Systems for Forecasting Flows and Their Uncertainty

by Konstantine P. Georgakakos, (M.ASCE),
Alexandre K. Guetter,
Jason A. Sperfslage,

Abstract: Discussed are design issues of distributed systems for operational flow forecasting. Results of sensitivity analyses are presented which illustrate forecast flow dependence on model discretization level and parameter estimation error. The appropriate model discretization level depends on the error properties of the model forcing, the catchment delineation errors resulting from the digital terrain elevation databases, and the model and associated parameter estimation errors. Sensitivity with data from Oklahoma shows that large-scale basin-average soil water content depends more on the precipitation input and soil water initial conditions than on the spatial distribution of important model parameters. Uncertainty in local soil water estimates, however, does depend on the uncertainty of local parameter values. Also, the results show that coarse spatial model aggregations store more water in the soil and produce less runoff volume than equivalent finer aggregations. The spatial structure of radar rainfall and the nonlinear catchment response generate fractional contents of model soil water that are scale dependent.

Subject Headings: Soil water | Errors (statistics) | Hydrologic models | Forecasting | Data processing | Uncertainty principles | Parameters (statistics) | Water content | Sensitivity analysis | Terrain models | North America | Oklahoma | United States

Services: Buy this book/Buy this article


Return to search