NEXRAD data (radar-based multi-sensor precipitation estimates) application to runoff routing using a hybrid hydrologic model (Distributed-Clark)

Precipitation is one of the primary inputs for hydrological modeling and related fields of work. In an effort to improve hydrologic application for flow predictions, NEXRAD radar-based multi-sensor precipitation estimates have been used by scientists and engineers particularly for GIS-based watershed-scale hydrologic model due to their spatial variability features; they can capture localized storms that completely miss rain gauges. However, to be used in conjunction with other geospatial products, the HRAP grid, which is a rectangular grid of approximately 4 km by 4 km and defined in a polar stereographic map projection using a spherical earth datum, based precipitation have to be translated into regular grid based data which use an appropriate map projection system (coordinate system) with watersheds well represented. Also, few feasible (simple) models exist for implementing spatially distributed rainfall inputs for hydrologic simulation.

In this study, a GIS-based tool was first developed for automation of required NEXRAD precipitation data (particularly for NCEP Stage IV national composite product) processing to generate spatially distributed rainfall inputs for hydrological modeling. It includes three steps of map projection transformation (to Albers Equal Area Conic map), modeling extent and NEXRAD grid subsetting, and raster and time series data generation. Then, a GIS-based spatially distributed Clark's unit hydrograph method (Distributed-Clark), a lumped conceptual and distributed feature model (hybrid hydrologic model) which combines the concept of Clark's unit hydrograph and its spatial decomposition methods, is introduced for use with NEXRAD data for runoff routing (runoff flow prediction); it can consider spatially distributed rainfall as model inputs. Differing from physically-based distributed hydrologic models, Distributed-Clark has relatively few parameters. Also, if preprocessing is completed, it can perform rapid computations for obtaining hydrographs at watershed outlets.

Case studies of model application were conducted in four basins (about HUC 10-digit sized): Illinois River near Tahlequah, OK; Elk River near Tiff City, MO; Silver Creek near Sellersburg, IN; and Muscatatuck River near Deputy, IN with (1) validation of NEXRAD radar-based rainfall estimates against available rain gauge observations, and (2) performance evaluation of the hydrologic model based on comparisons between simulation results and observed streamflow for spatially distributed radar-based rainfall and spatially averaged (lumped) gauged data.