A Minimum Crossed Entropy Model to Generate Information at Disaggregated Level

摘要

Montado ecosystem in the south of Portugal, integrated in Mediterranean ecosystems, has been extensively used by man for agro-forestry-extensive grazing systems. Sustainable development of this Montado ecosystem demands a long-term perspective about the consequences of today's activities, but a definition of the homogeneous units among this heterogeneous ecosystem is absolutely needed. This paper demonstrates the adherence of the minimum crossed entropy model as a process conducting to dynamic spatial information generation and disaggregation. With this model, it will be possible to use the disposable information to generate data disaggregated to homogeneous units that can be used on the estimation of these territories' soil occupation. The results quality will be evaluated using the weighted prescription absolute deviation (DAPP) and measuring the quantity of information that is changed due to aggregation process by the use of a "Disagegation Information Gain" (DIG), based on the cross entropy between the observed values of soil occupation at aggregated and disaggregated level and on the crossed entropy between the soil occupation estimated by the disaggregation model and those observed at HSU disaggregated level.