Predicting spanish sclerophyllous forests potentiality using artificial neural networks

Abstract

Holm oak and cork oak forests are between the most important sclerophyllous formations in the Mediterranean Iberia. In order to study their potentiality, an artificial neural network model, with a feedforward BP algorithm, has been applied. The elevation, continentality, insolation, annual rainfall, annual mean temperature, mean temperature of the coldest month and mean temperature of the warmest month are the used bioclimatic variables with a 10 km resolution. The neural networks seem a highly predictive powerful tool. Different patterns in the response of the studied forests have been shown. The holm oak presents a continuous and wide potential simulate range. Meanwhile the cork oak potential area is fragmented and restricted, in accordance with its actual distribution area. The lack of both forests in the eastern and southern warm zones of Iberian Peninsula is the main discrepancy with previous potential vegetation proposals.