XJ-SIF: a reconstructed 500-meter SIF dataset for revealing spatiotemporal vegetation dynamics in Xinjiang, China

However, existing satellite-based Solar-induced Chlorophyll Fluorescence (SIF) products emphasize large-scale spatial continuity, resulting in overestimation of vegetation signals in arid and semi-arid areas and constraining their applicability for fine-scale vegetation dynamics monitoring. In this study, we reconstructed a long-term (2000–2024) and high-spatiotemporal-resolution (500 m, 8-day) SIF dataset for Xinjiang, China(XJ-SIF), by using a weighted stacking ensemble learning framework that integrates a dynamic land cover masking strategy to effectively eliminate background noise from non-vegetated surfaces.

Model validation results show that XJ-SIF achieves superior accuracy, with an R² of 0.922 and an RMSE of 0.131 mW m-2 nm-1 sr-1.