 I'm Chaz Fant and I am going to talk to you about surface water availability in Morocco and the main point, so I'm going to talk to you about surface water but I really want to more discuss on this subtopic which is the balancing uncertainty that's more the focus and I this is some work that I did with Alyssa McCluskey is here and Kin Stresbeck who's coming a little bit later by the way I'm a civil engineer should be clear about that so I'm not going to go through the water cycle I know you guys know how the water cycle works the most part but I just want to point out that I'm talking specifically about surface water so I I'm not talking about storage I'm not talking about aquifers surface water so there's snowmelt and and other things which I'll get into so I'm part of the team and in this team we we have this integrated framework and you'll actually see this later I stole this from my advisors presentation is presenting later so I won't go through in detail but actually he doesn't know that I stole it so don't tell him but I what I'm presenting is the the rivers the runoff stream flow section of this and I'm presenting this because this is a fairly straightforward simple model and then we can get it into the some of the uncertainties and how do we present this these sorts of things quick overview of it's Clarence who is the surface rainfall runoff model so what we like about this model is that the focus is on climate impact so it nudging the climate we end up getting this change in runoff so there's a calibration validation process to build the base model and for the future like I said we just basically nudge the climate and see what happens to the runoff fairly simple this is just a quick schematic again input climate output runoff the structure if you care about these sorts of things it's two layers there's quick runoff and and slow runoff and alright so now now to Morocco so this is our basin map and so we did this for this was funded by a group who will remain nameless and they they told us we have people who can go out and get the data so they went out and got the data and they came back and they give us 16 basins with with measured stream flow and the other 21 we have no idea but we're you know we have to we have to model this entire country so there are our theories and hydrology to deal with these sorts of things which we we use the best that we could the simplest but still this this does introduce quite a bit of uncertainty I mean we have no no idea what the the measured stream flow and a lot of these basins might be so then there's the this other idea of natural versus actual or measured stream flow so actual is it rains it hits the ground runoff goes into the river so that's that would be natural but actual stream flow is people you know people come to the river they take from the river they irrigate they do all sorts of things with that or they build dams and they control the stream flow and so what you end up getting is is measured which is most of the time tampered with by by people which is good and I'm not I'm not saying that's a bad thing but the model that we're using is specifically designed for natural stream flow so it has the the climate signal so this is another thing that introduces uncertainty in the in the historical data so then we get to the GCMs which I'm I'm not going to go into a full account of the uncertainty of the GCMs I assume you guys for the most part understand that they can't predict for example what weather will be like on January 21st of 2053 but they are a good they're great models and they design for a specific purpose so in this study we were given downscaled from the Hadley model downscaled results at point one degrees and so and this again was this was the funders were giving us this data and I assume that they did a pretty extensive study to find out which model is the best model then they did spent a lot more money on trying to downscale using the best method etc etc and we said okay from another project we have this other sweet we have this fit these 56 scenarios that we can we can use and it's easy once we build the model it's it's fairly simple to just add add more scenarios we can run them through so just a quick for those of you aren't so familiar with the GCM models GCMs are really best at predicting long-term global changes and specifically in temperature so once you move to precipitation which of course is really important for for runoff there's quite a bit more uncertainty and so we want we sort of want to stay as close to that obviously we were doing a project that's local and where time matters so we can't we can't use that specifically but we want to stay as close as we can to to that and this is this idea of balancing how do we balance that and and using you know these super downscaled you know up to daily results so what we do is we take GCM historical and because each each GCM runs a historical run we take the future GCM results and we take those and we we figure out what's the change because GCMs don't do a great job of predicting what has already happened what we know has happened so we end up with with this change in climate and we take that change in climate and in this case we had a 30-year measured climate and we put those together so we we nudge the the measured climate or shock it if you're an economist and then we end up with with this climate that we put into the model and I should say for the for the future we ended up with average decadal changes for each month so 12 values for each month and I'm sorry 12 up 12 values total and we took the 30 years applied that change and then took the mean of that so we my point being that we were taking a lot of we're doing a lot of averaging and even with doing all these all this averaging we end up with so this is temperature on the left precipitation on the right for 2030 2050 and 2080 I know this is kind of hard to read and the the distribution and the boxing whisker plots is the 56 scenarios that we ran so you can see that there's there's an enormous distribution even by 2030 and remember this is a 30-year average each of those 10 years representing 30 more years so really this is a 90-year average from the model this is a lot of averaging and still we get changes in temperature with GCMs are fairly good at predicting changes in temperature between zero and almost up to two by 2030 which is you know in 20 years less than 20 years and then by 2050 you know you still get these these huge variations and precipitation as well and this is in percentage so this it's a percentage of so you get you know somewhere between negative 20 and positive 60 in the A2 scenario so then we ran it through the model and we end up with results so you know same sort of distributions is not so surprising in general precipitation is going down most of the medians are below the the zero mark and temperature is going up so obviously runoff is going to generally go down and so yeah so we end up getting these these pretty large distributions and then the X marks here so the the X on this side is the A2 the Hadley A2 and then on this side is actually B2 but I didn't really have a good place to put it just to show you you know where these where these end up falling and so you know obviously we chose a method taking a lot of doing a lot of averaging and and maybe downscaling can give you a better sense of what's happening you know this is for I forgot to mention this for a specific basin but you end up with at the end of the day you kind of end up with this this one point this one mark and you don't really know if this one mark is at the top or at the bottom of the full distribution and so this is change over all of Morocco's is you know again taking more more averaging taking more means over all of the basins and you know obviously the the distribution kind of shortens a little bit some are dry some are wetter and then the Hadley actually ends up kind of somewhere in the middle pretty much in within that 50% middle 50% so finally I had a little bit more time than I thought the so obviously the future is uncertain we all know that and I don't think a colleague of mine said the future is uncertain GCMs are terrible therefore what we should do and engineers often have a safety factor I'm sure you guys are familiar with that so he was saying what we should do is we should just change the safety factors bump it up a little bit so instead of 1.5 we'll use it to end of the day we don't need to really do these huge impact studies problem with that of course is it's more expensive and we really need to to to narrow that that you know where that is exactly so instead of maybe 1.5 we might decide we'll use 1.8 based on on long analysis because it's more expensive to build up to so I'm not saying that the the GCMs are bad they're great they're the best the science has to offer I think climate scientists have a really difficult really difficult job trying to predict these things but what I do want to say is I think it's it's really important that we start to and then this has been said before but that we start to understand these things with an uncertain future and start planning with an uncertain future just like we've done in the past and I think that understanding where that distribution falls is is really valuable information for for policymakers and that is the end