ICG2022-14
https://doi.org/10.5194/icg2022-14
10th International Conference on Geomorphology
© Author(s) 2022. This work is distributed under
the Creative Commons Attribution 4.0 License.

Machine learning-based identification of the geomorphological units of the Northeastern sector of the San Luis Province - Argentina

Ulises Rodrigo Magdalena1, Juan Pablo Zbrun Luoni2, Guillermo Ribeiro3, and Raul Reis Amorim4
Ulises Rodrigo Magdalena et al.
  • 1University of Campinas, Geosciences, Geography, Campinas, Brazil (ulisesrodrigo@id.uff.br)
  • 2Universidad Nacional de los Comechingones, Departamento Académico Arquitectura, las Culturas y Arte, Ciudad de Merlo – San Luis, Argentina (jzbrun@unlc.edu.ar)
  • 3Universidad Nacional de Córdoba, Facultad de Ciencias Exactas, Físicas y Naturales, Córdoba, Argentina (guillermoribeiro@gmail.com)
  • 4University of Campinas, Geosciences, Geography, Campinas, Brazil (raulreis@unicamp.br)

Advances in the access and processing of morphometric data from Remote Sensing and associated with the computational progress in terms of data storage and optimization through computational algorithms allow us to determine the boundaries of the geomorphological units demarcated by conventional procedures, enabling the exploration of hypotheses on landscape dynamics. Thus, this manuscript aims to delimit the geomorphological units of the Northwest of the San Luis Province - Argentina, through the Geological Chart 3366-II Villa de Merlo, of the National Program of Geological Charts of the Argentine Republic of the Servicio Geológico Minero Argentino (SEGEMAR). We used the random forest algorithm in this manuscript. We concluded that the results sometimes generalize the geomorphological units bounded by conventional methods. Still, in other cases, these boundaries get refined and allow us to explore the hypothesis of the subdivision of the “Piedemonte Occidental” Unit into “Piedemonte Superior” and “Piedemonte Inferior”. Furthermore, the results point out the shortcomings of the random forest presented as a black-box and eventually overestimate some geomorphological units in incorrect regions. However, this algorithm is still an important support tool for interpreting landscape dynamics.

KEYWORDS: geomorphology, random forest, landscape, spatial planning, digital cartography, remote sensing.

How to cite: Magdalena, U. R., Luoni, J. P. Z., Ribeiro, G., and Amorim, R. R.: Machine learning-based identification of the geomorphological units of the Northeastern sector of the San Luis Province - Argentina, 10th International Conference on Geomorphology, Coimbra, Portugal, 12–16 Sep 2022, ICG2022-14, https://doi.org/10.5194/icg2022-14, 2022.