 all right good morning folks welcome back welcome to panel number three which is about mitigating groundwater model uncertainties we have three great speakers to be on this panel that'll take a little bit of time to talk about their experiences with modeling as you can see we are slightly smaller panel than some of the other ones so feel free to contribute your feelings on groundwater uncertainties as we will probably have a little time compared to some of the others so JT Rieger it from JPL will be joining us online because he was unable to make it but we'll start off this morning with Laura Condon from the University of Arizona. Okay so I guess I don't really need much of a title I was just gonna talk a little bit about frontiers in groundwater modeling I'm generally I would count myself as a computational hydrologist so I build large-scale physically based models of groundwater surface water interactions and so we talked a lot about observations yesterday so I was just gonna talk a little bit about what's happening in modeling so I think this is just kind of the figure I have in my head of scale this has been talked about a lot especially this morning that models are a way that we can bridge some of the scaling gaps we have especially for a large-scale decision-making and trying to get for trying to be able to cover the really large-scale systems we have but also at a scale that's relevant for planning and decision-making and also that's important if we want to use physically based models that we can use them to try to understand better what's going on with processes so I just put this figure up I mean you could have a lot of examples of this but this is from Tom Mikesner just illustrating the fact that if you look at the Western US we have a lot of different recharge mechanisms that we might care about so it's not just about getting the water balance right it's about understanding what those mechanisms are at the large scale we have kind of like three main approaches to groundwater modeling I would say so there's lumped parameter global water balance models we also run land surface models globally and now there's a lot more integrated modeling going on where we solve 3d variably saturated flow in the subsurface and that's generally the first two we're doing globally the last one we've started doing regionally and more like the past decade and the reason for that is because it's much more computationally intensive and computationally demanding such as an example of like all the things going on I have kind of two examples of what's happening with integrated modeling over large scales so I work on modeling at the of the continental US there's also a group in Europe that's doing continental modeling of all of Europe and they're actually also coupled up to a weather model at a slightly lower spatial resolution there's also another modeling group in Canada that does a lot of integrated modeling for a lot of different provinces in Canada so just some examples of like what we can do with this we can if we simulate all of the groundwater we can use it to interrogate some of our more conceptual models that we use to try to understand water balances so this is actually a comparison I did looking with our continental simulations comparing to water table ratios which is something that is used a lot in large-scale modeling and just showing that like water table ratios really don't give you the answer this is not a one-to-one scatter line here also using integrated models we can do things like look at residence time distributions this is really important for understanding flow paths comparing to land surface behavior to see like contributions to transpiration on basically like all sorts of partitioning this is these are examples of things they're doing in Europe with the terse system P model that I talked about so really using this for more like short-term forecasting so they do a lot of data assimilation and do actually real-time forecasting with this coupled up with a weather model so really looking at how adding the groundwater storage into that system changes moisture convergence and actually predictions across Europe and so that's kind of like a little pitch of like what's going on in integrated modeling of course there's a lot of limitations so I think data has been mentioned a lot of times I point I have an illustration here for one year of simulation of our continental model we can gather about 1.2 million observation points like actual like in-search you observation points but if you're generating terabytes of outputs then a million observations is a millionth of your outputs so that's just to kind of like highlight the scale of what we're dealing with also we've talked about the increase in available subsurface data like with the glimpse data set from Tom Gleason but that's actually not really like a huge advancement in terms of the number of in-search observations we have so even if we have better gridded data sets to build our models from we have to be really careful about what actually how much actual data we have to use so some of the sky-tent stuff that was talked about yesterday I think is really interesting also I think that we still really don't know what's going on with long-term storage trends this is a paper from Bridget Scanlon where they compared grace from 2002 to 2014 with various land surface models and then two global water balance models so this isn't the integrated modeling that's not happening globally yet and basically the point of this map is just to show that like not only are they getting different answers they're like getting different directions in terms of storage trends so really all over the place and even if we have grace to compare to for the most recent you know 10 to 15 years we have huge uncertainty in what these storage trends should look like in the future and what the low-frequency variability is of that so I'm going past my time just a little since we have plenty of time okay then we also talked about human systems so there's a lot of examples for this is just a paper that came out recently that I thought was good that I'm just kind of showed that the way hydrologists like to think about hydrologic systems is like the typical global what a typical like water balance water cycle figure that everyone has in their mind from elementary school but really humans are dominating the hydrologic cycle all over the place and we don't have the right model in our mind and we're not modeling these things correctly really and so this is just like two examples of like what we can see with integrated models showing on the left some recent work we did where we looked at pre-development shallow groundwater versus post-development shallow groundwater just showing the impact this has on stream flows if you just take subsurface drying as its own basically isolate that out from irrigation or anything else and then the last one is an illustration of something that Manu also showed yesterday showing that if we have pumping happening in our models and we have that irrigation happening in conjunction with what's happening with our atmospheric forcing so when it's dry we use more water that the variability in the subsurface is completely driven by what the humans are deciding to do that's what I've got