Incorporation of ancillary information derived from satellite images applied on environmental variables evaluation

Geostatistical models for spatial prediction are based on observations points. When ancillary information related to the target variable is available, it can be used to improve the estimations. Some sources of information, like satellite images or digital elevation models, provide full information for the whole study area, improving even more the point estimation. It is useful when the target variable is difficult or expensive to sample. Also when is knew the observations contain errors and ancillary information can contribute to estimation procedure. The results are maps more realistic, incorporating physical knowledge about the processes under study. The aim of this work is to show how incorporate ancillary information derived from remote sensing products in spatial prediction model. Here is presented a study case using trends in water table depths as target variable and land use classification derived from Landsat image as ancillary information. The results were evaluated by cross validation and the use of ancillary information contributed to improve the spatial prediction. Key-words – geostatistics, ancillary information, universal kriging, groundwater. Palavras-chave – geoestatística, informação auxiliar, krigagem universal, água subterrânea.