RAL/HAP Seminar - Combining a large sample of catchments and a flexible hydrological modeling framework: benefits and challenges

RAL/HAP Seminar - Combining a large sample of catchments and a flexible hydrological modeling framework: benefits and challenge  Note - now at 2:30 pm
Nans Addor,  Post-doctoral scientist in RAL/HAP
22 June 2017, 2:30 pm, FL2-1001

Abstract:

In hydrological studies, the formulation of generally valid conclusions is often impeded by the small number of catchments and hydrological models involved. Many studies at the catchment scale rely on a handful of catchments, so it is usually unclear to which extent their conclusions apply to catchments in other locations. Similarly, when modeling is involved, the number of models is typically low, so one may wonder whether the modeling results are generally valid, or if they are conditional on the modeling setup. To progress on these two fronts, I am co-developing and using a large sample of catchments (CAMELS) and a flexible hydrological modeling framework (FUSE). CAMELS is a new data set involving 671 catchments in the contiguous USA (CONUS). It consists of i) daily atmospheric forcing and discharge time series and ii) a wide range of catchment attributes describing the topography, vegetation, soil, geology and network characteristics of each catchment. I will show how CAMELS can be used to assess the information content of hydrological signatures (indices characterizing hydrologic behavior) and to explore how well those signatures can be predicted using catchment attributes. I will summarize these findings by introducing a ranking of hydrological signatures, useful for the evaluation, calibration and selection of hydrological models. FUSE is a flexible hydrological modeling framework enabling the construction of conceptual models piece by piece. I adapted the original FUSE model and produced a new version optimized for parallel computing and that can be run either at the catchment scale or on a grid. I used FUSE to explore whether a flexible modeling framework can replace a small ensemble of hydrological models typically used for impact modeling. I ran different FUSE instances over the CONUS using 10 CMIP5 climate projections over 1980-2100. FUSE simulations are realistic and compare well with simulations from the hydrological models VIC and PRMS. I will discuss the potential of flexible modeling frameworks to provide more reliable and more tractable estimates of the uncertainty in hydrological projections.

 

Building:

Room Number: 
1001

Type of event:

Will this event be webcast to the public by NCAR|UCAR?: 
Announcement Timing: 
June 19, 2017 to June 22, 2017
Calendar Timing: 
Thursday, June 22, 2017 - 2:30pm to 3:30pm