Abstract

POVEDA-SOTELO, Yoan; BERMUDEZ-CELLA, Mauricio A.  and  GIL-LEGUIZAMON, Pablo. Evaluation of supervised classification methods for the estimation of spatiotemporal changes in the Merchán and Telecom paramos, Colombia. Bol. geol. [online]. 2022, vol.44, n.2, pp.51-72.  Epub July 18, 2022. ISSN 0120-0283.  https://doi.org/10.18273/revbol.v44n2-2022002.

In recent years there has been significant progress in Geographic Information Systems (GIS) and the development of supervised classification methods, but until now these had not been used to accurately calculate the surface extent of paramos in sectors of the Eastern Cordillera of Colombia. Furthermore, these methods had not been used to estimate the distance between the boundaries of these moors and major geological features. For this reason, in the present research, five different supervised classification methods were evaluated, with the purpose of determining which of them has a higher resolution in order to reproduce the extension and surface distribution of the paramos of Merchán and Telecom in Saboyá, Boyacá, belonging to the Merchán - Iguaque complex in the Eastern Cordillera of Colombia. For this purpose, satellite images of the study area by Landsat 8 for the year 2018 were chosen and classified into some algorithms based on Machine Learning (SVM, RF, DT, BC and ANN). To establish the accuracy and reliability of the classification data of the terrain features, the Kappa Index was calculated, which allowed determining that the most accurate method for this case was Random Forest. In addition, since the boundaries of the moorlands coincide with geological structures or contacts between formations, the distance between the edge of the moorlands and these features was estimated. The results obtained in this research are considered as an important input for future multitemporal analysis as in landscape metrics, which serve as a tool for the development and decision making in the management of natural resources, biodiversity, provision of ecosystem services, as in the land use planning for the municipality of Saboyá-Boyacá.

Keywords : Supervised classification; Kappa Index; Machine Learning; Paramo; Remote sensing.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )