TRMM SAMPLING OF RADAR-AMEDAS RAINFALL USING THE THRESHOLD METHOD

It is known that spatially averaged rainfall rate [R] is highly correl ated with the fractional area (F) of rain rate exceeding a preset thre shold (tau), when the area is large enough to include numerous convect ive systems in various stages of their life cycles. Using this fact, a method to estimate area-averaged rain rate from F(tau), which is obta ined from satellite observations, is proposed For Tropical Rainfall Me asuring Mission (TRMM). There have been numerous studies investigating F-[R] relationships and optimal thresholds at several radar observati on sites around the world but no studies to confirm the performance of the method within Japan. In this study an analysis of radar-AMeDAS (A utomatic Meteorological Data Acquisition System) precipitation data is presented. The F-[R] relationships of radar-AMeDAS rain data have bee n examined systematically, with the result that the optimum threshold that maximizes the correlation between [R] and F(tau) is neat 3.5 mm h (-1) in every year and season of available data. Using the threshold m ethod with the average coefficients obtained when the threshold is set to 3.5 mm h(-1), TRMM sampling of radar-AMeDAS rainfall is simulated. Fixing 5 degrees x 5 degrees areas, monthly mean area-averaged rain r ate is estimated from the observational coverage that would he obtaine d by TRMM during a month. The errors from the threshold method are onl y 3%-4% larger than the sampling errors (14%-19% on average) obtained by using the full dynamic range of observed rain rates. Considering th e dynamic range of TRMM sensors, the threshold method would be an effe ctive method to estimate area-average rain rate.