Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Master → Chili → Generación de una metodología de caracterización remota de suelos mediante un modelo predictivo, Región de Coquimbo, Chile

Universidad de Chile (2015)

Generación de una metodología de caracterización remota de suelos mediante un modelo predictivo, Región de Coquimbo, Chile

Macari Rosales, Orlando Antonio ;

Titre : Generación de una metodología de caracterización remota de suelos mediante un modelo predictivo, Región de Coquimbo, Chile

Auteur : Macari Rosales, Orlando Antonio ;

Université de soutenance : Universidad de Chile

Grade : Ingeniero Agrónomo 2015

Résumé
Remote soil classification is a technique that can be useful to generate an accurate approximation of the reality of this resource, especially in spatial conditions where information is scarce, biased, or nonexistent. This is the case of the Coquimbo Region, Chile, where a large area has not been classified in detail about their soils ; such that only in their agricultural valleys are soil survey studies. In this context, this work proposes soils classify of the Coquimbo Region according to Land Capability Classes system. To this goal, Geographic Information Systems (GIS) and Remote Sensing were used to generate the inputs needed to run a classification. The considered soil formation factors with this purpose were : Parental material, for which we used specialized geology information of the area ; Climate, in terms of rainfall and temperature data ; and Relief, based on landscape landforms, were considered using the Topographic Position Index (TPI) method. With this information, it was proceeded to design and implement a technical expert decision model, which uses expert knowledge in a discipline to generate the decision rule. The model resulted in a classification which covered 93.09% of the region, where the remaining 6.91% was considered inclusions, because representing spatially disaggregated units out of context. The classification included two categories with soils that were not associated to a particular Land Capability Class, but were included in a range : Classes I to IV category and Classes VI to VIII category. On the total surface classified as Land Capability Classes, the results were as follows : 0.62 % was classified as Class I ; 0.74% was Class II ; 1.05% was Class III ; 5.06% was Class IV ; 11.63% was Class VI ; 36.22% was Class VII ; 26.33% was Class VIII ; 6.61% was Classes I to IV category and 4.27% to Class VI to VIII category. Finally, a validation was performed based on expert criteria, evaluating seven specific points. Four points were not in agreement with the original result, which was due to two of the latter were within inclusions (San Julian and Maitencillo points) and a third point was a condition with chemical limiting condition (Huentelauquén point), outside the scope of model prediction.

Présentation

Version intégrale (3,9 Mb)

Page publiée le 8 mai 2022