Analysis on the Optimal Radar Scan Interval for Urban Flood Prediction

In the current practice, the radar scan method for precipitation estimation periodically observes regions where heavy rainfall is expected with the rotation of the radar. This is different from ground observation devices, such as rain gauge, which continuously observe cumulative precipitation. Different observation methods can cause the precipitation estimation accuracy to vary spatio-temporally. Therefore, quantitative precipitation estimation is needed to improve the accuracy of areal rainfall estimation according to the radar scan interval. For this, an error analysis has been performed according to the scan time in this study after instantaneous observation data were generated at each scan interval using the minute-unit precipitation data obtained by a ground gauge through an observation method that is similar to radar observation. The analysis result showed that there was a drastic increase in the uncertainty of rainfall estimation at scan intervals of 10 and 5 minutes compared with that at scan intervals of 3 and 2 minutes in all cases that relative error, absolute elative error, and correlation of the instantaneous estimation as to the continuous estimation. In the analysis with the actual radar observation data, the relative error tended to increase as the scan interval increased, and it tended to increase drastically with intervals equal to or larger than 3 minutes.