eLUE-GPP (MODIS): A global gross primary productivity product based on ecosystem light-use-efficiency model and MODIS EVI

Gross Primary Productivity (GPP) represents the cumulative amount of
carbon dioxide (CO2) assimilated by green plants through photosynthesis at
specific time intervals and spatial scales. It is the main component of
the carbon exchange between the terrestrial biosphere and the atmosphere,
and has a major influence on global climate and terrestrial ecosystem
functioning. Over the last two decades, the continuous and reliable
collection of global land surface variables by EOS-MODIS, and the parallel
development of the eddy-covariance flux tower network (FLUXNET) have
enabled the integration of MODIS observations with tower measurements for
the calibration and validation of remote sensing models to obtain global
GPP estimates. Despite the significant progress and success to date,
current remote sensing GPP models based on the light use efficiency (LUE)
concept share several limitations, including the difficulty in accurately
predicting LUE variability and the associated use of land cover maps and
look-up tables for biome specific maximum LUE, further down-regulated by
coarse resolution interpolated meteorological data, which introduce
significant uncertainties in the predicted GPP. To address the above
limitations, here we applied a simple yet ecologically sound remote
sensing GPP model based on the ecosystem light use efficiency (eLUE)
concept, using the more than two decades of global MODIS Enhanced
Vegetation Index (EVI) product and the publicly available FLUXDATA2015
dataset, to generate a global 5 km, 16-d GPP product (eLUE-GPP) from
February 2000 to December 2024. Cross-validation with 120 flux tower sites
(952.66 site/year) showed favorable accuracy of eLUE-GPP (hereafter
GPPeLUE) (R2 = 0.74, RMSE = 2.05 g C m-2 d-1). The uncertainty associated
with GPPeLUE is comparatively lower than that of the other global GPP
datasets (MOD17,  VPM, among others). We have also calculated the
uncertainty analytically for each GPP estimate based on the law of error
propagation, which allows quantification of the error budget in
applications such as Earth system model benchmarking and atmospheric
inversion. Our estimate of global total annual GPP, averaged over the
period 2001-2024, was 135.53±11.03 Pg C yr-1. Furthermore, we found a
significant increasing trend in global total annual GPP at a rate of
0.26±0.06 Pg C yr-1 (p < 0.001) from 2001 to 2024, particularly in
eastern Asia, northern India, Europe, eastern North America, and central
South America. We expect that the eLUE-GPP product will enable a more
accurate diagnostic analysis of the global carbon budget and thus
contribute to climate change research.