Verification for 24-Hour Forecast of Pinpoint Precipitation Using the Mesoscale Meteorological Model Mm5

In this study, verification is performed for the accuracy of 24-hour forecast of pinpoint precipitation using the mesoscale meteorological model MM5 with a high-resolution grid spacing of 2 km through comparisons with the precipitation measured at 9 weather stations in Gifu Prefecture over one-year period from June 2005 to May 2006. It is shown that the values of correlation coefficient of 3-hourly accumulated rainfall between the predicted values and the measured ones become almost 0.4 or more over one year although they are about 0.7 during the period from January to May. In addition, a radar data assimilation system to initialize 3-dimensional cloud microphysical fields accurately is newly developed based on the template matching technique. The forecast accuracy is improved not only in the large-area distribution of precipitation fields but also in the time series of the pinpoint precipitation.