Estimating Characteristics of Rainfall and Their Effects on Sampling Schemes: Case Study for Han River Basin, Korea

Abstract

This study characterized the monthly and regional variation of rainfall fields of the Han River basin using the Waymire-Gupta-Rodriguez-Iturbe (WGR) multidimensional rainfall model. The WGR model parameters were estimated using a genetic algorithm (GA) by comparing the first- and second-order statistics derived from point-gauge measurements and theoretically derived ones for the WGR rainfall model, which were also compared with the results using a nonlinear programming (NLP) technique (the Davidon-Fletcher-Powell algorithm). The WGR model was then applied to the sampling error problem for both rain-gauge network and satellite observation cases. The results of the study are as follows: (1) The GA provides more consistent and closer results for the observed properties of rainfall than the NLP. (2) The higher rainfall amount during rainy months (June to September) is due mainly to the arrival rate of rain bands and the mean number of rain cells per cluster potential center; however, other parameters controlling the mean number of rain cells per cluster—the cellular birth rate, rain cell intensity, and mean cell age—are found to be less sensitive to the rainfall amounts. (3) The number of rain bands (storms) in the upstream mountain area was estimated to be a little higher than that in the downstream plain area, but the cell intensity was a little lower; thus the monthly amount of rainfall remains almost the same for the whole Han River basin, even though more frequent but less intense storms are expected in the upstream mountain area. (4) The sampling errors estimated are not directly proportional to the rainfall amount, nor is its variability; rather, the sampling errors for the rainy months seem more or less the same, but a bit higher than those for the dry months. The sampling errors evaluated regionally (plain area versus mountain area) also did not show enough differences to distinguish one region from another. (5) Finally, the standard errors (a relative measure of sampling error in the rainfall variability) estimated monthly and regionally were estimated to be more or less the same, about a 1% level for the rain-gauge network case. This means that no obvious difference exists, especially for the sampling, to distinguish one region or month from another. We may say that the quality of rainfall data collected monthly and regionally remains almost the same.