Objective air temperature mapping for the Iberian Peninsula using spatial interpolation and GIS

This study presents an objective mapping of monthly mean air temperature over the Iberian Peninsula using the spatial interpolation of climatological data. The research focuses on an interpolation method (multiple regression with residual correction) that combines statistical global analysis with a local interpolation (splines and inverse distance weighting). Geographical information (the independent variables) is used to predict air temperature (the dependent variable) through the regression relationship and to test several approaches. A comprehensive GIS implementation makes it possible to address many cartographic issues and to produce mean minimum, mean and mean maximum air temperature surfaces for the entire Iberian Peninsula. The spatial resolution of the maps is 200 In and their temporal resolution is monthly and annual. An associated error level obtained through validation tests with independent data is computed for each map. These validation tests show that the best results have an average R-2 (mean of all months) of 0.91 (mean temperature), 0.87 (mean maximum temperature) and 0.83 (mean minimum temperature). When the objective is to map a large area, best results are obtained when the model uses all stations of the Iberian Peninsula together (general peninsular model) and not when using different local subsets (hydrographical basins regional models). However, when the objective is mapping a regional area, basin models produce better outcomes than the general model being applied to these regional areas. Copyright (C) 2007 Royal Meteorological Society