A Physically-based flood prediction model aided by a geographic information system
A physically-based flood prediction model, aided by a geographic information system (GIS), has been developed to derive synthetic flood frequency curves for ungaged or short-term gaged watersheds. A synthetic flood frequency curve is derived through Monte Carlo simulation in which the rainfall-runoff transformation is accounted for on the basis of physical processes. The GIS, which has been developed specifically to support hydrologic modeling, is able to automatically provide the flood prediction model with all parameters required for the simulation, based on the GIS database which stores various spatial data and numerous regional relationships. The essence of the proposed methodology is that a U.S. Soil Conservation Service (SCS) type method and a unit hydrograph are used to represent the physical rainfall-runoff transformation operating on random rainfall inputs having a certain probability distribution. In the process of Monte Carlo simulation, the hydrologic model deterministically converts many thousands of randomly generated storms into flood hydrographs by the convolution of an excess rainfall hyetograph with a synthetic unit hydrograph. A synthetic flood frequency curve is obtained by performing a frequency analysis for peaks of the many thousands of flood hydrographs. The SCS runoff curve number method, a method widely used for computing the depth of excess rainfall, has been modified by treating potential maximum retention and initial abstraction as random variables, in order to consider the variability of storm and watershed characteristics. Probability distributions for these variables are derived based on the analysis of local rainfall-runoff data. Furthermore, a regional dimensionless unit hydrograph and a regional lag-time relationship have also been derived to compute a synthetic unit hydrograph for a watershed of interest. The whole system, i.e., the hydrologic modeling interfacing with the GIS, has been implemented on a MS-DOS microcomputer linked to a graphic monitor, a printer, a plotter, and a digitizer. Performance of the system has been tested for a number of gaged watersheds located in forested foothills of southwestern Alberta, Canada. The validity of the derived regional relationships and proposed procedure is verified by comparing the synthetic flood frequency curves with the observed flood frequency curves for the tested watersheds. Moreover, the effects of using either the standard or the modified SCS method in Monte Carlo simulation on flood prediction results are also assessed. Simulation results show that integration of the proposed hydrologic model and the GIS results in an increased confidence with respect to the predicted flood magnitudes and their frequencies, as well as a significant improvement in efficiency of data handling and analysis for preparing the input to the hydrologic modeling. Replacing the standard SCS method with the modified SCS method markedly improves the simulation results.
Bibliography: p. 330-346.
Chang, C. (1992). A Physically-based flood prediction model aided by a geographic information system (Unpublished doctoral thesis). University of Calgary, Calgary, AB. doi:10.11575/PRISM/12018