 I'm a research professor at University of Albany Research Sciences Center and quickly in this talk I'm going to give a context for my work, explain the impetus for why I will be using open IFS. The workflow that I've planned and future plans, the advance is not working. Okay, my research foci cover about three foci. The first is assessing variability, long-term trends and change in the Earth's coupled water, energy and carbon cycles. The second is understanding the role of land atmospheric coupling and climate variability, climate extremes and the predictability of regional climate and the third is developing and applying process-oriented diagnostics that help identify and attribute model errors. So the underlying theme here is coupled and process-oriented diagnostics. And most of that work is framed through GWX panels that I serve on. So, GLASS, the Global Land Atmospheric System Study and GHP, which is the global GWX hydrocommetology panel. For my work, I rely on a number of observational data sources, satellites, sensors including mortis airs, amps or SMAP, routine in situ measurements, spanning the land atmospheric continuum models. I run some models in-house, so offline land models, coupled wharf model, coupled wharf with land data assimilation. Other models, I just analyze the outputs directly from the U.S. North American Land Atmospheric Simulation System, the CMIT 5, 6 climate models, and other global atmospheric reanalyses. Introducing land atmospheric coupling, I'd like to present first this MIC ECH diagram. We call this the ECH complexity of coupling. And you can see a number of positive and negative feedbacks. So we can start at sensible heating. So if we have positive anomaly and sensible heating, then we're going to, temperature will increase, we'll have a growth of the body layer, drier entrainments will increase. We may have a reduction in cloud cover and that will have a negative feedback on surface humidity and latent heat fluxes, which will drive further increase in surface temperature. So we have a positive feedback mechanism there. Alternatively, you could imagine if you have a low level jet of moisture, you may be triggering up like the easterly African low level jet. You may have increased sensible heating here, and you're triggering up that moisture, and you may trigger convection, and then you rain and you have a negative feedback pathway. So this is the complex version of coupling. For the GWX glass panel, we have what we call local coupling working group, and we've boiled that down to a simplified form. And we call it the local process chains. This is the GWX local process chain. We're talking about anomalies in the surface soil moisture. Bigot anomalies in the evaporative fraction at the surface may trigger subsequent changes in the lifting condensation level, which may trigger changes in the development of clouds and perhaps precipitation, which may ultimately feedback full circle to changes in soil moisture. So we say local because all coupling starts locally. The land signal, that's the sensitivity of the vegetation to soil moisture is necessary, but it's not sufficient because you also need that atmospheric state to be responsive to that variability in surface fluxes. Just a quick recap of how I define land atmospheric coupling. So coupling is the degree to which anomalies in the land surface, we're talking about soil wetness, soil texture, surface roughness, temperature, and overlying vegetation composition and structure can affect through complex controls on the partitioning of that surface flux, the planetary boundary layer, mesoscale circulations, and in extreme places, rainfall generation. So this is what people tend to focus on is this soil moisture precipitation feedback pathway. But I would agree with Alan Betts in the statement that this coupling strength is the single most fundamental criterion for evaluating, especially deterministic hydrologic and atmospheric model runs. So quickly when I think of coupling, when do I think it matters most? Well in some cases, at long time scales, coupling may not matter, but on the time scales that I'm certainly interested in, diurnal time scales, temperature and humidity, clouds and rainfall, the evolution of drought and the recovery of drought, heat wave severity, these are the cases when I believe that having realistic coupling in your model are paramount to prediction. It's also generally in terms of the condition you're talking about states when you have large-scale synoptic forcing is weak and the spatial gradients in the surface flexes are sufficient enough to drive mesoscale circulation. So in summary, Paul Durmauer puts it nicely in terms of this is the intersection of sensitivity of the land surface, variability of that system and memory. So when you have all three, this is when you can have a strong impact of coupling. So the one aspect of the land has to be strongly co-varying with one aspect of the atmospheric, could be atmospheric water demand, it could be just simply precipitation. And there's large anomalies. So we're talking about semi-arid regions where they have a large potential range of surface flux partitioning and then it also has to persist. So it has to be in a region where the frequency of rainfall is such that you don't have many, many small rain events which will eliminate the persisted memory of a long drought