New directions in semi-distributed hydrologic modeling

Rainfall-runoff models are required as a management and research tool for different hydrologic problems. The complexity of a model depends on the available input data, computing capacity, modeling capabilities, and output requirements. While many lumped parameter models are employed in practice, recent advances have seen the development of many complex distributed hydrologic models. Unfortunately the data and time requirements of distributed models are often large. The semi-distributed soil moisture routing (SMR) model, a relative simple water-balance model, was developed for small watersheds with shallow soils. This model is extended with procedures for infiltration excess overland flow, percolation, overland flow routing, and calibration. The newly developed SMR-V2 model can take advantage of temporally and spatially distributed input data, including NEXRAD rainfall data, and can better perform in larger watersheds with deeper horizontally layered soils. The new model is tested in 2 hypothetical 15 x 15 cell watersheds. SMR-V2 appears to represent many different processes occurring in a watershed and can account for overland flow generation and its movement across the watershed. The shuffled complex evolution (SCE) algorithm was employed to calibrate the SMR model using input data from the Dry Creek watershed in the Catskill Mountains of New York State. Comparison with the genetic algorithm employed previously in the same watershed indicates SCE to be more efficient and effective in calibrating the SMR model.