Error Generation Characteristics of the Areal Rainfall Estimation Interpolation Method Using Rainfall Radar Data

Abstract

This study evaluated the effects of watershed area, rain gauge density, rain gauge distribution, and rainfall movement direction on regional average rainfall estimates by comparing the spatial interpolation results. To this end, the interpolated watershed mean using radar grid data from the Ministry of Environment’s TM rain gauge point was compared with the watershed mean of all radar grids within the watershed. Results showed that eight or more rain gauges are required for accurate area average rainfall estimates in small watersheds of less than 500 km<sup>2</sup>. In addition, the inverse distance weighting method tends to underestimate area average rainfall, whereas the Thiessen method tends to overestimate it. Furthermore, as linearity of rain gauge distribution increased, the number of rainfall events with outliers increased. In particular, when the correlation coefficient exceeded 0.7, outliers appeared to rapidly increase, sharply increasing the possibility of overestimating or underestimating area average rainfall. This study clearly showed that there are limits to area-averaged rainfall estimates when using rain gauge rainfall data in small watersheds.