查看文章

Forest fires are a very common in India, especially in the hilly regions of the western and northeastern Himalayas, which puts adverse impacts on the environment and society. Himachal Pradesh and Uttarakhand are the states most prone to forest fires; therefore, this research is intended to map forest fire occurrence in both states using geospatial techniques and machine learning algorithms (MLAs). To fulfill this objective, we used the meteorological data such as evapotranspiration, precipitation, temperature and wind speed data regarding aridity, elevation, slope, aspect, curvature and land use/land cover (LULC) data; and three MLAs: support vector machine (SVM), random forest (RF), and logistic regression (LR) along with an ensemble learning model were used for the modeling of susceptibility of forest fires. This produced forest fire susceptible maps that were finally validated …