Extended Abstract (628K)10.2
A comparison of statistical and explicit short-term hydrological forecasting techniques
Glenn E. Van Knowe, MESO, Inc., Troy, NY; and K. T. Waight, M. Ceperuelo, J. Aymamí, S. Parés, S. Arumugam, and J. Oh
Real-time operational hydrological forecasts are needed for several reasons, including hydroelectric power generation planning and warning of possible water-related disasters. There are several approaches to real-time hydrological forecasts, each having strengths and weaknesses. In an attempt to assess the costs and benefits of one model approach compared to the other, we are in the process of comparing the performance of real-time forecasts produced by a statistically based model with forecasts produced by an explicit distributed hydrological model for both long and short-term forecasting.
The approach employed for the statistical forecasts uses Model Output Statistics (MOS), including Principal Components Regression, semi-parametric local likelihood, and constructed analogues with nonparametric resampling. The explicit model being used in the evaluation is the Distributed Hydrology Soil Vegetation Model (DHSVM). DHSVM is a spatially distributed hydrological model that explicitly represents the effects of diverse topography and heterogeneous subsurface conditions on the downslope redistribution of subsurface moisture that provides a dynamic representation of the spatial distribution of soil moisture, snow cover, evapotranspiration, and runoff.
The statistical forecasting method has two advantages; it is easier to implement and computationally very efficient. However, initial results indicate that this method may underestimate the rate of runoff as compared to the distributed model.
The results of several seasons of historical model simulations run with both methods in forecast mode (that is, only data available in real-time is used in the simulations) will be compared and shown in the presentation.
Session 10, Water Resources and Forecasting Applications
Wednesday, 20 January 2010, 4:00 PM-5:00 PM, B304
Previous paper Next paper