Microwaves from satellites can be used for rainfall rate retrieval by two different methods -- a statistical based algorithm or a physical based algorithm.Until now the statistical based algorithm has been more frequently applied because it is less complicated. However, the physical based algorithm provides the best method to reduce estimation retrieval bias. It can be simplified by reducing the number of unknown parameters such as surface emissivity, water vapor content, temperature profile, and cloud height. GMS infered data and RAOB data were used to estimated some atmospheric parameter and which were treated as known parameters. The most uncertain factor is surface emissivity over land which was resolved by NDVI(Normalized Differential Vegetation Index). Of the many physical based methods, the nonlinear simultaneous algorithm of determining rainfall rates proved to be the best. Rain gauge data over land was compared with satellite retrieved data. Area rainfall rate determined by the ATI (Area Time Integral) method was considered as ground true. The accuracy of the rainfall retrieval results in whole 1996 near Taiwan area were compared, most of nonlinear retrieval method are better than SI method which is the best statistical method in Taiwan area. The accuracy of stratus rain rate estimation was worse than cumulus rain rate estimation because cloud structure was difficult to determine and the cumulus pattern was assumed is all cases. Also when instant wind speed exceeded 30m/s and rainfall rate was over 50mm/hr, SSM/I rain information was reduced for it is close to instrument saturation.

*Section Chief / Meteorology Satellite Center / Central Weather Bureau

**Professor / Center for Space and Remote Sensing Research / National Central University

[go back]