 technical pieces of background information. Andy is the head of the GIS laboratory in Erie. Thank you everyone, I have a very impertinent request at the beginning of this presentation. I'm more than happy to share this presentation with anyone, so please save your camera batteries for how long they're in other sightseeing around Vietnam. I'm more than willing to give you a copy of this, so I'd be happy to do so in an effort to disseminate some of our preliminary results and also to get feedback from different groups on how we can try and improve this analysis. So just to give you an idea of what I'll be talking about, I think David and I can have introduced it quite well, trying to produce some quite rapid background assessments using spatial analysis to try and map out areas in Asia where rice is produced and the yield and the yield potentials. An awful lot of people have had an input into this process as well as the strategic assessment task force that David has mentioned. A lot of people have contributed a lot of time and effort into producing these preliminary assessments. So I'd just like to give acknowledgement to these people here. So this is the question that was posed by David. I joined Erie this time last year and this was one of the first tasks I was faced with, was if we look at the rice growing areas across Asia, what do we know about the cropping systems? What do we know about irrigation and rain fed and what do we think we know about the yields and what those yields may be in the future? And these are all questions that needed to be answered as kind of a background analysis for this assessment. So I'd split this into three separate tasks, determining where and how rice is cultivated in Asia, trying to get an idea of the yields, the actual yields, now and what they may look like in the future and then looking at rice potential yields and the climate change. So the first step, as David alluded to, is this whole concept of agri-ecologies. And we have an awful amount of, a tremendous amount of information on this by people like Hugh and Hugh and others who have collated statistical information and had a lot of expert roundtables and discussions on how rice is cultivated across Asia. And this is something that we hope we wanted to update. And after several round of discussions and several attempts to do this using other methods, we came back to this typology of what we wanted to try and use to describe the different rice systems across Asia. And we ended up with this eight class typology, if you like, of the different rice systems. Four of them are essentially irrigated and four of them are rain fed systems and we have everything from single season irrigated up to double and triple intensively irrigated rice systems. We have our rain fed systems again single and double. We can look at rice cropped with another crop and of course the upland category. So these are our eight typologies that we're using. And this has some relation to the data produced by Hugh and Hugh but it is still a slightly different way of looking at the agri-ecologies. And we had a lot of different data sources that we could use for this. And of course expert opinion has a huge amount of input into this as well but we also look back at all of the statistical publications out there and you can see that there's a lot of information for us to try and collate and to try and standardise to come up with a common measure of rice systems in Asia. So what we've done is to split Asia into 220 zones. We've gone down below the national level in terms of this assessment. And this splitting up of the region into zones is somehow a compromise between the available statistical information, the detail that we see in Hugh and Hugh and somehow related to the actual physical distribution of rice growing areas. And then we had to try and review and revise these estimates through an iterative process. Now this is the zones across Asia and you can see it's of course areas like India and China have much larger zones than other countries but in general we try to capture a sub-national concept of rice cultivation of rice areas. And this is the biggest issue we had at the beginning is that we have all the data on the left that comes in different formats and characterises the rice systems in different ways and it didn't really match the agroecology that we had conceived of as part of the round table discussions with the strategic assessment task force. And this has caused us to review and revise this several times to try and come up with how much rice has grown under each of these different agroecologies and this has been a major component of this background analysis. And this is what we came up with for Asia and at the bottom I've colour coded the different classifications, the different eight typologies and essentially the irrigated systems are in blue and the red systems are in the orange and red. And if you just look at Asia in general about 57% of the rice area is cultivated under irrigated conditions. And you can see there's quite an even split between the irrigated other, the double irrigated rice and the triple irrigated rice systems. When we break that down and look at what's happening in South Asia, East Asia and South East Asia you see very different trends. Evidently in East Asia it's predominantly irrigated systems. In South Asia it's more of a 50-50 split and in South East Asia rainfall is still slightly dominant. Of course we can break this down even further when going out into the country level. I don't want to spend too much time here, but this is one of the reasons why I'm really keen to disseminate this presentation after because we would really like to get a lot of feedback on this and how we can try and improve these figures. Now of course we can disaggregate this even more and go down to our zones and here if we look at the distribution I'm just going to give two examples here. This is the irrigated rice other system and each of these points represents about 25,000 hectares and you can quite clearly see some of the rice wheat areas in IGP and of course across China and other areas. Just to give you an idea of the kind of spatial detail where we're trying to pull together here across these different agroecologies. Another example will be the upland or dryland rice system and again you see very different distribution so this is what we're trying to pull together in this analysis. So that was the first component, trying to come up with a new system of defining rice agroecologies across Asia and then that would form the basis for a lot of the other work that was going on in the strategic assessment. The second