Assessment of rainfall estimates of TRMM satellite on Paraíba state

ABSTRACT The spatio-temporal rainfall variability, the low-density of rain gauges and operational problems are complexity factors for hydrologic studies in developing countries such as Brazil. This research evaluates rainfall estimates derived from Tropical Rainfall Measuring Mission (TRMM) 3B42 product, version 7, over Paraiba state, in Northeast of Brazil. This study analyzed time series of 212 rain gauges) and TRMM data within a spatial grid of 198 pixels, for a 14-years period (1998-2011). Analyses were carried out punctually for each pluviometer as well as spatially aggregated for the mesoregions Mata, Agreste, Borborema and Sertão, considering daily, 3-days, 7-days and monthly rainfall. Largest agreement between TRMM estimates and observed rainfall was obtained for Sertão region (correlation coefficient r daily = 0,89 and absolute average error EMA monthly = 16mm) and smallest agreement for seaside region (r daily = 0,75 and EMA monthly = 31 mm), i.e., the best agreement was obtained for regions with low rainfall rates. TRMM data underestimated observed rainfall in sites with large rainfall amount (monthly bias of -24 mm in Mata region) and overestimated in sites with low rainfall (monthly bias of +13 in Sertão and of +11 mm in Borborema). The spatially detailed characterization over Paraiba of the quality of the TRMM estimates showed in this work illustrates the potential of using such rainfall estimates for this state, because knowing its errors, their types and quantification tends to raise the trust in using them and also allows applying correction methods.