Real-Time Wildfire Detection Algorithm Based on VIIRS Fire Product and Himawari-8 Data

Wildfires have a significant impact on the atmosphere, terrestrial ecosystems, and society. Real-time monitoring of wildfire locations is crucial in fighting wildfires reducing human casualties property damage. Geostationary satellites offer advantage high temporal resolution are gradually being used for real-time fire detection. In this study, we constructed label dataset using stable VNP14IMG product random forest (RF) model detection based Himawari-8 multiband data. The band calculation features related brightness temperature, spatial features, auxiliary data as input framework training. We also recursive feature elimination method to evaluate these accuracy exclude redundant features. daytime nighttime RF models (RF-D/RF-N) separately analyze their applicability. Finally, extensively evaluated performance by comparing them with Japan Aerospace Exploration Agency (JAXA) product. exhibited higher accuracy, recall precision rates 95.62% 59%, respectively, rate small fires was 19.44% than that JAXA Adding well selection, effectively reduced overfitting improved model’s generalization ability. RF-D had RF-N model. Omission errors commission were mainly concentrated adjacent pixels clusters. conclusion, our VIIRS data-based can monitor location real time has excellent capability fires, making it highly