system, and then export at the river mouth. If anyone is interested in the details of it, all the model equations and the explanations of it are in a paper by May Orga that I'll show you at the end, the citation for that. So today I'm going to talk about a number of applications of the model. First the global application, then an application at smaller scales, one into the Yangtze River watershed, and then some sub-basin applications. And then if we have time, I'll say a little bit about some scenarios for future conditions that we've run. So first the basics, the global watershed application. The inputs are at half a degree by half a degree. The output is for the whole watershed, so the export at the river mouth itself. So while the inputs are all at half a degree by half a degree, most of those inputs into the watershed are then averaged for the model calculations, going into the model calculations to calculate an output from the watershed as a whole. And you'll see this in a few minutes in some of the output. The run for this is basically very early 2000 year, not that explicitly 2005 exactly, of course, because we don't have all the inputs for an exact year, but basically early 2000, and as I said there are over 5000 watersheds in this. So the time step is annual average. So why annual? Well the basic reason is that while we would really like to have much higher temporal resolution than that, seasonal, daily, even shorter time, the constraining factor really is that the input databases were not available. Most of them were not available at all. We had to create them. And the information that we had to create those was not really of the kind that we could do seasonal inputs. For example, while of course global watersheds don't change on timescales that we're dealing with, and we do have some information on changes at shorter timescales and annual on water runoff and precipitation intensity, for the other types of information that we needed, that just was not information available at lower than annual scale. For example, for the fertilizer use, we had to use FAO data which is reported at the national scale annually. We then took land use to distribute that fertilizer use as well as the different crop types within that grid, and the fertilizer use by crop type to calculate the input for that particular data layer. So there was not information available then for how fertilizer use varied, for example, over the course of the year, especially when you're looking at the global scale. Similarly, we didn't have information on, not enough information to understand how in river processes, biological processes globally change at seasonal scales or shorter timescales. But also very importantly, and overall at the global scale, there was not data to look at, to either calibrate in certain cases or validate the model output. Most of the river loading data, when you look at a global picture, the data available is really usually reported just on an annual basis, the annual loading. And this gives you some idea of the amount of information that was available for global rivers on inorganic nitrogen export, for example. You can see there's quite a few rivers, but nowhere near, you know, anywhere near the 5,000 watersheds that we were dealing with, and similarly for dissolved organic carbon. But this also shows you how the model output and the measured compare. So quite, quite favorably, and also we had data over a fairly wide range of export rates, as well as geographic distribution, all the way from the Mississippi, the Ganges, the Rhine, down to some rivers in the Arctic region. This data, and for almost all of these rivers, was also only available for an annual export for comparison with the model. And this just shows you the model output at the global scale. And this is really the first global picture of nitrogen, phosphorus, carbon, and silica export by rivers globally. And this brings you back also to the point of spatial scale. As I said, while the input was at the half a degree level, for most of the parameters, we averaged that then for the watershed to do the calculation. And so the output here is the average output that gets to the river mouth averaged over the whole watershed as a yield. So it would be the average inorganic nitrogen export per kilometer squared of watershed per year. And as you can see, there are many hot spots both in Europe and in the eastern portion of the US and Central America, but also very importantly in Asia. And I'll come back to that towards the end of the talk. But it really gives us a global picture of what the current conditions are with respect to, in this case, inorganic nitrogen export on an average watershed basis. This is called a yield. So now I want to move on to another application. So we developed this model to give us an annual sort of average for some sort of decadal range originally. But we also wanted to see if we had input data on an annual basis, how good would the model compare to the measured export by a river? And very fortunately, one of the people working with us, Yan, from China, was able to compile all the inputs we needed for the Yangtze River Basin at the province level over a 30-year period, so annually for a 30-year period. So what we did is we were able to take that data, which he compiled for each of the provinces for every year between 1970 and 2003. This is some of the data you can see. You can see the changes in the big increase in nitrogen input to the watershed. And then run the model just for the Yangtze and compare it to the measured export of the Yangtze River. And we were quite pleasantly surprised to see that even doing these calculations annually over this 30-year period, that the model I think captured very well the inner annual variability, the overall trend, although before 1985 there's clearly a difference between the measured and the model. An offset of the model predicted being greater than the measured. We're now trying to go in and understand more about why those differences happened. Is it different, you know, maybe the input data weren't as good before 1985? There have been obviously some big changes in China around about that time and compared to after that time. But basically we were quite surprised and happy. There was no change in the model configuration at all that we had used at the global scale, simply applying it and running the model annually. So what I've been talking about so far is, as I said, just the output at the mouth of the river averaging the inputs over the whole watershed to do the calculation. Of course, we also would like to see within basin variability, within basin differences in transport of elements. So John Harrison was able to run the dissolved inorganic phosphorus model at a half a degree scale for, again, early 2000 conditions. The difference here was that he took the same input data at a half a degree scale. But using a river routing was able to capture the upstream inputs for each grid of dissolved