OEMC use case proposes an open workflow that enhances SIF spatial resolution to about 1 km for better GPP flux estimation

Proposes an open workflow that enhances the spatial resolution of SIF estimates to about 1 km by synergistically using other remote sensing sources such as LST and NIRv with a semi-empirical approach. Prior SIF downscaling approaches by Duveiller and Cescatti (2016) and Duveiller et al. (2020) exist; this use case builds on them to develop a first prototype workflow in Julia and advances it as an open tool. The text describes the workflow design and the prior methods it relies on, but does not provide quantitative validation metrics for downscaling accuracy.

Contributed to a new SIF retrieval based on Sentinel-5P TROPOMI data in which machine learning is used in the retrieval itself to enhance retrieval quality and reduce noise, yielding ungridded instantaneous SIF estimates. Introduces machine learning into the SIF retrieval itself rather than only in post-hoc downscaling, improving signal quality at the source. The text states that machine learning is used in the retrieval to enhance quality and reduce noise, but does not provide specific numbers for retrieval accuracy or noise reduction.

Gridded the ungridded instantaneous SIF estimates into a dedicated dataset at 0.05 degrees (about 5 km) with 8-daily frequency for the period 2018-2025, and applied the spatial enhancement tool to it. Provides a gridded SIF dataset with a long time span and fixed frequency as the input basis for the downscaling tool. The text gives the dataset's spatial resolution, temporal frequency, and time range, but does not describe its validation or uncertainty quantification.

Ported the spatial enhancement tool to the Copernicus Data Space Ecosystem using the OpenEO framework, enabling users to perform on-demand spatial downscaling over their target areas based on their downscaling model of choice, with a dedicated notebook illustrating how it can be done. Moves from a local prototype to a cloud-based on-demand service, lowering the barrier for distributed participants such as RECCAP to use the downscaling tool. The text describes the porting goal and the existence of the notebook, but does not report user testing or service performance data.

This work targets participants in regional carbon assessments such as RECCAP, aiming to provide downloadable data subsets and on-demand downscaling tools for scenarios that need higher spatial resolution to discriminate GPP by land cover type, detect disturbances and land cover change, and improve representation of coastal and island regions. The tool is deployed on the Copernicus Data Space Ecosystem via the OpenEO framework, allowing users to process target areas based on their downscaling model of choice. For evaluation, the work matches satellite grid cells to eddy covariance flux site GPP ground measurements and quantifies the effect of spatial heterogeneity, laying the groundwork for subsequent validation of the value of downscaling for GPP estimation.

The text is a project use case description and does not provide quantitative results for downscaling accuracy, the magnitude of SIF retrieval quality improvement, or GPP estimation improvement, so readers still need to watch for subsequent validation studies. The dataset covers 2018-2025, and the end of that time range may involve near-real-time or provisional versions, so version updates should be noted. When matching satellite pixels to eddy covariance sites, how spatial heterogeneity is quantified and how it affects GPP estimation remains to be clarified. In addition, allowing users to choose their own downscaling model means result quality may vary with model choice, and the tool itself does not provide a unified quality control scheme in the text.