Modelling of permafrost distribution in the Sierra de Telera (Central Pyrenees, Spain) by using Geographic Information System techniques

This work analyzes the permafrost distribution in a low altitude mountainous area, the northern slope of the Sierra de Telera (Central Pyrenees, Spain), by using a GIS empirical-statistical modelling approach. The ground thermal conditions in the area were determined by conducting 223 BTS measurements. Preliminary exploratory analysis showed that incoming potential short-wave solar radiation may be the main factor explaining permafrost distribution, a variety of models was generated including elevation, snow cover persistence and three different solar radiation inputs (annual, summer and May solar radiation values). Three multiple linear regression analyses were then performed to model the relationship between BTS measurements and the predictor variables. A cartography of the predicted permafrost distribution based on BTS measurements was obtained for each model (MMSR, MASR and MSSR). In order to assess the accuracy of the models, Kappa values and additional quality measures derived from the error matrix were obtained. The three models were statistically significant, but their performance was different: MMSR and MASR predicted moderately well the BTS distribution in the study area (MMSR K = 0.505; MASR K = 0.458) while MSSR predicted BTS location only fairly well (MSSR K = 0.296). The overall accuracy of the models ranged from very good (MMSR and MASR) to good (MSSR). The observations and analysis conducted strongly suggest the importance of solar radiation in order to explain ground thermal conditions in low altitude mountain areas with steep topography such as Sierra de Telera, placed well below the regional permafrost limit.