component of the background analysis is looking at actual rice yields and trying to estimate what the yields are in 2010, the contemporary rice yields across Asia at the same level of spatial detail and then to look at the future yields. And if we are seeing that what we've seen over the last few years is going to continue a kind of business as usual trend that was really the only way we could conceive of kind of modelling actual yields so we're trying to extrapolate what we've seen in subnational statistics in the future. Of course extrapolating it out to 2010 is quite plausible extrapolating it out to 2035 is another question but it's really the only approach that we could conceive of to do this in a quite short space of time. So we went back to our rice time series data where area along with its partners has collected a huge amount of information over the years from different countries on rice production area and yield statistics and so we were able to go back sometimes 50 or 60 years to look at the trend in rice yield. And on the right we see the number of zones that we observed to each country. This is our kind of spatial disaggregation of this information. So how we go about doing this and this is just an example from Bangladesh looking back over 50 years of rice yields and we see a lot of fluctuation from year to year but we use our lowest method to smooth out that trend to just try and capture what has been the rice yield trend over the last 50 years. And then we use Arima forecasting models to extrapolate that out to 2010. Now if we're lucky we have current yields from maybe 2007, 2008, 2009 depending on how long it takes for a country to publish these official statistics so we still have to estimate 2010 and then we go ahead and estimate 2035. Just to give you an idea of what this looks like this is an example from Andhra Pradesh over 50 years and this is showing in the grey line the actual yield reported in the statistics from year to year. The black line is showing our smooth trend and then we have the extrapolation in the future in red and with standard errors in blue. Another example from Haryana is a very different pattern in the historical time series and the influence that change has on our projections out into the future. The standard errors become much wider because we've observed historical changes in the trend and that means we have less confidence in projecting what that trend might be in the future. And again starting at the continental level these are the yield estimates for 2010 in tonnes per hectare across all of Asia about four and a half tonnes per hectare. I've seen variations when you're looking South Asia and East Asia and South East Asia. Of course East Asia having again the higher yields. South Asia and South East Asia slightly less projecting out to 2035. We generally see an increase of about one tonne also in 25 years. And if we were to assume that the area of production stays constant that translates to a change in production of 630 million tonnes to 760 million tonnes in the future. Of course we can disaggregate that and go back to our 220 zones and look at the actual yield estimates for 2010. Again we see much higher yields in the North, particularly in Northern China and yields coming down to 2 to 3, 3 to 4, 4 to 5 tonnes per hectare in other parts of Asia. Moving on to 2035 generally seeing substantial yield gains in different areas. I'm just going to flick back and forth between these two just for a while so you can focus on a particular area and observe the change. Translating that into a changing yield to tonnes per hectare by extrapolating the trends that we observe we see the yield gains in most places. Some of them are substantial gains, 2 to 4 times, sometimes higher. Other places 0 to 1, 1 to 2 tonnes. That's quite a lot of spatial variation in what we've observed in the trend if we were to extrapolate that out. Now as I mentioned before in general, when we look at 2010 we seem to have plausible results but there is no forecasting methods in the world that is going to allow you to come up with reliable results at sub-national level 25 years in the future. We're just trying to do the best we can with the data and information that's available. But we feel that the method is quite robust and it's making the most of what we have. Now the final step in this process was to look at potential yields. And again David alluded to this it's part of the process of computing yield gap estimates at the same kind of detail. And the way we've approached this is to use a rise of 2,000 crop growth model and climate data in order to generate potential yield estimates for 2010 and 2035. Now the rise of 2,000 requires daily climate data from a range of variables, rainfall, temperature, radiation, vapor pressure, and wind speed. And the only way for us to generate that is to downscale some of the results from the global climate modeling exercises that are going on. Now downscaling, it's a very computer-intensive process where you take these global models which are very coarse spatial resolution and try to generate higher resolution climate data for your area of interest. And to do that we have to split Asia into three regions. Each region takes about three months of processing on some fairly powerful computing facilities and that's three months if everything goes well. There are a lot of technical issues in doing this. And it has been somewhat unfortunate that we know that many people are doing this exercise but there has been somehow an unwillingness to share the information which means that we've had to do a lot of this ourselves. Again, these are the zones that we've looked at across Asia where we split up into three zones for the climate modeling purposes. Now again, I want to acknowledge the Malaysia Electrological Department for sharing their data with us. It's been invaluable. We've just finished the run of South Asia here at Erie and the Hadley Met Center in the UK has been aiding us enormously with this and they're running the model for China and the rest of East Asia. And this is still ongoing. Now as we know there are many different global climate change models and many different scenarios and it's simply impossible to try and downscale all of them. It's just too much work. So we chose one model which is the Hadley CM3 model and we chose one scenario which is a kind of balanced emission scenario. As it turns out, since we're really looking at the 2035 that's not a big time scale in terms of climate change modeling. And most of the divergence in scenarios occurs after that date. So the choice of scenarios is actually not that critical in this case.