The use of TRMM satellite daily information (Tropical Rainfall Measuring Mission) is proposed to estimate extreme precipitation over ungauged areas and time periods. Assuming precipitation non-stationarity as signal fluctuations due to the ENSO impact, it is possible to find common features between in-situ and TRMM data sets by multi-resolution analysis (MRA) with wavelet transform, especially during the wet period where maximal precipitation series are obtained. This method could be applied on other stations as a regionalization for obtaining rainfall datasets as a solution of data scarcity. It is shown some results obtained into the objectives of a local project of improving the statistical downscaling of TRMM and its applicability in water engineering projects.