Publications

The availability of accurate rainfall data with high spatial resolution, especially in vast watersheds with low density of ground-measurements, is critical for planning and management of water resources and can increase the quality of the hydrological modeling predictions. In this study, we used two classical methods: the optimal interpolation and the successive correction method (SCM), for merging ground-measurements and satellite rainfall estimates. Cressman and Barnes schemes have been used in the SCM in order to define the error covariance matrices. The correction of bias in satellite rainfall data has been assessed by using four different algorithms: (1) the mean bias correction, (2) the regression equation, (3) the distribution transformation, and (4) the spatial transformation. The satellite rainfall data were provided by the Tropical Rainfall Measuring Mission, over the Brazilian Amazon Rainforest. Performances of the two merging data techniques are compared, qualitatively, by visual inspection and quantitatively, by a statistical analysis, collected from January 1999 to December 2010. The computation of the statistical indices shows that the SCM, with the Cressman scheme, provides slightly better results.