Analyzing the Relations Between Spatial Variables in Khanmirza Plain: Comparison of Geological Weighted Regression and Ordinary Least Square Models

Usually environmental science and geography researcher use OLS model for variable spatial relations analyzing. This model has some lacks and shortagea in spatial outcome especially in local scale. In recent years, GWR model are used by some scientists for analyzing the relation of spatial variables in local scale. In this article, for analyzing and comparison of these models, land use change (as dependent variable) are assessed in relation with drawdown and withdrawal of groundwater resources (as independent variable) in years of 2001-2011 in Khanmirza Plain (Chaharmahal va bakhtiari province). In order to study the models efficiency, the Standardized residual variation's coefficient, Spatial Local dependencies, Morn’s Index, Corrected Akaike Information Criterion and Local coefficient of determinationwere used. Result indicate that based on Standardized residual coefficient of variation, GWR model has better ability to adopt data on variables respect to OLS. The findings of the research showed that based on the standardized coefficient of variations, GWR model has the ability to adapt data than OLS. Also, based on the results of the explanation coefficient on the variables of the research, GWR model creates a favorable local fit between regression and sample points. Based on Morrow index, GWR pattern represents the least similarity of the amount and location in the adjacent positions of the data samples and prows the performance of GWR model in providing spatial outputs relative to the OLS pattern