• 514 sites-years of flux data were used to analyze the potential of physiological and phenological metrics in explaining the variability of forest NEP; • Summer physiological metrics performed better than phenological metrics in explaining IAV of NEP; • Ecosystem respiration played an important role in controlling the variability of NEP in forest ecosystem; • MCUI exhibited a great potential in explaining both IAV and SV of NEP. Understanding the feedback of ecosystem carbon uptake on climate change at temporal and spatial scales is crucial for developing ecosystem models. Previous studies have focused on the role of spring and autumn phenology in regulating carbon sequestration in forest stands, but few on the impact of physiological status in summer. However, plant accumulated the most carbon in summer compared with spring and autumn, therefore, it is of great significance to explore the role of summer phenological metrics on the variability of carbon sequestration. Using 514 site-years of flux data obtained at 40 FLUXNET sites including three forest ecosystems (i.e. evergreen needleleaf forest (ENF), deciduous broadleaf forest (DBF) and mixed forest (MF)) in Europe and North America, we compared the potential of physiological and phenological metrics of Gross Primary Production (GPP) and Ecosystem Respiration (RECO) in explaining the interannual and spatial variability (IAV and SV) of forest net ecosystem production (NEP). In view of the better performance of physiological metrics, we developed the maximum carbon uptake index (MCUI), which integrated the physiology metrics of photosynthesis and respiration in summer, and further explored its ability in explaining the IAV and SV of NEP. The results suggest that the MCUI had a better ability than respiration-growth length ratio (RGR) in predicting NEP for all three forest types. The interpretation of MCUI based on meteorological variables illustrated that the controlling meteorological factors of MCUI differed substantially among ecosystems. The summer shortwave radiation had the greatest influence on MCUI at DBF sites, while the soil water content played an important but opposite role at ENF and DBF sites, and no significant meteorological driver was found at MF sites. The higher potential of MCUI in explaining IAV and SV of NEP highlights the importance of summer physiology in controlling the forest carbon sequestration, and further confirms the significant role of peak plant growth in regulating carbon cycle of forest ecosystems. Understanding the drivers of peak plant growth is therefore of a great significance for further improving the precious of ecosystem model in the future.

• PhenoCam data at 13 sites were used to analyze its potential of phenology modeling. • GCC and RCC performed well in capturing GPP-based SOS and EOS at DBF sites. • RCC showed unrecognized importance than GCC for phenology modeling at ENF sites. Vegetation phenology has received increasing attention in climate change research. Near-surface sensing using digital repeat photography has proven to be useful for ecosystem-scale monitoring of vegetation phenology. However, our understanding of the link between phenological metrics derived from digital repeat photography and the phenology of forest canopy photosynthesis is still incomplete, especially for evergreen plant species. Using 49 site-years of digital images from the PhenoCam Network from eight evergreen needleleaf forest (ENF) and six deciduous broadleaf forest (DBF) sites in North America, we explored the potential of the green chromatic (GCC) and red chromatic coordinates (RCC) in tracking forest canopy photosynthesis by comparing camera-derived start- and end-of-growing season (SOS and EOS, respectively) with corresponding estimates derived from eddy covariance-derived daily gross primary productivity (GPP). We found that for DBF sites, both GCC and RCC performed comparable in capturing SOS and EOS. However, similar to earlier studies, GCC had limited potential in capturing GPP-based SOS or EOS for ENF sites. In contrast, we found RCC was a better predictor of both GPP-based SOS and EOS for ENF sites. Environmental and ecological explanations were both provided that phenological transitions derived from RCC were highly correlated with spring and autumn meteorological conditions, as well as having overall higher correlations with phenology based on LAI, a critical variable for describing canopy development. Our results demonstrate that RCC is an underappreciated metric for tracking vegetation phenology, especially for ENF sites where GCC failed to provide reliable estimates for GPP-based SOS or EOS. Our results improve confidence in using digital repeat photography to characterize the phenology of canopy photosynthesis across forest types.