This study presents the Spatial Lag Model (SAR) and Conditional Autoregressive Model (CAR) in order to investigate the association between soybean yield and agrometeorological variables related to medium temperature and global solar radiation. The study was realized with data from the agricultural years from 2005/2006 to 2007/2008 crops in the West Region of the state of Parana. As the agrometeorological data are available only for eight cities of the region in study, the estimates were obtained through the use of Thiessen polygons. The estimation of the parameters of the adjusted models was obtained using the method of maximum likelihood. The evaluation of the performance of models was held based on the coefficient of determination (R²), maximum value of the logarithm of the likelihood function and Bayesian Information Criterion of Schwarz (BIC). This study also allowed to verify the correlation and the spatial autocorrelation between soybean yield and the agrometeorological factors by analyzing spatial area, by uses of Global and Local uni and bivariate and significance tests. The study demonstrated that by means of performance indicators used, the SAR and CAR models offered better results than the classical multiple regression model.
spatial autocorrelation; spatial statistics area; spatial SAR and CAR models
