Consideration of Rainfall Intermittency and Log-normality in the Merging of Radar and Rain Gauge Data

Merging radar and rain gauge data is important as it can generate accurate rain rate field. By combining the point rain gauge data and spatial information from radar, it is expected to get higher-quality rain rate field. Data assimilation techniques such as successive correction method, conditional merging, co-kriging, Gaussian merging, etc., have been generally used as merging techniques. However, these techniques were not successful to combine the radar and rain gauge data due to the intermittency and log-normality of rain rate data. In this study, to overcome this problem, a method was proposed to consider no rain (zero) measurement in the data merging. Summary of this research includes, first, rain rate data was classified into 4 cases: following normal distribution and not including 0 (case 1), following normal distribution and including 0 (case 2), following log-normal distribution and not including 0 (case 3), and following log-normal distribution and including 0 (case 4). Previously, cases 1 and 3 were mostly evaluated in many literatures. This study mainly focused on cases 2 and 4. Second, to show the problems in cases 2 and 4, the conventional co-kriging technique was applied, whose result was then compared with those of cases 1 and 3. Finally, the proposed method was evaluated with the radar and rain gauge data collected over the Chungju Dam basin, Korea.