period. So you have to have the intersection of these three characteristics for coupling to play a large role in your predictability. So I mentioned the models. I want to talk about model coupling. So this is probably a well-known result from the Global Land Atmospheric Coupling Experiment 1 results and they're spanning a number of models here and looking at the spread and the co-variability between the soil moisture and lifting condensation level. You can see nine models here and I like to point out the GFDL model is the most tightly coupled, that's very strongly coupled and CCMA what I would say is probably the most weakly coupled. There's very few cases where we have observational verification constraints on these models so it's unclear which is more accurate other than places like the RM-SGP site and Cabal where we have land atmosphere continuum observations but this is the state of models, lots of uncertainty and little verification data to determine who is the winner of the beauty contest per se. I personally looked at intercomparing models to Ameriflux Fluxnet stations and you can see these bars are the estimates from the local Fluxnet site for coupling the soil moisture vapid diffraction so that land leg of atmospheric coupling. You can see by and large the models off both offline and re-analysis models are always more strongly coupled than the local estimate from observations so my sense is that the models are generally too strongly coupled. An operational example of wind coupling matters in the US we're giving the NOAA CFS team a lot of slack because what happened is that in the CFSR they noticed their two meter temperature skill was very poor and so they decided let's make a quick fix and we'll lower all the rooting depths for just this crop vegetation type highlighted in green. We're going to allow these crops to see the deepest layer in the soil moisture column so what really happens is that these crop types always have water so they can always evaporate and that will cool by evaporative cooling the lower screen level temperature two meter. So this is the GLS model so this is a land model forced with observed base precipitation they had a bias of about four watts per meter squared in latent heat fluxes and their forecast was 27 watts per meter squared so they wanted to reduce this bias and that two meter temperature they decided to extend this rooting zone depth when they did that they certainly reduced the sensible heating bias reversed it from 18 watts per meter squared to negative 10 watts per meter squared and you can see this clear delineation of that crop land mask crop vegetation land mask where they made that modification but in doing so yes they solved their two meter temperature warm bias but they also create in the meantime they created the problems with the boundary layer simulation because there's a perpetual fog over this region so we say here you know the temperature arrow is reduced but we get the right result from the wrong reason so this is the implications for this for earth system models is tragic because you're trying to get the coupled carbon energy and water cycle properly predicted coincidentally and so here you you solve one problem which is a two meter temperature bias and create another problem which is this problem with the boundary layer and the realistic atmospheric demand for water at the surface why was I interested in open IFS so with John Palo Basamo I discussed a proposal and the proposal was successful it's called the role of soil moisture and weather predictability over the U.S. Great Plains it's funded by NASA for the next three years the underlying science question that I asked was how will assimilation of high resolution NASA's map data refined model made atmospheric coupling and lead to improvements in short-term weather and wind energy forecast so I have here a map of the equipped wind energy regions in the U.S. and my focus here is the southern Great Plains this accounts for about 30 percent of the United States wind energy production my approach is to undertake a series of idealized experiments and they're designed in a manner that will allow me to clearly distinguish between the roles of model physics local remote soil moisture effects SMAP data assimilation and synoptic weather on the forecast scale my hypothesis that drove using open IFS was that the best data assimilation results will derive from the most realistically coupled model at the same time we have NOAA and HTSL have benefited from 20 plus and 30 plus year operational development histories respectively and there should be more direct inter comparisons in the past ECMWF did not allow sharing of their model so these inter comparisons were limited and they were never done in the same system with the same parameters and the same meteorological forcing thirdly in the U.S. currently we have a transition operationally from the NOAA 36 to the NOAA MP land surface model it's already been implemented in our national water model and it's soon to be implemented in the in the suite of NCEP operational products and LDAS, GLDAS, GFS, and CFS so my my suggestion in my proposal was based on this idea that it's an opportune time now in the U.S. for this inter comparison as part of the needed NOAA MP critical evaluations why not also look for potential added value from lessons learned using HTSL my plans to use