The lack of precipitation data distribution is an obstacle to understand and model its variability, thus to obtain information for regions which do not have meteorological stations or have missing data in your database by interpolation is needed. Thus, the objective of this study consists in using Artificial Neural Networks (ANN's), proposing different procedures for its utilization for spatial interpolation of rainfall data over the Alagoas State. For this study we used 245 rainfall stations located in the States of Alagoas and Pernambuco, from which the information of latitude, longitude, altitude and rainfall near the interpolated station were used as input parameters of the neural networks. Using ANN's in the estimation of lacked rainfall data, showed statistical difference in only one procedure adopted by neural networks. The estimates for the November month showed more consistency with those observed in the base stations, due to lower spatial variability of rainfall during this month.
Infilling missing; goestatistical; variograms; spatial variability
