Precipitation

Using hourly thermal infrared images, cloud levels are classified on the basis of their cloud top temperatures: low, medium low, medium high, high and very high or "cold" clouds. The presence of these cloud classes is counted during the period of interest (day, dekad) leading to cloud durations (CD) for each level. Rain gauge data are received in near real time through WMO-GTS Synops reports. These data are related to the satellite cloud durations by means of local regression. This means that a regression equation is established for each rainfall station considered and its 11 surrounding rainfall stations:

R = a + a₁.CD₁ + a₂.CD₂ + a₃.CD₃...+ residual

The regression coefficients (a) and the residual are assigned to the station. This procedure is repeated for every rainfall station. Hereafter the regression coefficients and residual are interpolated to each pixel by means of inverse distance weighing. Finally the rainfall is calculated for each pixel and the complete rainfall field is obtained.

Monsoon precipitation dynamics in Asia