The application of Geographically Weighted Regression (GWR) and Ordinary Least Squares (OLS) Models in estimation of Mediterranean system precipitation values

Document Type : Original Article

Authors

1 shahid beheshti university

2 Assistant Professor of Geography, University of Paymnoor, Iran

Abstract

Relationships between rainfall and altitude can have significant consequences for rainfall, hydrology and environmental management studies. Designing a rainfall model for areas lacking sufficient stations, especially mountains, can play a significant role in environmental planning. The objective of this paper is to investigate the spatial variations of the Mediterranean rainfall system based on the altitude variable in western Iran using GWR and OLS. In this study, at first, the most suitable two-day Mediterranean rainfall pattern were selected for study. The spatial relationships between rainfall and altitude variables were modeled using Geographical Weighed regression techniques and classical regression. Comparison of these two methods showed that the AIC values of the GWR model is less than the OLS model, which indicates a lower error and better fit of the GWR model than the OLS model. The results of spatial modeling in the form of scatter plot and interpolation maps showed that the WGR method is well modeled with a suitable fit of spatial relationships of rainfall with altitude. In general, the GWR model estimates precipitation increases with altitude toward the central and eastern parts of the regions, along the Zagros high mountain range and its internal slopes

Keywords