The aim of this study was to evaluate the monthly rainfall estimates obtained by two data sources (3B43_V6 algorithm and University of Delaware) for the State of Rio Grande do Norte, during the 1998 to 2008 period. Statistical analysis based on the linear correlation coefficient and on the performance index from a 2x2 contingency table was done. Coastland and semiarid precipitation regimes which depend on geographical location and on soil type coverage were considered. It was noted that the 3B43_V6 underestimated the coastland rainfall and overestimated the rainfall over arid zones. The Delaware mean rainfall agreed well with observed rainfall, but shows discrepancies concerning to interpolation method and grid spacing. The results suggests that both 3B43_V6 and Delaware rainfall dataset are able to capture the rainfall mean variability both in time and space; however, they show deficiency in identifing extreme precipitation events.
3B43_V6; Delaware; contingency table; TRMM; semiarid
