Comparative analysis of interpolation methods in the middle Ebro Valley (Spain): application to annual precipitation and temperature

Abstract

This paper analyzes the validity of various precipitation and temperature maps obtained by means of diverse interpolation methods. The study was carried out in an area where geographic differences and spatial climatic diversity are significant (the middle Ebro Valley in the northeast of Spain). Two variables, annual precipitation and temperature, and several interpolation methods were used in the climate mapping: global interpolators (trend surfaces and regression models), local interpolators (Thiessen polygons, inverse distance weighting, splines), geostatistical methods (simple kriging, ordinary kriging, block kriging, directional kriging, universal kriging and co-kriging) and mixed methods (combined global, local and geostatistical methods). The validity of the maps was checked through independent test weather stations (30% of the original stations). Different statistical accuracy measurements determined the quality of the models. The results show that some interpolation methods are very similar. Nevertheless, in the case of precipitation maps, we obtained the best results using geostatistical methods and a regression model formed by 4 geographic and topographic variables. The best results for temperature mapping were obtained using the regression-based method. The accuracy measurements obtained by the different interpolation methods change significantly depending on the climatic variable mapped. The validity of interpolation methods in the creation of climatic maps, useful for agricultural and hydrologic management, is discussed.

Comparison of regression techniques for mapping fog frequency: application to the Aragón region (northeast Spain)

“…We demonstrated that the local and spatially complex nature of fogs makes them more difficult to map than other climatic variables such as precipitation and temperature for which it is common to obtain high R 2 values for regression models, usually higher than 0.7 (Ninyerola et al, 2007a(Ninyerola et al, , 2007b. R 2 values higher than 0.7 have been obtained previously for the Aragón region in monthly regression models of precipitation and temperature (Vicente-Serrano et al, 2003;Vicente-Serrano et al, 2007;López-Moreno et al, 2007). In contrast, fog-frequency models for this region yield R 2 values between 0.48 and 0.72 for the winter months, and only consider non-linear relationships by means of GAMs.…”

“…Accordingly, the modelling of fog frequency is a more complex task than that of modelling other climate variables. Precipitation and mean temperature are usually well modelled in terms of elevation, aspect, influence of water bodies, and latitude (Vicente-Serrano et al, 2003).…”

“…The inclusion of these variables in a geographic information system (GIS) facilitates their calculation and helps to reveal their interrelation with climatic data. The most commonly used variables include those that affect the spatial distribution of climate, including elevation, longitude, latitude, distance to the ocean, and areas of topographic shadow (Ninyerola et al, 2000(Ninyerola et al, , 2007aVicente-Serrano et al, 2003. These variables are usually sufficient in mapping climate variables that show gradual spatial variations; however, for variables that show very large spatial variability and notable local features, such as the frequency of fogs, other spatially varying predictors that record local geographic and topographic characteristics are potentially useful in modelling.…”

“…We used a relatively high value of the tension parameter (400) in order to obtain interpolated residuals close to the values observed at the stations. This approach was employed because it has proved to be more suitable than other local techniques such as inverse-distance weighting or geostatistics (Nalder and Wein, 1998;Vicente-Serrano et al, 2003;Ninyerola et al, 2007a;). The sum of residual interpolation and predicted regression maps modified the initial results of the model, and real values were obtained at the locations of the weather stations:…”

“…Many previous studies have mapped climate parameters such as precipitation (Ninyerola et al, 2000(Ninyerola et al, , 2007aBrown and Comrie, 2002), temperature (Vicente-Serrano et al, 2003;Ninyerola et al, 2007b), evapotranspiration (Martínez-Cob, 1996;Vicente-Serrano et al, 2007), and snow depth (López- Moreno and Nogués-Bravo, 2005). The climate variables that have received the greatest attention are precipitation and temperature, as maps of these variables are of great interest for numerous applications.…”