Bioclimatological mapping tackling uncertainty propagation: application to mainland Portugal

We present a bioclimatological diagnosis of mainland Portugal, namely the thermotype and ombrotype maps following Rivas-Martinez's worldwide bioclimatic classification system. In order to obtain this diagnosis, we produced maps of bioclimatological indices using, as base data, geostatistical interpolations of air temperature and precipitation. We performed uncertainty propagation obtaining uncertainty measures for the produced maps: mean absolute errors and root mean squared errors. For the non-linear indices, besides the usual approximation using Taylor expansion, we devised error formulae, for which we showed that the propagated uncertainties are upper bounds on the true uncertainty measures. We compared the obtained uncertainty measures to those reported on a previously published work, which used a different methodological framework to obtain the same diagnosis. Although the approach we used here implies a great number of interpolations and subsequent calculation steps, it permitted the use of a large amount of data relative to precipitation. An F-test showed that the estimated mean squared errors for the maps of ombrothermic indices were significantly lower than those produced by the former methodological framework.