Submitted to: Journal of Hydrology
Publication Type: Peer reviewed journal
Publication Acceptance Date: 3/29/2004
Publication Date: 10/1/2004
Citation: Diluzio, M., Arnold, J.G. 2004. Formulation of a hybrid calibration approach for a physically based distributed model with nexrad data input. Journal of Hydrology. 298(1-4):136-154. Interpretive Summary: Spatial variability of rainfall is a major source of error in watershed modeling. Often only one or two rain gages are found in a watershed, which may miss rainfall from localized thunderstorms and cause large errors in modeled stream flow. In this study, precipitation from NEXRAD radar (as seen on television weather forecasts) was input to the SWAT watershed model. The radar data gives estimates of precipitation for each 4 square kilometer grid cell, thus 'filling in' between rain gages and providing more accurate stream flow predictions.
Technical Abstract: This paper describes the background, formulation and results of an hourly input'output calibration approach proposed for the Soil and Water Assessment Tool (SWAT) watershed model, presented for 24 representative storm events occurring during the period between 1994 and 2000 in the Blue River watershed (1233 km2 located in Oklahoma). This effort is the first follow up to the participation in the National Weather Service- Distributed Modeling Intercomparison Project (DMIP), an opportunity to apply for the first time within the SWAT modeling framework, routines for hourly stream flow prediction based on gridded precipitation (NEXRAD) data input. Previous SWAT model simulations, uncalibrated and with moderate manual calibration (only the water balance over the calibration period), were provided for the entire set of watersheds and associated outlets for the comparison designed in the DMIP project. The extended goal of this follow up was to verify the model efficiency in simulating hourly hydrographs calibrating each storm event using the formulated approach. This included a combination of a manual and an automatic calibration approach (Shuffled Complex Evolution Method) and the use of input parameter values allowed to vary only within their physical extent. While the model provided reasonable water budget results with minimal calibration, event simulations with the revised calibration were significantly improved. The combination of NEXRAD precipitation data input, the soil water balance and runoff equations, along with the calibration strategy described in the paper, appear to adequately describe the storm events. The presented application and the formulated calibration method are initial steps toward the improvement of the simulation on an hourly basis of the SWAT model loading variables associated with the storm flow, such as sediment and pollutants, and the success of Total Maximum Daily Load (TMDL) projects.