In Cuba rainfall is a meteorological variable of importance at the national level, economic and social losses could have been avoided knowing the exact position where they would occur. To solve this problem, the NS was created, a system that provides, among many others, precipitation forecasts, although it has errors when providing the quantity, as well as the position of the precipitation to occur. That is why this project aimed to improve the precision of the system's precipitation forecast by implementing the sliding window method. From its implementation, the results obtained were that the spatial error presented by NS was reduced by using a window of size N = 15 and the maximum and average instructions; the quantitative error was more optimally decreased with the mean instruction, using the same window size, and this last instruction was the one that most completely improved the precipitation forecast provided by NS, both in position and in the amount of forecast precipitation.