open IFS are one to implement common surface parameters at one kilometer resolution in both land schemes that involve surface soil properties leaf area greenness and albedo will become prescribed on a daily real-time basis generally these are climatological monthly values but the we're looking at the short-term forecast six to 30 hours so it's important that we have the real time the latest real-time surface conditions after I standardize the surface parameters for both the land models I will add HTSL to the NASA land information system I will then quantify the offline NOAA MP and HTSL uncertainty using this process processing that vacation verification toolkit as part of NASA lists once I've verified their offline performance I will then couple lists HTSL to the NASA unified wharf so then we're in a coupled modeling environment and then I will look at integrating the soil moisture data simulation performance and list new work using both land schemes so here is my inner domain outer domain my focus is wind energy forecasting so I plotted here this is the 925 hectopascale murdo wind mean wind speed and vectors so this is the low-level jet during jj a in the southern grid plane so that explains why there's such a dense high-density wind farms in this region I just wanted to this is just my monumental first open IFS HTSL run so that's for posterity I guess and a really exciting thing that we've done for the land scheme it's highly dependent on the soil parameterization so just like precipitation intensity matters well the land scheme that's seeing that precipitation intensity also matters in terms of the soil if you imagine you have a on the gradient of soil hydraulic conductivity if you have a high intensity of precipitation and low hydraulic conductivity prescribed as a soil then you'll have a lot of runoff if you just change that by a little bit the response of the land scheme to a given intensity of precipitation varies drastically so we in the US there's a new 30 meter probabilistic soil series map and it's available at each of these layers so we're going to apply a new layering scheme in HTSL and NOAA MP so I have this consistent layering scheme that's consistent with the available soil texture properties so here we have mapped three of the soil texture properties residual soil moisture saturated soil metric potential pore size distribution index and there's three others that that we use as well here's the citation but I think this is very exciting work that previously there weren't available products I think this is the parallel to the HADIST probabilistic distribution of sea surface temperature period this is the equivalent for soil moisture or soil texture and land surface schemes quickly the list in the new warp system here's the overview here this is the land off-line land system we have the LVT is the pre-processing toolkit so that pre-processes the parameters and the met data LVT is the post-processing so everything from LIS can be passed to LVT for post-processing uncertainty estimation parameter estimation and calibration this is coupled into new wharf wharf has the new wharf has the wharf air WN car model as its core but it has a number of proprietary NASA schemes Goddard visit Goddard this is NASA Goddard micro physics radiation chemistry EDA the has a number of model forcing capabilities so this is enables NASA scientists to really leverage best all their the suite of satellite instrumentation to me the key utility of this is that you can perform a long-term offline spin-ups of the soil moisture in the precise configuration that you're going to be running in the coupled mode so if I were to run my same data simulation experiment with this vanilla wharf I would have to run it in very expensive coupled mode for five or six years but if I'm running this very cheap offline model I can do the same thing much cheaply much more much less expensively and at the same time I can do an analysis across the sensitivity to different parameters like the soil textures that I mentioned so this quick summary of what we've done so far we've completed updates to the surface parameters and soil layering for the U.S. NOAA MP model that's now sorted out so H. Tesla will be next I'm very interested in anyone willing to help test H. Tesla parameter sensitivity and to compile a couple of land atmosphere observation verification data sets as part of the GX global hydrochromatology panel I'm helping organize a new North American regional hydrochromat experiment and the proposal for that is to span implement these AmeriFlex stations with additional boundary profiling capabilities so LiDAR and radar this is one proposal over New York State where I am located we have a new mesonet 125 meter logical stations of which 17 have enhanced profiling capabilities with LiDAR and radiometers and flux stations so these would be the test beds providing the verification data for soil moisture planetary bony layer covariance that we plant that is critical to forecasting this low-level wind so here's the website for the New York Mesonet if you're interested this is a nice storm cloud development sequence from yesterday morning I thought I might share with you thanks yeah this is all in the all in the harbor here yesterday morning it took about one hour the entire cycle is less than an hour and you take that okay thanks at all so questions okay