Geographically weighted regression, Gilan, Isohyets, Ordinary least square regression, Regression Decision tree.

Document Type : Original Article

Authors

.

Abstract
Among various approaches for mapping isohyets are regression analyses techniques that use relationship between precipitation and geographical factors. Current study was carried out to compare global regression methods (multi polynomial regression and ordinary least square methods), local regression methods (local polynomial regression, geographically weighted regression) and decision tree regression. Average of 20 years annually precipitation data of 185 meteorological observations over Gilan Province and its neighboring stations were used for modeling of spatial distribution variations of mean annual precipitation by using other variables like elevation and point locations from the sea level. Comparison between results using cross validation technique showed that geographically weighting regression method has the highest accuracy to estimate mean annual precipitation (R2=87 and RMSE=147mm) and can be used to map isohyets in Gilan province.

  • Receive Date 15 September 2011
  • Revise Date 05 February 2012
  • Accept Date 04 April 2012