2024_China_TF-AEPNet_FireProduct

I) DESCRIPTION: We propose a thermal feature-aware attention-embedded pyramid network (TF-AEPNet) for hourly wildfire detection with Himawari AHI data, and employ this model to construct the first deep learning-based hourly 2 km wildfire product that provides the spatial-temporal coordinates of each detected fire pixel across China in 2024. II) DATA FORMATTING The TF-AEPNet wildfire product is stored in "2024_China_TF-AEPNet_FireProduct.xlsx". The file records the year, month, day, hour, and longitude and latitude information of all fire pixels.