Spatial analysis of annual precipitation of Khuzestan province; An approach of spatial regressions analysis

Knowing of precipitation values in different regions is always of main and strategic issues of human which has important role in short- term and long-term decisions. In order to determine of precipitation model and forecasting it, there are different models, but given that the precipitation data have a spatial autocorrelation, the spatial statistic is a powerful tool to recognition of spatial behaviors. In this research, for determine of precipitation model and predicting of it with geographical factors e.g. altitude, slope and view shade and latitude- longitude by using spatial regressions analysis such as ordinary least squares (OLS) and geographical weighted regressions(GWR), 13 synoptic stations of Khuzestan province from establishment to 2010 were used. Results showed a powerful correlation between precipitations with geographical factors. Also results of modeling through OLS and GWR representative that forecasting of GWR is close to reality, so that in GWR, the sum of errors of residuals is less, the  is more and there aren't any spatial autocorrelation in residuals and the residuals are normal. The of OLS can only justify 75 percent of precipitation variations with spatial factors while in GWR this quantity is 82- 97 percent. Accordingly, it was found that, in east, northeast and north of province the altitudes, in east and northeast and Zagros Mountains the view shade and slope are the most important spatial factors, respectively.