Rainfall data presents variability and the attainment of the estimate of its space distribution is essencial in agricultural and environmental planning. The objective of this work is to compare the weighted least squares estimation method for semivariogram modeling with the trial and error fitting estimation method through the jack-knifing validation technique to annual average rainfall in the São Paulo State. Rainfall observations corresponding to the period from 1957 to 1997, from 379 rain gauge stations covering entire São Paulo State, representing an area of approximately 248,808.8 km² were used in this study. The parameters for the hole effect model, used to describe the periodicity exhibited by the semivariograms, were more precisely estimated by using the weighted least square estimation method when compared with the manual fitting method. The jack-knifing technique showed itself to be an adequate tool for the definition of methods and models to be used for semivariance. It also allowed for the definition of 16 neighbors as the ideal number to estimate by kriging the unsampled locations.
periodic; hole effect model; jack-knifing; validation method
