Modelització climàtica mitjançant tècniques SIG i la seva aplicació a l'anàlisi quantitativa de la distribució d'espècies vegetals a l'Espanya peninsular

Abstract

Developing a numerical model to interpolate climatic data, using this information for building a fitogeographic model and mapping these numerical models are the main objectives of this thesis. This work proposes an empirical methodology to model and map some monthly and annual climatological variables (mean minimum temperature, mean temperature, mean maximum temperature and precipitation). The method consists in a spatial interpolation of the information available in the meteorological stations based in a multiple regression analysis where the independent variables are the geographical ones (altitude, latitude, continentality and solar radiation). Moreover, this methodology had been implemented in a Geographical Information System. Obtaining objective maps with a known error level is an interesting information that in addition have been compared with other interpolation techniques as classical (contour lines) and geostatisitcal (kriging). For mapping the model we have reproduced the equation of the multiple regression analysis using map algebra. Previously, however, we have built a raster matrix (map) of each one of the independent variables. We have worked with a 200 m resolution for the whole peninsular Spain. The obtained maps (potential maps) and have been corrected using the residual error of the regression analysis. These correctors have been interpolated for obtainig the anomaly maps. Using this anomaly maps we can correct the potential maps to obtain the real maps. We have used 60% of the stations to build the model and the remaining 40% for validation purposes. Nevertheless, the final maps have been built using the whole set of stations. The results are very interesting because we obtain values (R2 of the test) between 0.81-0.89 as mean values for the different tested models. We have used climatological maps, orographic maps (altitude and slope) and information about the distribution of vegetal species (plots of the second National Forestry Inventory) to build the ecological range of the most important trees of the peninsular Spain. From the ecological ranges we have calculated the descriptive statistics (mean and standard deviation) that allow us to obtain three different levels: range within one standard deviation (68% of the plots), two standard deviation (95%) and minimum-maximum (100%). For each involved variable we have reclassified the raster matrix with values of 1 (inside the range) and 0 (outside the range). Finally, using map algebra, we have obtained maps of potential distribution of vegetal species based in a suitability index (number of coincidences). Also, we have developed some applications of the climatological and fitogeographical models, being the most interesting one the prediction of the changes in species distribution assuming a climatic change.

Date of Award	4 Apr 2001
Original language	Undefined/Unknown
Supervisor	Xavier Pons Fernandez (Director) & Joan Roure Nolla (Director)