SURVEYING SOIL CHEMICAL WEATHERING IN PARANA STATE/BRAZIL: A DATA MINING-GIS HYBRID APPROACH

Unsupervised artificial neural network (ANN) analysis and geographic information system (GIS) derived models were used to investigate the nonlinear relationships associated with geochemical weathering processes at local and regional scales. The data set consisted of 18 B-horizon soil variables (P, C, pH, Al, total acidity, Ca, Mg, K, cation exchange capacity, base saturation, Cu, Zn, Fe, B, S, Mn, radiometrics and magnetic susceptibility measures) and 5 topographic variables (elevation, slope, hydrological accumulated flux, horizontal curvature and vertical curvature) characterized at 304 locations from a quasi-regular grid spaced about 24 km across the state of Parana. This data base was split into two subsets: one for analysis and modeling (274 samples) and another for validation (30 samples) purposes. The self-organizing maps and clustering methods were used to identify and classify the relations among solid-phase chemical element concentrations and GIS derived topographic models. The correlation between elevation and k-means clusters related the relative position inside hydrologic macro basins, which was interpreted as an expression of the weathering process reaching a steady-state condition at the regional scale. Locally, the chemical element concentrations were related to the vertical curvature representing concave-convex hillslope features, where concave hillslopes with convergent flux tends to be a reducing environment and convex hillslopes with divergent flux, oxidizing environments. Stochastic crossvalidation demonstrated the ability of the ANN model to produce unbiased classifications and quantify the relative amount of uncertainty. This work strengthens the hypothesis that, under B-horizon steady-state conditions, the terrain morphometry were linked with the soil geochemical weathering in a two-way dependent process, the relief was a factor on environmental geochemistry while chemical weathering was for the terrain feature delineation.