Rainrate estimation using the GMS-5 IR data and verification

Many efforts have been made to estimate rainrate using the satellite data which provide fairly regular spatial and temporal coverage despite of their intrinsic limitations in the rainfall estimation. It is especially an important tool in the field of nowcasting like the flash flood forecasting system. In this study, we compared three different rainfall estimation methods using the observed rainfall data spanning from 20th July to 20th August 1998 in hourly basis. The time period is selectively chosen because of the severe damages on the Korean peninsula by the mesoscale convective complex.

The methods used are the probability matching method (PMM, Sohn et al., in preparation), the look up table method (Kurino, 1997) and the regression method. The satellite derived rainfall estimations are validated by using the more than 400 automatic weather stations (AWSs) spreaded over South Korea. The mean distance between any two AWSs is about 40 km. The root mean square (RMS) error and the mean error show that all three methods give excellent results when the heavy rainfall occurs in the area of the low equivalent blackbody temperature of the GMS5-IR1 (TB1) and of the weak rainfall at the high TB1. However, there are several cases which produce unrealistic rainfall estimation especially when the low TB1 is not associated with the high rainfall area even the cirrus cloud effect is removed in our estimation.

Each method has its own characteristics. The PMM has a tendency to underestimate the weak rainfall and to overestimate the heavy rainfall. On the contrary, the regression method overestimates the weak rainfall and underestimates the heavy rainfall. The estimation from the look up table method is so smooth that it is impossible to pick up the heavy rainfall case. A detailed error analysis of the case study periods and their application results for the summer of 1999 over Korean peninsula will be given in the presentation.

Toshiyuki Kurino, 1997: A Rainfall Estimation with the Infrared Split-Window and Water Vapor Measurement. Meteorological Satellite Center Technical Note No. 33, JMA, 91-100.

Sohn, B.J., H.J. Oh, H.S. Chung, E.A. Smith, 1999: Rainfall estimates using probability matching method applied to combined GMS-5 and SSM/I data. To be submitted to J. Met. Soc. Japan.