摘要：

Self-organizingap (SOM)nd geographic informationystem (GIS)odels weresedo investigatehe nonlinearelationshipsssociated with geochemical weathering processest local (~100 km~2)ndegional (-50,000 km~2)cales.he dataetonsistedf 1) 22-horizonoil variables: P,, pH,l,otalcidity,a,g, K,otalation exchangeapacity,umf exchangeableases,aseaturation,u, Zn,e,,,n, gammaspectrometry (totalount, potassium,horium,ndranium)ndagneticusceptibilityeasures;nd 2)ixopographic variables: elevation,lope,spect,ydrologicalccumulatedlux,orizontalurvaturend verticalurvature. It isharacterizedt 304 locationsrom quasi-regular gridpacedbout 24 kmcrosshetatef Parana.his dataase wasplit intowoubsets:neornalysisndodeling (274amples)ndhetheror validation (30amples) purposes.heelf-organizingapndlusteringethods weresedo identifyndlassifyheelationsmongolid-phasehemical elementoncentrationsnd GIS derivedopographicodels.heorrelationetween elevationnd k-meanslusterselatedheelative position insideydrologicacroasins, which was interpretedsn expressionfhe weathering processeachingteady-stateonditiontheegionalcale. Locally,hehemical elementoncentrations wereelatedohe verticalurvatureepresentingoncave-convexillslopeeatures, whereoncaveillslopes withonvergentluxendsoeeducing environmentndonvexillslopes with divergentlux,xidizing environments.tochasticross validation demonstratedhatheOM producednbiasedlassificationsnd quantifiedheelativemountfncertainty in predictions.his worktrengthensheypothesishat,t-horizonteady-stateonditions,heerrainorphometry were linked withheoil geochemical weathering inwo-way dependent process:heopographicelief wasactorn environmental geochemistry whilehemical weathering wasorerraineature delineation.

展开