QeRSIF: a quality-enhanced red solar-induced chlorophyll fluorescence dataset for TROPOMI satellite

Solar-induced chlorophyll fluorescence (SIF) is a proxy for monitoring vegetation photosynthesis. Red SIF (RSIF) presents a stronger relationship with gross primary production (GPP) in theory as it is much more closely linked to photosystem II. However, due to its low signal-to-noise ratio in weak atmospheric absorption bands and strong reabsorption effect by chlorophyll within leaves and the canopy, satellite-based retrievals of RSIF remain highly uncertain to date, resulting in limited applications compared to far-red SIF (FRSIF). Motivated by this limitation, we develop a two-step quality enhancement framework to provide a novel quality-enhanced RSIF satellite-derived dataset. First, a data-driven algorithm was applied on TROPOMI to obtain the original RSIF retrievals, followed by spectral denoising based on principal component analysis and machine learning making use of the original radiance spectra. Subsequently, spatiotemporal filtering was implemented by integrating MODIS albedo data and ERA5 radiation variables, yielding a global quality-enhanced red SIF product (QeRSIF) eventually. The comparison results demonstrate that QeRSIF product exhibits markedly improved spatial detail, temporal smoothness, and outlier suppression than the original RSIF. At the global scale, QeRSIF shows substantially stronger correlations with vegetation indices (NDVI and NIRvP) and FLUXCOM GPP. Its mean R² against GPP reaches 0.54, compared to 0.13 for the original RSIF, and it even outperforms the widely used Caltech FRSIF dataset (R² = 0.41). The QeRSIF dataset covers the period from June 2018 to March 2025, gridded at 0.05° spatial resolution with an 8-day temporal compositing period. As a novel global terrestrial satellite-based quality-enhanced RSIF product, QeRSIF provides critical data support for advancing studies of vegetation photosynthesis, global carbon cycling, climate change responses, and agricultural monitoring. The dataset is openly available at https://doi.org/10.5281/zenodo.17460923 (Du et al. 2025) and will be continuously updated at https://data.casearth.cn/thematic/Liulab.