(2024) 20:57 | Fahad Shahzad1†, Kaleem Mehmood45†, Khadim Hussain3, Ijlal Haidar45, Shoaib Ahmad Anees6, Sultan Muhammad5, Jamshid Ali4, Muhammad Adnan7, Zhichao Wang1* and Zhongke Feng12*
This study investigates the prediction of vegetation fires in Pakistan using machine learning algorithms. The research focuses on forest fires, crop fires, and other vegetation fires, leveraging data from the MODIS Global Fire Atlas, topographic, climatic conditions, and vegetation types between 2001 and 2022. Four models—logistic regression, random forest, support vector machine, and extreme gradient boosting—were tested and selected based on their performance metrics. The models achieved prediction accuracies ranging from 78.7% to 87.5% for forest fires, 70.4% to 84.0% for crop fires, and 66.6% to 83.1% for other vegetation fires. The random forest model showed the highest accuracy and AUC values, making it the most optimal model. The study also generated maps to analyze the risk of vegetation fires in Pakistan, highlighting areas with high, moderate, and low fire risks. The findings provide valuable insights for wildfire management and policy formulation.This study investigates the prediction of vegetation fires in Pakistan using machine learning algorithms. The research focuses on forest fires, crop fires, and other vegetation fires, leveraging data from the MODIS Global Fire Atlas, topographic, climatic conditions, and vegetation types between 2001 and 2022. Four models—logistic regression, random forest, support vector machine, and extreme gradient boosting—were tested and selected based on their performance metrics. The models achieved prediction accuracies ranging from 78.7% to 87.5% for forest fires, 70.4% to 84.0% for crop fires, and 66.6% to 83.1% for other vegetation fires. The random forest model showed the highest accuracy and AUC values, making it the most optimal model. The study also generated maps to analyze the risk of vegetation fires in Pakistan, highlighting areas with high, moderate, and low fire risks. The findings provide valuable insights for wildfire management and policy formulation.