黑龙江流域10米分辨率森林火烧迹地数据集(2020-2023年)

本数据集以Sentinel-2遥感影像为核心数据源，利用优选光谱特征识别潜在火烧迹地，并融合活跃火数据与增强型火烧迹地指数（EBAI）进行数据精炼，最后结合时空一致性后处理算法，生产了2020—2023年黑龙江流域逐月森林火烧迹地产品（ARB_FBA）。利用独立参考单元及多种主流公开火烧迹地产品进行对比验证，结果表明：（1）ARB_FBA数据集的错分误差介于7.65%—8.46%之间，漏分误差为9.43%—13.02%，Dice系数高达89.94%—91.21%。（2）与GABAM、FireCCI51及MCD64A1产品相比，该数据集在空间细节刻画上更为准确，有效提升了对小规模火烧斑块的捕捉能力。该数据集涵盖 2020—2023 年黑龙江流域逐月森林火烧迹地信息，存储格式为 SHP，由 4 个数据文件组成，数据总量为 336 MB。

This dataset takes Sentinel-2 remote sensing images as the core data source, identifies potential burned areas using optimized spectral features, fuses active fire data and the Enhanced Burned Area Index (EBAI) to refine the dataset, and finally combines spatiotemporal consistency post-processing algorithms to produce the monthly forest burned area product (ARB_FBA) of the Heilongjiang River Basin from 2020 to 2023. Using independent reference units and multiple mainstream publicly available burned area products for comparative validation, the results show that: (1) The commission error of the ARB_FBA dataset ranges from 7.65% to 8.46%, the omission error is between 9.43% and 13.02%, and the Dice coefficient is as high as 89.94% to 91.21%. (2) Compared with GABAM, FireCCI51 and MCD64A1 products, this dataset is more accurate in spatial detail depiction and effectively improves the detection capability for small-scale burned patches. This dataset covers the monthly forest burned area information of the Heilongjiang River Basin from 2020 to 2023, stored in SHP format, consisting of 4 data files with a total data volume of 336 MB.