Experimental downscaled TROPOMI SIF dataset for continental Europe

The present dataset represent the attempt done within the Sen4GPP project to produce a prototype downscaled SIF for continental Europe. The objective was to adapt an existing downscaling methodology (Duveiller et al. 2020) and apply it to selected sentinel data in order to downscale TROPOMI SIF data (Guanter et al. 2022) from a 10 km grid to a 1 km grid. The method relies on a locally calibrated model linking fine spatial resolution explanatory variables to the coarse spatial resolution target using a moving window. In this case, the explanatory variables are the Sentinel-3 OLCI green vegetation index (OGVI), and Sentinel-3 SLSTR daytime land surface temperature (LST), which are preprocessed into 8-daily composites by project partner U. of Southampton. For details on the downscaling algorithm, the reader is directed to the ATBD document of the Sen4GPP project.

The dataset covers the TROPOMI period from 2018-05-11 until 2020-12-29 for continental Europe. The data is in provided in sinusoidal projection widely used with the MODIS land products, and it covers the area of the MODIS tiles v2 to v5 and h17 to h20. The dataset is divided in separate NetCDF files, with each file covering the entire spatial domain for a single time slice, and each slice representing a period of 8-days. The main variable of interest in each individual file is the predicted downscaled SIF at 1km (variable name: sif) that is mapped on the main dimensions (easting, northing) that cover 4800 by 4800 pixels of circa 1km in the Sinusoidal projection.