Improved estimation of gross primary production with NIRvP by incorporating a phenophase scheme for temperate deciduous forest ecosystems

NIRvP, the product of the remotely sensed near infrared reflectance of vegetation (NIRv) and photosynthetically active radiation (PAR), has been demonstrated to be a good proxy for terrestrial gross primary productivity (GPP) due to their strong theoretical correlation. Currently, a linear relationship with a constant slope parameter (k) is used to estimate GPP with NIRvP. However, this parameter may vary with plant phenology, potentially leading to uncertainty in GPP estimation. In this study, we hypothesized that considering the influence of phenology on k could improve the accuracy of GPP estimation. To test this hypothesis, we used pair measurements of ground eddy covariance-derived GPP and satellite-derived NIRvP across nine temperate deciduous forest sites, and compared the modeling results of two parameterization schemes: one with a constant k throughout the entire growing season (termed the global scheme) and another with dynamic k parameters across different phenophases (e.g., greenup, mature and senescence phases; termed the phenophase scheme). Our results show that relative to the global scheme, the phenophase scheme significantly improved GPP estimation with a reduction in bias by 63.71 ± 16.46%, 53.71 ± 26.17% and 32.96 ± 32.07% in the greenup, maturity and senescence phases across all study sites, respectively. Moreover, the efficiency of this improvement is meaningful in terms of the Akaike Information Criterion (AIC). Further regression analysis revealed that k varies with phenophase, which can be attributed to variations in photosynthetic capacity and canopy optical properties due to leaf aging and canopy structure development. These findings collectively improve our understanding of the influence of phenology on the GPP-NIRvP relationship and emphasize the effectiveness of using the phenophase-dependent k parameterization scheme, which would ultimately contribute to improving the accuracy of GPP estimation in temperate forest ecosystems.