Precipitation Estimated by Remote Sensing in the Baixo São Francisco Physiographic Region: A Performance Analysis of TRMM (3B42) and CHIRPS

The use of data estimated by Remote Sensing has become an alternative to make up for the deficit of meteorological/rainfall stations in Brazil. Therefore, this research aims to evaluate the performance of precipitation estimated by TRMM and CHIRPS products in the Baixo São Francisco (Baixo SF) physiographic region. To this end, using the following statistical parameters: linear correlation coefficient (r), root mean square error (RMSE), relative bias percentage (PBIAS) and Nash-Sutcliffe efficiency (NSE), data from precipitation from 67 rain gauges with the respective values estimated by TRMM and CHIRPS, in the period from 2000 to 2015. With values of r = 0.88 and 0.87; RMSE = 31.01 and 144.98 and NSE = 0.76 and 0.71, CHIRPS performed better than TRMM on the monthly and annual scales. On the daily and decennial scales, neither product obtained significant results. In regions further away from the coastal zone, where altitudes range between 400 and 700 meters, data from CHIRPS and TRMM had better accuracy. It is concluded that only the CHIRPS estimates, in the monthly and annual periods, have good accuracy to represent the rainfall in the Baixo SF in the studied period.