inorganic phosphorus coming down the river, plus the inputs at that grid, and then do the calculation for many different, for each about half a degree grid globally. And I'll show you the output for that. And that's John up here in the left-hand corner. Has been very important in the whole global news activities. So he spent a lot of time getting data globally for dissolved inorganic phosphorus at subbasin scales. And you can sort of see, hopefully, some of the information he was able to capture. But you can also see that most of the data still is at the river mouth. It's very hard to get global coverage or anywhere near global coverage for within basin information. But when he did compare the model output run at that smaller scale with the measured, again, there's a fairly good relationship there between the measured and the model output. So, again, taking the same model configuration that was developed for a whole watershed calculation and using it at finer scale resolution did reasonably well. This shows the comparison, then, of the global application at average watershed scale at the top, with the blue colors again being low yield and the yellow and redder colors being higher yield, to the application at the half a degree scale on the bottom here, so subbasin scale. And as you might expect, there's a considerable similarity, but much higher resolution and information at a smaller scale. So this was a very interesting, I think, quite successful application of the model at that much smaller scale. So, again, here we see most of the hot spots being in eastern U.S., a little bit on the western U.S., a little bit in central Asia, certainly Europe, and, again, southern and eastern Asia. He also has applied the dissolved organic carbon model at small watersheds in central California, using inputs at one kilometer squared resolution. So even a smaller scale application. About 14 watersheds in the central valley of California shown here, where there were inputs available from a variety of sources at high resolution, comparing, again, the modeled DOC output with the measured. Again, a reasonably good comparison there. And, again, it was the same model that we had used, the same configuration we used for the global scale calculations, but now used with a much higher resolution input information. So, finally, then, in the last few minutes, how many minutes do I have left, Balash? A couple minutes? Five minutes? Okay. I wanted to show you another reason that we actually developed this model, was to explore what some potential future conditions might look like, and how nutrient export around the world might vary under a range of different scenarios. So, in this case, what we did is, and I'll explain it a little bit in a minute, we used the same, you know, approach that we used for the first application, the global application, input at a half a degree, average output for the watersheds. And in this case now, we developed inputs under different scenarios for the year 2020-30 and 2050. I'll just talk a little bit about the 2030 now. But the idea was that we wanted to help, we wanted to explore how different policy options and costs in watersheds to control nutrients and nutrient use would translate into the amount of nitrogen and phosphorus used, produced by food production and energy production in watersheds globally, how that then would translate into the nutrient export, and eventually to be able to look at coastal ecosystem effects and how those might be changing. So, what we used were the millennium assessment and millennium ecosystem assessment storylines to develop input databases that were consistent with the storylines. I'm going to talk, going to talk about two of those scenarios, the global orchestration and adapting mosaics, sort of funny names, but that's what they call them. The point being is that the global orchestration you can think of as being a very reactive approach to environmental problems with the world being very globalized in terms of decision making, in terms of trade, et cetera. And in sort of a very contrasting scenario, adapting mosaic is one in which there's a real proactive approach to addressing environmental problems using nutrient, very good nutrient management on the landscape, et cetera, and a more regionalized approach to making decisions about various transactions. So, I'm going to call the global orchestration so it's easy to remember a business as usual, because really it's kind of the trajectories that we're on now. And the adapting mosaic you can think of as a better case scenario. Now, all of the assumptions that went into this include a wide range of social, economic policy and ecological considerations. And of course we had to develop the input databases, gridded input databases for these scenarios for these years. So it was a huge effort in developing these input databases. I'm just going to give you a little bit of a flavor of what the difference in these two is. Take the, I guess I call it the worst case scenario, but business as usual in some cases. The fertilizer use is high, still more like it is in the US. Nutrient management on the landscape is not really optimized. Meat consumption is high similar to what we, in many regions of the world, similar to what we have in the US. In other words, you might think of the southern and eastern Asia, which has a low meat consumption now under this scenario is growing and increasing more towards what we have. And access to sewage treatment is complete, full. So in other words it's all going into pipes and discharged into water systems basically. Under the better case scenario, moderate use of fertilizer, a very efficient nutrient management on the landscape, a much more moderate meat consumption, etc. But the thing to remember is that each country had different trajectories for these things. We had to consider what the current condition is now and how that might change in the future. So the specifics in other words vary by country. And this next slide is, I want you to just get a general impression of how we developed these input databases. Basically we used the image model, the integrated assessment model to develop these inputs, taking into consideration the potential, the expected distribution of population and what their food demands or needs would be, considering the different options there of how much meat, etc., they would be eating. Then that was taken, also taken into consideration is the available land for growing that food, the available water for irrigation as needed. Depending on the amount of meat consumption and the assumptions there, the types and number of animals that had to be raised were considered. That of course results in manure production and of course the amount of meat production also has big implications for the amount of crops that you have to grow to feed those animals, in addition to the crops that are eaten directly by humans. And there's a climate model behind this, fairly simple climate