Using Landsat satellite imagery to deploy the reforestation at Suoi Mo, Luc Nam district, Bac Giang province

DOI: 10.18173/2354-1067.2023-0024

Abstract

Bac Giang is a mountainous province with suitable conditions for the cultivation and development of timber such as acacia, eucalyptus for the needs of market. However, the planted forest area itself also changes according to the growth-logging cycle of each tree species. Therefore, it is necessary to determine the newly harvested forest area for estimating the number of seedlings that must be prepared for replanting in this area in order to optimize the process of planting and exploiting production forests. This study applied several image processing algorithms to increase the accuracy of land covers classification. Bare soil after timber harvesting has been separated from other types of land cover by object-based image classification approach with Support Vector Machines (SVMs) algorithm due to its efficiency and flexibility. The obtained map shows that the area of bare soil after harvesting is 246.6 ha in 2022 corresponding to the number of seedlings needed of 70144 for the new planting season. The validation is done by comparison with classification results using independent samples selected through visual interpretation from high resolution Google Earth image captured during similar period. The confusion matrix shows an adequate confidence with the overall accuracy of 81.38% and a Kappa coefficient of 0.87. Such results can help local authorities propose appropriate solutions and policies for the development of forest resources and socio-economic development in a more harmonious manner in this mountainous area.