The rainfall over Northeast Brazil is quite complex and is considered seasonal, besides suffering strong influences of El Niño, La Niña and other weather systems such as dipole, acting on the ocean basins Tropical Atlantic. In this study the mathematical-computational interpolation technique Kernel Smoothing was applied to rainfall data collected over the Northeast in the period from 1904 to 1998, from 2,283 weather stations. The calculations were developed using the software "Kernel", written in C language and Cuda, enabling interpolation of over 26 million rainfall measurements, allowing to generate maps of rainfall intensity over the whole region, and to calculate statistics for precipitation in the Northeast on monthly and annual scales. The resulting interpolations were used to identify the years with the most and least precipitation along the studied period, the spatial distribution of rainfall in each month, as well as the characteristics of rainfall in times of El Niño and La Niña.
interpolation; smoothing; precipitation
