Dependencia espacial de datos topográficos a escala de ladera y pequeña cuenca agrícola

Abstract

[Abstract] Landscapes are characterized by both, random and non-random variability components. Random variability of topographical data, summarized by the semivariance, may be used to elaborate DEMs. The main objective of this work was the study of the spatial dependence pattern of topographical data, using geostatistical techniques. The experimental data sets were directly measured by means of an Abney level in six relief units, hill slopes or elementary first-order small catchments ranging from about 0,62 to 5,72 ha. Forall the six landscape units, the spatial variation of elevation could be expressed as the sum of a deterministic term given by a lineal function and a stochastic component given by spatially correlated height residuals. For the residual elevation data sets, the experimental semivariograms were best fitted by gaussian isotropic models with a small nugget effect. Scaled semivariograms, using the sample variance as scaling factor, allow the comparison of the variability pattern for different landscape units. Cross-validation was used to determine the number of data points for DEM elaboration by block kriging.