New reconstruction of China's spatio-temporal forest over the past 300 years

Reconstructing long-term forest spatial distribution data is crucial for revealing the long-term ecological effects of land use change and supporting research on ecosystem carbon cycles. To address the limitations of existing statistical reconstruction methods in terms of spatial distribution characterization, and the limited spatiotemporal coverage and difficulty in accurately capturing forest dynamics of remote sensing inversion methods, this study proposes a historical forest reconstruction method that integrates ground survey temporal area and remote sensing spatial information. This method integrates multi-source remote sensing data to construct probabilistic features of forest distribution, improving the reliability of forest spatial distribution identification. It introduces historical cropland and construction land as constraints on the spatial allocation of forest by human land use, reconstructing a dataset of China's annual forest spatial distribution from 1700 to 2020.