The contribution of remote sensing for the development of a Green-Holistic IoT Platform for Forest Management and Monitoring: Reforestation and Deforestation Modules

Abstract. The Green-HIT project focuses on effective and efficient forest monitoring and management, which holds the promise for climate change mitigation, ecosystem conservation, and biodiversity loss reduction. This project is funded by the Cyprus Research & Innovation Foundation (CODEVELOP-GT/0322) and is currently being implemented in Cyprus. Cyprus is located in the Eastern Mediterranean, an area frequently affected by various incidents that impact the preservation of forests (for example, forest fires, illegal logging, hunting, trespassing, and other activities that are damaging to biodiversity), especially during the summer season. Specifically for forest fires, several factors contribute to the increased risk of fire, such as prolonged drought, hot summers, strong winds, steep forest slopes, and flammable vegetation. Early warning and direct management facilities are paramount to efficiently tackling such disastrous events. To this end, the Green-HIT project aims to develop a holistic IoT platform for supporting productivity, competitiveness, and growth of the economy and the promotion of digital and green technology via forest management and monitoring in a post-pandemic world by (a) offering support for prevention, detection and reaction to forest fires, (b) providing afforestation and/or reforestation recommendations, (c) protecting forests from illegal logging and hunting, (d) monitoring forests and forest areas, and (e) offering forest mapping and inventory facilities by collecting, combining and analyzing field and remotely sensed data. This study will present the deforestation and reforestation module of the Green-HIT platform, which aims to identify and suggest (to relevant authorities), possible areas for reforestation. This module was developed using remote sensing data. Specifically, a change detection technique using the Euclidean distance was used for the identification of deforested areas achieving an Overal Accuracy equal to 67.7 %. Also, for the reforestation module, a multicriteria analysis was applied using several parameters like dNBR, land cover, fire history, soil erosion, etc., using the Google Earth Engine platform. For the purposes of this study, the Argaka fire event was selected to evaluate the accuracy of the developed model.