Improved GPP upscaling from instantaneous measurements to daily sums using the light-use-efficiency-based model

The rapid development of satellite technology has greatly contributed to the estimation of global terrestrial gross primary productivity (GPP). However, most satellites only obtain instantaneous photosynthetic signals at overpass time. Given the demand for all-day GPP in application studies, this leads to the problem of gaps. Therefore, a temporal upscaling process is needed for daily GPP estimation. Most traditional methods consider the change in the total amount of incident light during a day while ignoring the impact of light use efficiency (LUE) on the diurnal trajectories of GPP, which produces uncertainties in the temporally upscaled results. In this study, using simulations from the SCOPE model and tower-based measurements, we find that the diurnal shape of GPP trajectories is dependent on the latitude of site, DOY, and LAI. By investigating how the canopy structure (LAI employed in this study) influences the light response of LUE and subsequently affects diurnal variations in GPP, we proposed a temporal upscaling method considering the diurnal variation in LUE at the canopy scale. To account for the impact of varying temperature and moisture throughout the day, an asymmetric correction factor was also included. Compared with previous upscaling schemes based on the cosine of SZA, PAR, PAR×NDVI, etc., our LUE-based method showed higher R² and lower RMSE values on both simulations (R² of 0.993, ∆ R² up to 0.008; RMSE of 0.274 gC m⁻² d⁻¹, ∆ RMSE up to 0.058 gC m⁻² d⁻¹,which is 1.763 % of the mean simulated GPP) and tower-based observations (average R² of 0.970, ∆ R² up to 0.09; average RMSE of 0.453 gC m⁻² d⁻¹, ∆ RMSE up to 0.410 gC m⁻² d⁻¹, which is 7.70 % of the mean observed GPP). Moreover, it outperforms traditional methods in low LAI conditions and at low-latitude sites where upscaling is more challenging. Therefore, the LUE-based method exhibits higher robustness, smaller disparities across sites, and less dependence on the time window, which is of great importance for the better estimation of daily GPP from instantaneous satellite observations.