 First of all, I want to thank you all for hanging on to the bitter end today. I know some of you are probably fighting jet lag and probably the temptation to go and have a beer, which is very much on my mind, too. Anyway, so my task is to start off with giving you the climate basis for this study. In particular, what I'm going to hopefully get you through is not only the technical details, but hopefully at a very superficial level, but also just sort of the philosophical approach that we've used to take what you could regard as a climate model or an earth system model and reduce it down into the level of detail that we feel is necessary to make it computationally efficient so that we can run this model through a number of different, what we would call plausible outcomes, which then gets translated into risk and the impacts that you'll see in just a second. So again, I am merely a representative here of a larger team of scientists, two sitting up here, three sitting up here, excuse me, as well as a number of other folks that I'm privileged to work with and privileged to give you the results of today. James did a very good job of providing the impetus and segue into this work, and it really boils down to the one hand and the other hand, and really it does come down to when we look at climate changes, for example, these two maps here, which are just showing precipitation changes from two climate models that you're probably all very familiar with when we talk about the IPCC. When you take a step back and look at this, you might think, well, it's getting wetter in the tropics and this seems to be drying in the extra tropics and it all seems like it's consistent. The problem is that when you get down to regional details, you start seeing different signs. You see precipitation going up in one place and in that same place in the different model, you see it going down. And so how do we actually wrap ourselves around this problem and give these climate outcomes in a way that's useful to impact an economic assessment and even to guide strategy? So to do that, I'm going to start with some philosophical thoughts here. And the first one is probably one that you've heard a lot about, heard many times, maybe said in different ways. But the famous quote that all models are wrong, but maybe some are useful. Well, I think as we've all talked about in many of the parallel sessions today, it's probably more a case of let's just accept the fact that no model will ever be perfect. We will never have a perfect model. We'll never have a perfect climate model, particularly for the work that we do. But many of them are suitable. Many of them are not misleading. And that's probably the key issue here, is that we have at our disposal a number of models that are useful, that are not misleading, that are insightful. And we need to glean as much information as we can from these models. So if we have a suite of models, what do we do with them? How do we pick one model over the other? And there's been a lot of work up until now that's tried to look at models and say, well, if this model is more skillful at doing something than this model, we should trust this model more. Well, as it turns out, there's actually a growing body of evidence more recently that shows that that sort of metric doesn't really always apply. What I mean is that you could pick over a certain region for a certain given time, a certain variable, say temperature. And you find that one model does a great job of modeling temperature for that particular region for that particular time over another. And if you pick a different time period, that association or that relative skill can flip, could go the exact opposite. The model that was once worse now becomes the better model. And so how do we sort of take that result philosophically? Well, for our purposes, we're going to basically say that we are considering all plausible outcomes. We can't really say for sure whether one model can be favored over another model outcome. But we at least have to recognize that both of them are plausible. Okay, finally, so if we have all these models, we say to ourselves, well, let's just assume that they're equally plausible. The temptation to doing impact studies is to say, well, let's just take the mean of all these results and run it through our impact models. Well, there's a caveat to that. When you take a mean of all different models, you smooth out some of the extreme results. Those are actually the most impactful and probably the most insightful and could be because they're equally plausible the most costly. So rather than taking a model mean approach to climate input data, we would like to take the distribution of those outcomes and feed them through the impact model. Now what happens is that you then end up with this wealth of information, almost an overload of climate model information. We have to figure out a way to whittle it down to the important details. And this is in essence our approach and this is what we're trying to get at. So we've talked about the fact that models aren't perfect. We can't really precisely predict the response of the atmosphere, the ocean, the land, what makes up our earth system precisely. We know we can't really do that. What we can do is use observations to help us guide where we think some of the responses are probably unlikely, in a statistical sense. And I'll get to that in just a second. The third issue has nothing to do with climate, but it has to do with the fact that we have to somehow predict emissions. How many of these important trace gases or what we call greenhouse gases? How much of these emissions are we going to emit? And we have to do it in a way that's consistent with our model framework, okay? We can't just throw emissions in willy-nilly. We have to have some economic basis to these