International Journal of Wildland Fire · 2022 · 51 citations · 92 references
ComputingForestryMachine LearningPredictive AnalyticsRemote SensingFire DetectionFire ModelingFire Risk AssessmentGeospatial Machine LearningLand Surface ModelingSupport Vector MachinesArea Under The Receiver Operating Characteristic CurveFire Risk PredictionFiresForest Fire RiskReceiver Operating Characteristic AnalysisSatellite-based Fire MonitoringSupport Vector ClassificationSupport Vector Machine Classification
Increasing numbers and intensity of forest fires indicate that forests have become susceptible to fires in the tropics. We assessed the susceptibility of forests to fire in India by comparing six machine learning (ML) algorithms. We identified the best-suited ML algorithms for triggering a fire prediction model, using minimal parameters related to forests, climate and topography. Specifically, we used Moderate Resolution Imaging Spectroradiometer (MODIS) fire hotspots from 2001 to 2020 as training data. The Area Under the Receiver Operating Characteristics Curve (ROC/AUC) for the prediction rate showed that the Support Vector Machine (SVM) (ROC/AUC = 0.908) and Artificial Neural Network (ANN) (ROC/AUC = 0.903) show excellent performance. By and large, our results showed that north-east and central India and the lower Himalayan regions were highly susceptible to forest fires. Importantly, the significance of this study lies in the fact that it is possibly among the first to predict forest fire susceptibility in the Indian context, using an integrated approach comprising ML, Google Earth Engine (GEE) and Climate Engine (CE).
92
Google Earth Engine: Planetary-scale geospatial analysis for everyone
Noel Gorelick, M. Hancher, Mike Dixon et al. · Remote Sensing of Environment · 2017 · 14K citations · Full text
High-Resolution Global Maps of 21st-Century Forest Cover Change
Matthew C. Hansen, Peter Potapov, Rebecca Moore et al. · Science · 2013 · 11.6K citations