The Retrieval Of Precipitation Over Land In China With TRMM/TMI Microwave Data By Using Physical Algorithm

In order to construct the initial cloud-radiation database, 11 cloud-precipitating systems over land surface in China were simulated by MM5 to obtain the rainfall profiles whose upwelling radiative brightness temperature were computed by Monte-Carlo radiative transfer model. Then as the priori databases, 4 databases named Database_m11, Database_m12, Database_m21 and Database_m22 were extracted from the initial cloud-radiation database by using check list and compositve criteria method. The rain rates of three cloud-precipitation systems during July 2007 were retrieved by importing those 4 databases to the GPROF algorithm respectively. Comparing with the results from TMI 2A12 product and three statistic algorithms containing NESDIS, GSCAT and PCT-SI, the rain rate from Database_m12 was close to TMI's and better than that from Database_m11 significently, even better than those from statistical algorithms sometimes. The rain rate of Database_m22 was closer to TMI than that of Database_m21. The rainfall intensities of PR were used to inspect the results calculated by the above algorithms. It is found that TMI 2A12 product shows the best results. Among the four databases, Database_m12 has better impression in general. Database_m11 and Database_m12 controlled by check list display better behavior than Database_m21 or Database_m22 obtained by compositive criteria method. And the databases controlled with compositive criteria method would cause rain rate more intensive. In the same way, the rainfall intensities calculated from the Doppler Radars on land surface were also utilized to compare the results from the above databases. The rain rates from physical methods are more compatible with those observed by the Doppler Radars than those from the statistical methods. Especially the results retrieved by Database_m12 are even better than those from TMI 2A12 product. The reason is although the Database_m12 is based upon the database of GPROF 6.0, it contains the cloud-precipitation profiles which represent the rainfall character of Meiyu front on that area in China. The better expression of Database_m21 and Database_m22 can be found when one-hour accumulated precipitation observed by rain gauges is supposed to test the above various algorithms. Generally PCT-SI has the best behavior than the other two statistical methods. Among the four kinds of hydrometeors retrieved by Database_m12, the precipitable water (pw) displays consistent with that from TMI 2A12 product or the cloud liquid water retrieved by Radar on land. The vertical distributions of other three hydrometeors including cloud water (cw), cloud ice (ci) and precipitable ice (pi) are close to those of TMI. However the contents of those three kinds of hydrometeors are sometimes higher than those from TMI respectively. The hydrometeor structures obtained from other three databases are similar to those from Database_m12 except that the contents of those hydrometeors are lower generally. The distribution place of pw retrieved by the databases controlled with compositive criteria is consistent with cloud liquid water computed from Doppler Radar. When the databases neglected the less than 0.01mm/hr in precipitating profiles, there is almost no negative effect on the retrieved rainfall intensity, but the better hydrometors vertical distribution would be produced. Also the computing time can be reduced by 1/4. Thus, the algorithm constructed in this study has a positive effect on retrieving rainfall intensity and the profiles of hydrometeors. And it can be regarded as a helpful attempt to improve the physical algorithm applied in China with which the rainfall intensity and hydrometeor structures can be retrieved from brightness temperature.