Discrimination between climate and human-induced dryland degradation

In this study we present a technique to discriminate between climate or human-induced dryland degradation, based on evaluations of AVHRR NDVI data and rainfall data. Since dryland areas typically have high inter-annual rainfall variations and rainfall has a dominant role in determining vegetation growth, minor biomass trends imposed by human influences are difficult to verify. By performing many linear regression calculations between different periods of accumulated precipitation and the annual NDVI_max, we identify the rainfall period that is best related to the NDVI_max and by this the proportion of biomass triggered by rainfall. Positive or negative deviations in biomass from this relationship, expressed in the residuals, are interpreted as human-induced. We discuss several approaches that use either a temporally fixed NDVI peaking time or an absolute one, a best mean rainfall period for the entire drylands or the best rainfall period for each individual pixel. Advantages and disadvantages of either approach or one of its combinations for discriminating between climate and human-induced degradation are discussed. Depending on the particular land-use either method has advantages. To locate areas with a high likelihood of human-induced degradation we therefore recommend combining results from each approach.