Aplicabilidad del enfoque de conjuntos borrosos a la clasificación de suelos de la depresión del lago de Valencia, Venezuela

conventional soil classifications produce discrete groups with sharp boundaries, but soil variation is continuous. the fuzzy set theory offers an opportunity to overcome this limitation, expressing the degree of membership of each individual to every class by means of a function which varies continually between 0 (it does not belong) and 1 (it belongs). this study applies the fuzzy set theory to classify soils from the valencia lake basin, venezuela, and evaluates the results. data from 339 soil profiles, interpolated at 15 and 50cm depth, were used. the following variables were selected by means of a principal component analysis for their greater contribution to the whole variation: sand and clay percentages, ph, electrical conductivity and caco3 equivalent percentage. additionally, the following morphological attributes were considered: soil horizon sequence (a/c, a/bw/c or a/bt/c), soil drainage class, presence or absence of vertic properties and presence or absence of duplex (layers of contrasting soil textures). the fuzzy c-means with extragrades algorithm, available in the fuzme program, was applied. soil data were clustered into 9 typical classes plus one extragrade class. the within-class variability is low, except for the percentage of soil organic c and electrical conductivity at 15cm. the base saturation and the cation exchange capacity, as laboratory variables from special analyses, are predictable by the fuzzy soil classes. finally, an association was found between the fuzzy soil classes and landscape units, suggesting that these classes are amenable of mapping.