*Appendix A.1. Tropical Rainfall Measuring Mission 3B42 Dataset (TRMM)*

TRMM 3B42, one of the TRMM Multi-satellite Precipitation Analysis datasets, provides 3 h precipitation estimates for the quasi-global coverage of 50◦S to 50◦N latitude, with coverage going back to 1998. TRMM 3B42's algorithm combines rainfall estimates from infrared and microwave satellites with gauge data and is operated in four major steps. (1) The Goddard Profiling Algorithm is used to convert and combine passive microwave data from plentiful sensors (TMI: the TRMM microwave imager, AMSU: Advanced Microwave Sounding Unit, AMSU-E: Advanced Microwave Sounding radiometer-Earth, and SSM/I: Special Sensor Microwave Imager) to 3 h precipitation estimates. (2) The infrared precipitation estimate from geostationary satellites is produced by histogram matching of the passive microwave precipitation estimate. (3) Then, microwave and infrared precipitation data are blended using infrared data to fill in the missing data where microwave data are not available. (4) The rainfall data are added up to monthly totals and blended with GPCC monthly precipitation data by using the inverse error variance weighting method to obtain the TRMM 3B43 values. The monthly sums of precipitation estimates are adjusted to make the results equal to the TRMM 3B43 monthly value. The final adjusted estimates are the TRMM 3B42 [68].
