Evaluation of Fire Products Using Spatio-Temporal Clustering Method

With the deepening understanding of fire research and the rapid development of remote sensing technology, remote sensing data has become the main basis for fire monitoring. This study aimed to test the accuracy of four selected fire products based on fire events in Sichuan Province, including VNPI4IMGTDL_NRT (hereinafter referred to as VNPI4DL), MCD64Al, Fire_CCI, MCDI4ML. The reference data is provided by Sichuan Forestry and Grassland Bureau, and the fire products are verified by the spatio-temporal cluster analysis method and classification accuracy evaluation method. The results show VNP14DL has the highest fire detection accuracy, with Fl-score reaching 0.490 in 2014, followed by MCD14ML (Fl-score=0.457), MCD64Al (Fl-score=0.390), and Fire_CCI (Fl-score=0.330).