Monitoring of Forest Changes in Mount Kenya National Park Based on Domestic High-Resolution Satellites

Abstract. With the increase in population and the growing demand for timber, especially fuelwood, Kenya's forests are facing the threat of serious deforestation and illegal logging activities. At present, remote sensing has become an important means of monitoring the dynamic changes of forest resources with its advantages of wide monitoring range, fast speed and low cost. This paper selected Mount Kenya National Forest Park as the study area, employing domestic 2-meter resolution satellite images (Gaofen-1 satellite constellation, Gaofen-6 and the Ziyuan-3 satellites) to conduct dynamic forest change monitoring from 2019 to 2023. The optimized semantic segmentation model DeepLabv3+, which is based on the Pytorch deep learning framework, was used to achieve fine segmentation of forest element boundaries via a spatial pyramid pooling module and an encoder-decoder architecture. The method of manual annotation was used to mark the forest changes and their causes from 2019 to 2023. The newly added forest was mainly restored by artificial planting, covering an area of 3078.71 ha. The total area of forest reduction was 2,425.91 ha. There were 542 patches of forest land occupied by new arable land, covering an area of 1,613.41 ha. It was the most important cause for the forest reduction. 327 patches of forest reduction were due to artificial logging, covering an area of 408.34 ha, which was the secondary cause of forest loss. Human activities were more constrained at higher altitudes, hence, the most of planting and lost forest patches usually occurred below 2,600 meters above sea level. Effective implementation of environmental protection policies was an important reason for the emergence of large areas of new plantation while the occupation of arable land and logging were main factors for the forest reduction in Mount Kenya.