Plant Phenology Index leveraging over conventional vegetation indices to establish a new remote sensing benchmark of GPP for northern ecosystems

Abstract

Northern ecosystems, encompassing boreal forests, tundra, and permafrost areas, are increasingly affected by the amplified impacts of climate change. These ecosystems play a crucial role in determining the global carbon budget. To improve our understanding of carbon uptake in these regions, we evaluate the effectiveness of employing the physically-based Plant Phenology Index (PPI) to estimate gross primary productivity across ten different ecosystems. Based on eddy-covariance measurements from 65 sites, the vegetation index (VI)-driven GPP models (six different algorithms) are calibrated and validated. Our findings highlight that the Michaelis–Menten algorithm has the best performance and PPI is superior to the other five VIs, including NDVI, NIRv, EVI-2, NDPI, and NDGI, at predicting gross primary productivity (GPP) rates on a weekly scale (with an average R${}^2$ of 0.64 and RMSE of 1.70 g C m${}^{-2}$ d${}^{-1}$), regardless of short-term environmental constraints on photosynthesis. Through our scaled-up analysis, we estimate the annual GPP of the vast 37 million km${}^2$ study region to be around 22 Pg C yr${}^{-1}$, aligning with other recently developed products such as GOSIF-GPP, FluxSat-GPP, and FLUXCOM-X GPP. Derived from a climate-independent approach, the PPI-GPP product offers distinct advantages in exploring relationships between climate variables and terrestrial ecosystem productivity and phenology. Furthermore, this product holds significant value for assessing forestry and agricultural production in northern regions and for benchmarking terrestrial biosphere models and Earth system models.