emissions. And some numerical basis for the uncertainty of those emissions. So you put all that together and you get to this rationalization that we cannot base any sort of assessment or interpretation or impact study on one prediction. And I use the term prediction very loosely, very loosely, okay? We can't trust a single prediction. So what we have to do is perform as many simulations as we can, that we know span the ranges of uncertainty that we know exist. And we can constrain through observations, all right? And do that in a way that both takes in the economics of the problem, the emissions of the problem, and the earth system aspects of the problem. And this is the tool that we use, okay? It's not the only tool, it is a tool, but it's a tool that we've employed. And we use it because it incorporates a lot of the elements that I've just spoken about, okay? It's what we refer to as the integrated global system model. And I want you to think of more of this as not just a model but a framework. There are many different aspects to what's going on here. Depending on what question we're asking, we may put more detail in one of these elements as opposed to another, okay? So it's a very flexible framework. We don't, we recognize that we have certain computational limits that we hit very hard when we have to run these computer simulations. We have to make decisions about what level of detail we put in. And so the model itself consists of a global, computable, general equilibrium model of the global economy. It's a multi-sector model. And I could spend my entire 20 minutes explaining the model. But really in the context of this discussion, that last, the rightmost arrow is our most important arrow for this discussion, which is emissions, all right? The economic model has a framework to span a range of uncertainties in what we think the economy is going to do over the course of the next few decades, based on some policy. And it provides emissions. And we use those emissions in our Earth system model, which we regard as an Earth system model of intermediate complexity, an important qualifier, intermediate complexity. Not the full complexity, but intermediate complexity. Elements that we know are important that we need so we're not misleading ourselves when we look into the future, all right? So these two models work together to provide outcomes of what we think is going to happen in the next few decades. This just gives you a summary of what the human system models, the EPA model, emissions prediction, emissions policy and prediction analysis model does. There are some important uncertain parameters in the model. I can't go through all of them, but you see a few of the big ones there. The main output, all right, again, is emissions for the Earth system model. For the work that Ken and Channing will talk about, some of the other important issues or outputs from the model are prices. So the global prices from the economic model actually getting fed into this regional economic model that Channing will talk about in just a second. This is trying to summarize everything that we think we know, or at least we think we're uncertain about for regards to the Earth system model. And there's the key bullets here. Couple of things just to sort of keep in mind when thinking about all this. Probably one of the main uncertain issues that you've probably heard a lot about, at least in the media, is this issue of climate sensitivity, all right? You put a certain amount of trace gases in the atmosphere, it will produce a certain amount of global temperature response, all right? That's what we think, that's in essence what climate sensitivity is, all right? And that in and of itself is a very hard thing to observe, let alone to model, all right? There are many different ways of getting climate sensitivity. We know that it's uncertain. We've done studies such as on the right here by Chris Forrest and co-authors that can give us a range of what we think the acceptable values of climate sensitivity are. In tandem with other issues of the climate system, which is the response of the ocean, the impact of aerosol forcing, the impact of or the response of terrestrial ecosystems to an increase in CO2, which we call the CO2 fertilization effect. Plants may become more productive if the ambient CO2 concentration gets higher. That actually has some bit of uncertainty to it, which we factor in. And there's also something that we use that looks at the precipitation frequency trends, but it's not directly relevant to this study. So what do we do? We take this model, we know that emissions are uncertain, okay? We also know that the climate's response is uncertain, and we run scenarios. We just run a range of scenarios. Now there's a lot of work, the two papers that you see up there that have presented these results in the peer reviewed literature. And what you're looking at the right here are just examples of the types of outputs that we look at. The key thing here is that we look at this in a distributional sense, all right? So what we're looking at here on the right are frequency distributions of temperature outcomes on the top, under these various policy scenarios, and they're kind of cryptic, but I'll get to that in just a second. The black one that you see is what we refer to as our reference scenario. Some people call it no policy, but then argue the peak or argue well, it's not, there is a policy, you're not doing anything. So we try to sort of view this as our unconstrained emissions scenario, okay? But there is a policy, the policy is we decide to do nothing. And then there are a range of stabilization scenarios. One through four, okay? Level one, which is the green curve on both of those panels, is our strictest stabilization scenario. And so basically what we do is we tell the model that it has