model and also an energy model because there's biofuel production as well. But this basically is just to give you an idea of the different, actually model components, additional model components that were used to develop the input databases so that we could explore different policy options. So just one graph of output showing the results. This for inorganic nitrogen now is mapped out as the change in inorganic nitrogen export by rivers for Asia and Africa and Europe. And you can see under the global orchestration which is the very, you might think of it as aggressive or not very environmentally optimized scenario, there's a, I'm going to focus on southern and eastern Asia right now, is that you see that under sort of a business as usual scenario, business as usual being that the current trajectories are basically continued. That there's a large increase would be expected in this case nitrogen export by rivers in this area. Already a hotspot under current conditions but a very much larger increase over the next, well this was over a 30 year, next 30 years. And this compares then or contrasts to the scenario in which there was much more efficiency of nutrient use on the landscape, changes in technology, etc. And in that case the changes in export are much smaller than under the global orchestration scenario and in some cases even a decrease in nitrogen export. So this just gives us some idea of what some of maybe possibilities are, a way to explore some future possible conditions. And then we can also back out of the model what the improvements in nutrient use efficiency and technology that led to these differences in the two scenarios. Really mainly increased fertilizer management on the landscape, lower meat consumption and a little bit from sewage increased efficiencies but primarily resulting from the agricultural sector. So I guess I've tried to show you then, I hope I've showed you and you've seen how we've developed this global model for looking at how land use in watersheds globally is affecting nutrient transport to coastal systems, how the model works under in a very specific watershed over annually over a 30 year period. It's application at sub-basin scales and also how we're using it now to explore some different scenarios of possible future conditions. We've also run it for 1970 as well so looking in some hind casting. So if anybody's interested in more details on the model equations and explanations, Mayorga at all paper, Emilio Mayorga's paper in environmental modeling and software can give you a lot of details. The input databases and a lot of the information that I talked about today is also in a special issue of global biogeochemical cycles published last year. So thank you. I'm really curious about what's in the model but I know we can't get into that. But the last time I saw a diagram like the wiring diagram like that I suppose was a long time ago in the population models that came out, oh geez what in the 60s I suppose. And you mentioned that at least in the nitrogen case the equations are linear and so you're presumably what inverting a whole system of linear algebraic equations to time step forward. Can you say more, I guess my question is can you say more about modeling aspects and where you see the critical problems? The critical problems. Let's see first of all the wiring diagram was only a wiring diagram of the image model which is in the global news model. That's an integrated assessment model and it was just to try and show you some of the different components. And that was what was used to develop just the input databases for the scenario applications for 2030 and 2050. But back to your question about the news model itself. Some of the challenges and some of the what we would like to see as improvements in the future would be more of a dynamic modeling approach where we can look at time lags in the system where we can look at for example smaller time steps because in reality the effects in the coastal zone don't happen on an average annual scale. The biology is very fast and responsive and so we would certainly like to be able to look at more seasonal variation. We have been doing some seasonal variation modeling now exploration for the inorganic nitrogen and phosphorus and John says that's going quite well but we don't have the publications from that yet. So I would say that more of a dynamic system would be very useful and actually what we also really need is more data to see how well the model is doing. We can do all sorts of you know much more detailed calculations and time varying but unless we have the measured data in rivers to compare it to then we're on very thin ground. So that's one of the things that we certainly would like to have more of most of that type of information is only available for the US so a very very small portion of the world and under very limited conditions compared to what we see in other regions. So those are some of the things that we'd like to see improvements on. Very nice talk. I had no idea this was even going on this is great so I'm happy this is not part of my domain. A couple of questions one is there are similar types of development of models of usage of these kinds of nutrients in the ecosystems that are going to be affected by coastlines and the second one is that are there plans or do you have access to the data for catastrophic events that may transport more of these nutrients all at once and cause the kinds of things you were talking about that the model doesn't quite get to. So thank you for those questions. One of the things that the news working group and it's about 30 people are working on now is being able to trying to quantitatively relate nutrient inputs to coastal ecosystem effects also within rivers or within the reservoirs or lakes within the river system as well. That's a huge challenge because it really you have to take into consideration not only the nutrient inputs but the physical and hydro hydrological hydrographic conditions in the coastal zone as well and so there's progress being made on that but we're not there yet and it also involves a different suite of people as you might expect then we're in to some extent that that we're involved in the the river modeling work itself but a very important point and in fact that was the reason for doing this to begin with and then your second question was oh catastrophic events yeah we would love to be able to to do to get at that one of the problems or challenges there again goes back to usually during those catastrophic events there are not measurements to capture those events to see how well even a model might be doing but we should not let that stop us in that there is some information for some rivers on that and again if one was trying to just optimize the modeling for one particular watershed then that you can do that but we're really trying to get a global picture or at least a very large-scale regional picture but that's a very important thing to do absolutely