Downscaling Sun-Induced Chlorophyll Fluorescence (SIF) in the Saihanba Region, China

The Sun-induced chlorophyll fluorescence (SIF) holds significant representativeness in monitoring vegetation structure and physiology, playing a crucial role in tracking vegetation changes. However, the limited availability of high spatial resolution SIF severely constrains its application in local areas. To address this issue, this study employs a downscaling approach based on the extreme gradient boosting (Xgboost) model to generate high-resolution SIF data. Choosing the Saihanba region in China as our experimental area, we first reconstruct high-quality Sentinel-2 Vegetation Indices (VIs) time-series data for the Saihanba area. These time-series VIs data serve as the primary parameters for the downscaling of SIF. Subsequently, leveraging the constructed Xgboost model, we successfully downscaled the coarse resolution SIF (0.05°) to a higher resolution (0.0005°). To validate the accuracy of the downscaled SIF data, we utilized the gross primary productivity (GPP) from two flux sites within the research area for verification. The results indicate a high level of precision in the downscaled SIF. Through this study, we have effectively generated precise SIF data for the Saihanba region, offering a convenient and reliable method for monitoring vegetation in local areas.