to stabilize CO2 equivalent concentrations by the end of this century to about 560 parts per million. The other curves that you see there, blue, orange, red, are just sort of relaxing that stabilization scenario. By about 110 parts per million, CO2 equivalent increase per level of stabilization that you see. So from here and in slides that follow, you're gonna hear jargon like in the level two stabilization you heard this and the level one stabilization you heard we get that. And that's what this is referring to here is that for a given target of stabilization of the CO2 equivalent gas concentrations by the end of the century, that's how we're gauging our levels here. So level one again is 560 parts per million CO2 equivalent by the end of the century. I might point out that right now we're at 480 or thereabouts. So we really don't have a lot of room to get there. But we have a lot to do to actually achieve that stabilization if we think that that's possible. Level two is a 650 parts per million CO2 equivalent. I'm gonna show you results from the level two stabilization and the unconstrained emissions scenario. Channing's gonna focus on the level one stabilization because economically there's sort of an interesting feature in there that you can glean some results off of. But that's basically our scenarios. From those frequency distributions, somebody mentioned before how it's very difficult for scientists to convey important results. This is one of the things that we do from these frequency distributions is we actually turn them into roulette wheels. And we ask people, well, if you don't wanna do anything about climate, this is the wheel that you have to spin for some warming to occur, global warming to occur by the end of the century. Well, if you embrace or adopt a sort of a moderate stabilization policy, the level two policy, you get to spin that wheel. Which wheel do you wanna spin? It's a gamble, it's a risk. Which we found this is a very useful and effective way to sort of convey to the general public what's at stake here, what are the odds? The problem is, not problem, I shouldn't say that. I strike that. The issue is that when we think about a regional climate change, those global wheels don't really apply. It's not really what we're trying to get at here. We're really trying to get down to the regional information. I mentioned before that the IGSM is an intermediate complexity model. And what that means, among other things, is that the atmosphere only has, it can only describe the atmosphere in a latitude sense. We've removed the longitudinal detail of the model. Simply because we have a computational constraint that we have to meet. And so we've removed that level of detail. So if we look at the atmospheric output from the model, this is looking at precipitation change from increases in CO2 concentration. All we can really do directly from the model is if this is the Zambezi, this is where the Zambezi sits, the Zambezi River, which is our focus of our study. These are the latitude bands that the model is able to explicitly tell you something about. And that really doesn't get down to what we really wanna get at either. It tells us something interesting that if we were to look at precipitation changes over those latitude bands, we could see both increases and decreases in precipitation. We're not precisely sure what it's going to be, but in a distributional sense, you can see that the central tendencies here sometimes lie below zero, so we might see a drawing, or it may be above zero. So how do we take that information and get it to a point where it actually geographically is something that's useful and we can actually, is palatable to these impact models? And this is a study here, again, I don't have time to go through the whole study. It's a way to numerically fuse that latitude new detail with information from climate models from the IPCC exercises. And so this is the only equation that you'll see here, okay? And really all it's showing you is that we've taken the zonal information from the IGSM's atmospheric model. We've constructed a climatological downscaling to the regional detail, and then we've used the information from climate change experiments to put together what we call these regional climate change kernels. And all it's saying is that for a given unit of temperature change, it could be cooling, it could be warming, these are the shifts in the patterns that we would expect in a linear expansion, right? This equation here is essentially a Taylor expansion and it's truncated to a linear expansion. I can talk about the nonlinear stuff later if we need to. But for our purposes, we found this very effective, particularly over the next few decades, to at least glean out from climate models what we think is most salient and insightful when thinking about this whole issue in a sort of a probabilistic or a statistical sense. Okay, so some of the maps here that you're gonna see are just sort of showing you that, yes, we can construct these climatological downscalings of the climate for South Africa. So this is just showing you from data that we can get, you can get just about anywhere, temperature variations over Southern Africa for its summer and its winter time. Not surprising, it gets warm in the summer and it gets cool in the winter. Hey, not surprising. What is probably worth noting though is that hopefully you can see it in the back of the room there is that I've outlined two boxes. One is for the Western Zambiae Basin and the other one's for the Eastern Zambiae Basin. And what is interesting to see is that the seasonality in the Western Zambiae Basin is a lot greater than the Eastern Zambiae Basin. And that'll be kind of interesting when we talk about what we see in the way of shifts in the future. What you're seeing here is just again the