Un modelo basado en sensores remotos e inteligencia artificial para la estimación de la degradación del bosque
Fecha
Autores
Autor corporativo
Título de la revista
ISSN de la revista
Título del volumen
Editor
Compartir
Altmetric
Resumen
Forest degradation and deforestation are environmental problems that diminish the provision of ecosystem services. A possible solution for the identification and monitoring of these problems are monitoring systems based on remote sensing; information from active and passive satellites offers a great deal of useful information. Despite this, there is a limitation in tropical areas due to the high presence of clouds throughout the year, which consequently limits the amount of information. Thus, this study proposes a model that makes use of remote sensing and artificial intelligence for the identification of forest degradation with limited information using NDVI, its variation between time windows in the period 1990 - 2019 and machine learning tools. This study was developed in tropical rainforest in the municipality of Mapiripán (Meta, Colombia), an area where problems of illicit crops and deforestation have been identified. To address this problem, a model was proposed based on geographic information systems for the identification of degradation related to deforestation. When comparing 7 machine learning algorithms, it was found that the neural network algorithm of three hidden layers (model with the best performance, 75.25% accuracy) and the second model corresponds to the linear discriminant algorithm (73.25% accuracy) show better performance despite its limited information. Additionally, it was possible to identify that 60% of the deforested areas suffered some degree of intervention that led to deforestation, which could be avoided through a monitoring and early warning system.