Assessment of k-Nearest Neighbor and Random Forest classifiers for mapping forest fire areas in central Portugal using Landsat-8, Sentinel-2, and Terra Imagery

• Published in: Remote sensing (Basel, Switzerland) Vol. 13; no. 7; pp. 1 - 25
• Main Authors: Pacheco, Admilson da Penha, Junior, Juarez Antonio da Silva, Ruiz-Armenteros, Antonio Miguel, Henriques, Renato F.
• Format: Journal Article
• Language: English
• Published: Basel Multidisciplinary Digital Publishing Institute (MDPI) 01.04.2021; MDPI AG
• Subjects: Algorithms; ASTER; Biodiversity; Burned; Classification; Classifiers; Climate change; Conservation areas; Desertification; Fires; Forest & brush fires; Forest fires; k-Nearest Neighbor; Landsat; Landsat 8; Landsat satellites; Mapping; MODIS; Random Forest; Remote monitoring; Remote sensing; Satellite imagery; Science & Technology; Sensors; Sentinel 2; Spatial discrimination; Spatial resolution; Terra; Vegetation; Portugal
• ISSN: 2072-4292; 2072-4292
• DOI: 10.3390/rs13071345

Forest fires threaten the population’s health, biomass, and biodiversity, intensifying the desertification processes and causing temporary damage to conservation areas. Remote sensing has been used to detect, map, and monitor areas that are affected by forest fires due to the fact that the different areas burned by a fire have similar spectral characteristics. This study analyzes the performance of the k-Nearest Neighbor (kNN) and Random Forest (RF) classifiers for the classification of an area that is affected by fires in central Portugal. For that, image data from Landsat-8, Sentinel-2, and Terra satellites and the peculiarities of each of these platforms with the support of Jeffries–Matusita (JM) separability statistics were analyzed. The event under study was a 93.40 km 2 fire that occurred on 20 July 2019 and was located in the districts of Santarém and Castelo Branco. The results showed that the problems of spectral mixing, registration date, and those associated with the spatial resolution of the sensors were the main factors that led to commission errors with variation between 1% and 15.7% and omission errors between 8.8% and 20%. The classifiers, which performed well, were assessed using the receiver operating characteristic (ROC) curve method, generating maps that were compared based on the areas under the curves (AUC). All of the AUC were greater than 0.88 and the Overall Accuracy (OA) ranged from 89 to 93%. The classification methods that were based on the kNN and RF algorithms showed satisfactory results. Research was supported by PAIUJA-2019/2020 and CEACTEMA from University of Jaen (Spain), and RNM-282 research group from the Junta de Andalucia (Spain). Special thanks to the four anonymous reviewers for their insightful comments.