Upscaling near-real-time biospheric CO2 fluxes over Europe with a modified Vegetation Photosynthesis Respiration Model (VPRM)

Monitoring ecosystem carbon dioxide (CO₂) exchange is crucial for assessing the impacts of climate extremes and constructing carbon budgets to inform land management and enforce international climate treaties. To this end, we present here gridded hourly ecosystem CO₂ fluxes upscaled from eddy covariance observations at 0.1° × 0.1° resolution and updated at low latency. Sentinel-2 indices are used to drive a modified Vegetation Photosynthesis Respiration Model (VPRM) following Bazzi et al. (2024) with a restructured Ecosystem Respiration equation and explicit soil moisture stress functions. VPRM parameters are optimized to half-hourly eddy covariance Net Ecosystem Exchange (NEE) and Gross Primary Production (GPP) datasets for each of 36 FLUXNET sites. Additionally we modify the temperature dependence of GPP by optimizing minimum and maximum temperatures as parameters and estimating optimum temperatures from mean annual temperature. We find these temperature modifications reduce RMSE for NEE and GPP respectively by 11% and 12% overall, 16% and 18% at evergreen needleleaf forests, 14% and 12% at grasslands, and 12% and 16% at mixed forests. Using site data on meteorology and vegetation, we train a random forest to produce mapped VPRM parameters representing the spatial heterogeneity in ecosystem characteristics. Gridded VPRM NEE estimates are presented based on both modeled parameter maps and multi-site optimizations by plant functional type, and upscaled products can be produced within hours of satellite data availability.

 

[1] Bazzi, H. et al. "Assimilating Sentinel-2 data in a modified vegetation photosynthesis and respiration model (VPRM) to improve the simulation of croplands CO₂ fluxes in Europe." International Journal of Applied Earth Observation and Geoinformation 127 (2024): 103666.