3-11
Forest Fire Risk Assessment and Mapping Using Support Vector Machine Algorithm, A Case Study in Nghe An Province, Vietnam
Authors: Thi Nam Phuong Doan, Van Trung Nguyen, Thi Thanh Hoa Pham, Thanh Ha Tran, Thi Thu Ha Le
Number of views: 23
In recent years, forest fires have occurred frequently in Vietnam due to the influence of climate change and human activities. This paper presents the results of modeling the risk of forest fires in the west of Nghe An province (north-central Vietnam) from remote sensing and GIS data. 09 factors affect the risk of forest fire, including vegetation cover (NDVI index), soil moisture (NMDI index), elevation, slope, aspect, wind speed, land surface temperature, average monthly precipitation and population density are used to build a model for mapping forest fire risk based on Suppor Vector Machine (SVM) algorithm. The past forest fire data is collected from the database of the Forest Protection Department (Vietnam Ministry of Agriculture and Rural Development) to evaluate the accuracy of the model. Different values of cost parameter (C) are tested to select the value with the highest accuracy in predicting forest fire risk. The results obtained in the study can be used effectively for monitoring and early warning of forest fire risk in the localities, helping to reduce damage caused by forest fires.