similar type of plot, but for precipitation. Important thing to point out here is that during the summertime, there's a lot of precipitation that falls in the northern flank of this basin. It's a lot, okay? And in the winter time, you almost see a complete reversal. Very dry conditions in the winter. And important transitions periods between the winter and the summer, the spring period, which probably very important for planting and getting ready for the growing season. There's some very important transitions that play out particularly when we think about changes in these quantities into the future. Okay, so here are climate change kernels, okay? I'm running out of time, so all that's showing you is that again, from all this information from the climate models, these are the different patterns and temperature shifts that we would expect for a given unit of warming, okay? Mostly what you see is that over the land, you see a stronger amount of warming than over the ocean, okay? That's a very well understood problem and feature of most climate models, and it's encouraging that we see it here, okay? But the center of action tends to float around. So for one model, you may see a lot stronger warming over one location than another. And this plays out with the agricultural impacts and many other things that we'll get into in just a second. Similarly for precipitation, and you might say just by looking at this, there's a lot more texture and a lot more variety in the changes in precipitation between model to model. And again, this approach is to try to embrace all of that, get it all in together, normalize it and synthesize it and fuse it with the IJSM information. So what do we get? Here is a result for the Western Zambezi Basin, and what you're looking at, changes in temperature by the middle of this century for the unconstrained emissions case and the level two stabilization, okay? And one thing to point out is that everything in the abscissa is greater than zero. So everything's warming here, everything. For the unconstrained emissions scenario, what we're seeing is a distribution with a mode of about two degrees, okay? And very, very small chances of warming on the order of about four degrees Kelvin and four degrees Celsius, okay? Pretty big temperature increase. What's encouraging is that for the level two stabilization, what we essentially do is to take this unconstrained emissions result and move the entire distribution to the lower half of the unconstrained emissions scenario. So everything to the right side of the central tendency of the unconstrained emissions distribution is essentially not possible in the level two stabilization scenario. There's a substantial decrease in the mode of the distribution and also a shrinking of the range of that distribution. You see, tend to see more subtle effects for other months and for other regions. What you're seeing here is the Eastern Zambezi result and because again of the fact that the Eastern Zambezi seems to have not as strong a seasonality and the climate seems to be a bit more constant and a bit more stable in terms of the responses that we're seeing. What you see here is a little bit more subtle, okay? You don't see that big shift in the mode but what you do see is a change in the skewness of the distribution. To the right of this mode of the distribution, you see a pretty large bar that goes from 1.5 to 1.75 degrees. In the level two stabilization, even though the mode hasn't shifted much, what's very interesting is that this right hand side of the distribution has now been shifted to the lower half of the level two stabilization scenario. So there's been an important shift in the skewness of that distribution and the question is, is it important? Does it mean anything? These next two results are just simply for the precipitation changes. This is probably the most striking result that you see which is for the Western Zambezi for springtime conditions and for the unconstrained emission scenario, you see a very broad distribution crosses both sides, both positive and negative precipitation changes. What's interesting though is that the mode sits in a drying scenario. What happens in the level two stabilization scenario is that not only have you shrunk the distribution, you've also moved the mode to just about no precipitation change. And again, very compelling, very striking, is it important? You'll find out in just a second. And I've run out of time, so I'm gonna stop there and basically these bullets here are just sort of regurgitating what I've told you already. Again, the biggest features of this work, we've found at least on the climate side is that we see some very marked responses not only in the seasonality of the results but the modes and the skewness. And again, as you'll see in just a second, we'll find out what difference does anything make? There are some issues in this approach about linearity, the Taylor expansion approach that we use. We've gotten a lot of questions about, well, do you feel it's linear? The short answer to that is for temperature, we see no real concern in terms of its linearity of this expansion. Out to 2050, precipitation is fairly linear. When you go past 2050, you start to bring in features or model responses from the climate models that become non-linear. But fortunately for this work, we've only been looking at 2050. So we feel like we're okay with that. There's also a lot, a lot of interest in getting more details. We've used global climate models, which are very coarse. Well, relatively speaking, our course in the spatial detail. The good news is that this approach is very palatable to regional climate models that have greater spatial detail. But for this study, we're not going to get to that. But it's an area of active research. With that, thank you. And Ken, you're up.