Evaluating the contribution of auxiliary observations for climate mapping: a case study in the Guadalquivir region

Abstract

In order to improve the detection of climate variability patterns, we assess the impact of incorporating auxiliary meteorological networks into the AEMET network to produce temperature and precipitation maps in the Guadalquivir basin. We employ multiple linear regression and residual spatial interpolation to create monthly precipitation and air temperature (mean minimum, mean and mean maximum) maps using two different datasets: Official (only AEMET stations) and Extended (both AEMET and auxiliary network data). Comparison of the performance of both datasets focuses on three key indicators: adjusted R2, cross-validation measured with RMSE, and the percentage of significant independent variables. Overall, the results indicate that the inclusion of auxiliary networks did not consistently or significantly enhance regression models or reduce map inaccuracies. The extended dataset shows a slight decline in adjusted R2 for most of the variables, with a maximum decrease of 0.082 in R2. However, it allowed for the inclusion of more independent variables in the regression models. Notably, altitude, distance to the Atlantic Ocean, and distance to the Mediterranean Sea emerged as crucial predictor variables for both precipitation and temperature. The impact of auxiliary networks on the error metric lacked a consistent pattern. They led to decreased RMSE values for only half of the variables, with a maximum improvement of 1.24 d°C for temperature models and 6.27 dmm for precipitation models when using the extended dataset.