A global dataset of vegetation fluorescence emission efficiency derived from TROPOMI sun-induced chlorophyll fluorescence and MODIS reflectance

1. Datasets Summary Vegetation is a crucial component of the ecosystem, regulating the exchange of carbon and water between the surface and the atmosphere through biological processes such as evaporation and photosynthesis. However, a reliable global product for vegetation photosynthetic physiology was still lacking. In this study, we developed a global vegetation fluorescence emission efficiency (eF) by combining TROPOMI Sun-induced chlorophyll fluorescence and MODIS reflectance observations. The eF data were computed by normalizing SIF using absorbed photosynthetically active radiation (iPAR) and canopy greenness (fluorescence-corrected vegetation index, FCVI), providing a proxy related to vegetation photosynthetic physiology. The eF product was provided at a global scale, with a daily temporal resolution and a spatial resolution of 0.05°. The dataset spanned from 1 May 2018 to 29 February 2024, following the availability of high-quality TROPOMI SIF observations. In addition to the original eF product, a temporally smoothed eF dataset was also provided to facilitate the analysis of seasonal dynamics and long-term trends. 2. Data Generation Method The fluorescence emission efficiency is computed as: eF=πSIF/(iPAR×FCVI) FCVI= R_nir- R_(vis) ̅ where: SIF is sun-induced chlorophyll fluorescence derived from TROPOMI, R_nir is the reflectance values at the NIR band. R_(vis) ̅ is the broadband reflectance in the visible region, over the 400 ~ 700 nm range. MODIS reflectance value at band2 (841 ~ 876 nm) for R_nir. The average reflectance values of band1 (620 ~ 670 nm), band3 (459 ~ 479 nm), band4 (545 ~ 565 nm) to approximate R_(vis) ̅ . IPAR is the instantaneous incident photosynthetically active radiation calculated from MCD18C2 product. The MCD18C2 product provided iPAR data at a temporal resolution of 3 hours. Each daily file contained global iPAR fields at eight GMT time points, spaced at three-hour intervals. To ensure temporal consistency with TROPOMI observations, MODIS iPAR values were linearly interpolated to 13:30 local solar time. 3. Temporally Smoothed eF data To accurately characterize the seasonal dynamics, Savitzky-Golay (SG) filter combined with an iterative upper-envelope fitting approach was employed to smooth time series eF. Prior to filtering, spurious high-value outliers in eF, defined as sudden increases exceeding the annual mean value, were removed and linearly interpolated. Then SG filter was used to smooth the eF time series. To account for lower-than-expected observations caused by residual atmospheric effects or undetected clouds and shadows, we applied an iterative upper-envelope fitting method. Specifically, we iteratively replaced data points falling below the fitted curve with the SG-fitted values. The process continued until convergence, defined as when over 90% of the values met the criterion that the absolute difference between successive iterations was below a specified threshold of 1×10^(-6) . Convergence was typically achieved within 30 iterations, which was set as the maximum to avoid overfitting. 4. File Format and Naming Convention All data were provided in NetCDF-4 (.nc) format. For ease of access and distribution, the daily files are organized into yearly .ZIP archives. Each archive contains the individual daily global files for that specific year. The archive named as: Global_eF_product_YYYY.zip Within each ZIP archive, the daily global files follow the convention: Global_eF_YYYYMMDD.nc (Where YYYY, MM, and DD represent the year, month, and day, respectively.)