Fig. 1. 0.5° grid average distribution of precipitation samples, average surface rain rate, cloud microphysics and spectral parameters during Meiyu period over the Yangtze-Huaihe River Valley. (a) Precipitation sample number; (b) rain rate; (c) reflectivity of 0.63 μm channel; (d) reflectivity of 1.6 μm channel; (e) black body temperature of 3.7 μm channel; (f) black body temperature of 10.8 μm channel; (g) cloud effective radius; (h) cloud optical thickness; (i) cloud water path

Fig. 2. Scatter of predicted rain rate of random forest model and observed rain rate for training set and testing set. (a) Training set; (b) testing set

Fig. 3. 0.5° grid average distribution of precipitation intensity prediction, observation and their difference. (a) Predicted surface rain rate over the Yangtze-Huaihe River Valley from 1998 to 2007; (b) observed surface rain rate over the Yangtze-Huaihe River Valley from 1998 to 2007; (c) difference of predicted and observed surface rain rates over the Yangtze-Huaihe River Valley from 1998 to 2007; (d) predicted surface rain rate in orbit 9121; (e) observed surface rain rate in orbit 9121; (f) difference of predicted and observed surface rain rates in orbit 9121

Fig. 4. Scatter of predicted and observed surface rain rate of proposed model. (a) All the rain rate during Meiyu period over the Yangtze-Huaihe River Valley from 1998 to 2007; (b) all the rain rate in orbit 9121

Fig. 5. Probability distribution of precipitation intensity at different levels (light rain, moderate rain, heavy rain, torrential rain, extra-torrential rain) under different cloud characteristic parameters. (a) CWP; (b) TB4; (c) CER; (d) COT

Fig. 6. Relationship between precipitation intensity, cloud characteristic parameters and cloud top temperate as cloud top height. (a) RR; (b) CER; (c) COT; (d) CWP

Table 1. Meiyu periods over the Yangtze-Huaihe river basin from 1998 to 2007

Table 2. Model input predictive factor table

Table 3. Importance of forests random variables

Table 4. Statistical classification table of cloud microphysical parameters and cloud top temperature (each segment contains only the right boundary value)