A Machine Learning Approach for Mapping Forest Vegetation in Riparian Zones in an Atlantic Biome Environment Using Sentinel-2 Imagery

Danielle Elis Garcia Furuya,Mayara Maezano Faita Pinheiro,Danillo Roberto Pereira,Michelle Taís Garcia Furuya,Ana Paula Marques Ramos,João Alex Floriano Aguiar,Veraldo Liesenberg,José Marcato Junior,Jonathan Li,Wesley Nunes Gonçalves,Nayara V Estrabis,Lucas Prado Osco

doi:10.3390/rs12244086

Abstract

Riparian zones consist of important environmental regions, specifically to maintain the quality of water resources. Accurately mapping forest vegetation in riparian zones is an important issue, since it may provide information about numerous surface processes that occur in these areas. Recently, machine learning algorithms have gained attention as an innovative approach to extract information from remote sensing imagery, including to support the mapping task of vegetation areas. Nonetheless, studies related to machine learning application for forest vegetation mapping in the riparian zones exclusively is still limited. Therefore, this paper presents a framework for forest vegetation mapping in riparian zones based on machine learning models using orbital multispectral images. A total of 14 Sentinel-2 images registered throughout the year, covering a large riparian zone of a portion of a wide river in the Pontal do Paranapanema region, São Paulo state, Brazil, was adopted as the dataset. This area is mainly composed of the Atlantic Biome vegetation, and it is near to the last primary fragment of its biome, being an important region from the environmental planning point of view. We compared the performance of multiple machine learning algorithms like decision tree (DT), random forest (RF), support vector machine (SVM), and normal Bayes (NB). We evaluated different dates and locations with all models. Our results demonstrated that the DT learner has, overall, the highest accuracy in this task. The DT algorithm also showed high accuracy when applied on different dates and in the riparian zone of another river. We conclude that the proposed approach is appropriated to accurately map forest vegetation in riparian zones, including temporal context.

Highlights

Monitoring the spatial-temporal dynamics of land cover and use in riparian zones is essential to understand the numerous surface processes that can occur in these areas [1]
Our particular interest was to investigate the potential of machine learning algorithms and measure their variation in performance when applied to Sentinel-2 images for this task in riparian zones
Studies [4,15] have shown that Machine learning (ML) techniques are an efficient approach to map different land use and land cover classes, including forest vegetation, and our trials demonstrated that some algorithms can perform this task with higher accuracy than the others

Summary

Introduction

Monitoring the spatial-temporal dynamics of land cover and use in riparian zones is essential to understand the numerous surface processes that can occur in these areas [1]. Deforestation and inadequate use are some of the most notorious problems in many environmentally fragile riparian zones. These regions have an important role in environmental conservation, providing multiple ecosystem services [2]. With the increasing loss worldwide of wetlands and riparian areas [2], an accurate mapping of forest vegetation is required to define strategies for both monitoring and conservation. Riparian zones offer an ecological function essential to wildlife and human communities that are gathered around its proximities. In this regard, investigating methods that provide an accurate description of these forest fragments is a relevant scientific task

Results

Discussion

Conclusion