Influence of Rain Gauge Density and Temporal Resolution on the Performance of Conditional Merging Method

Abstract

This study aims to evaluate the performance of conditional merging (CM) technique to merge radar and in-situ rain gauge data. The merging method contains the following steps: (1) the ordinary kriging(OK) technique is applied to obtain the gauge-only and radar-only kriged rainfall fields; (2) the correction field is calculated by subtracting the radar-only kriged rainfall field from the original radar field; (3) the composite rainfall is derived by adding the correction field to the gauge-only kriged field. We investigated 33 10-min rainfall fields; there were 22 instances where the CM-based rainfall fields were more accurate than the gauge-only fields. The correlation coefficients of the CM and the gauge-only OK methods varied between 0.63 and 0.83, and 0.01 and 0.74 respectively. The performance of both methods improved with greater gauge density and coarser temporal resolution. Furthermore, the OK and CM techniques’performances were more sensitive to gauge density and temporal resolution respectively. We also found that the addition of gauge significantly improved the performance of both methods when the existing gauge density is approximately 1 gauge/100 km2. However, this effect is absent when the density reaches 4 gauges/100 km2.