Files in this item



application/pdfMelchingYenWenzel_1987.pdf (6MB)
(no description provided)PDF


Title:Incorporation of uncertainties in real-time catchment flood forecasting
Author(s):Melching, Charles Steven; Yen, Ben Chie; Wenzel, Harry G., Jr.
Contributor(s):University of Illinois at Urbana-Champaign
Subject(s):Water resources center
Water resources center--Illinois
Hydrology and hydraulics
Decision making
Flood control
Flood forecasting
Flood protection
Fflood warning
Mathematical model
Real-time forecasting
Geographic Coverage:Illinois (state)
Abstract:Floods have become the most prevalent and costly natural hazards in the U.S. When preparing real-time flood forecasts for a catchment flood warning and preparedness system, consideration must be given to four sources of uncertainty -- natural, data, model parameters, and model structure. A general procedure has been developed for applying reliability analysis to evaluate the effects of the various sources of uncertainty on hydrologic models used for forecasting and prediction of catchment floods. Three reliability analysis methods -- Monte Carlo simulation, mean value and advanced first-order second moment analyses (MVFOSM and AFOSM, respectively) - - were applied to the rainfall -runoff modeling reliability problem. Comparison of these methods indicates that the AFOSM method is probably best suited to the rainfall-runoff modeling reliability problem with the MVFOSM showing some promise. The feasibility and utility of the reliability analysis procedure are shown for a case study employing as an example the HEC-1 and RORB rainfall-runoff watershed models to forecast flood events on the Vermilion River watershed at Pontiac, Illinois. The utility of the reliability analysis approach is demonstrated for four important hydrologic problems: 1) determination of forecast (or prediction) reliability, 2) determination of the flood level exceedance probability due to a current storm and development of "rules of thumb" for flood warning decision making considering this probabilistic information, 3) determination of the key sources of uncertainty influencing model forecast reliability, 4) selection of hydrologic models based on comparison of model forecast reliability. Central to this demonstration is the reliability analysis methods' ability to estimate the exceedance probability for any hydrologic target level of interest and, hence, to produce forecast cumulative density functions and probability distribution functions. For typical hydrologic modeling cases, reduction of the underlying modeling uncertainties is the key to obtaining useful, reliable forecasts. Furthermore, determination of the rainfall excess is the primary source of uncertainty, especially in the estimation of the temporal and areal rainfall distributions.
Issue Date:1987-09
Publisher:University of Illinois at Urbana-Champaign. Water Resources Center
Genre:Report (Grant or Annual)
Sponsor:U.S. Department of the Interior
U.S. Geological Survey
Rights Information:Copyright 1987 held by Charles Steven Melching, Ben Chie Yen, Harry G. Wenzel, Jr.
Date Available in IDEALS:2016-05-18

This item appears in the following Collection(s)

Item Statistics