Correlated change in normalized difference vegetation index and the seasonal trajectory of photosynthetic capacity in a conifer stand

Seasonal changes in canopy photosynthetic activity play an important role in carbon assimilation. However, few simulation models for estimating carbon balances have included them due to scarcity in quality data. This paper investigates some important aspects of the relationship between the seasonal trajectory of photosynthetic capacity and the time series of a common vegetation index (normalized difference vegetation index, NDVI), which was derived from on site micrometeorological measurements or smoothed and downscaled from satellite-borne NDVI sensors. A parameter indicating the seasonality of canopy physiological activity, P-E, was retrieved through fitting a half-hour step process model, PROXELNEE, to gross primary production (GPP) estimates by inversion for carboxylation and light utilization efficiencies. The relative maximum rate of carboxylation (V-rm), a parameter that indicates the seasonality of CO2 uptake potential under prevailing temperature, was then calculated from PE and daily average air temperature. Statistical analysis revealed that there were obvious exponential relationships between NDVI and the seasonal courses for both canopy physiological activities P-E and V-rm. Among them, the on-site broadband NDVI provided a robust and consistent relationship with canopy physiological activities (R(2=)50.84). The relationships between satellite-borne NDVI time series with instantaneous canopy physiological activities at the time of satellite passing were also checked. The results indicate that daily step NDVI time series (data downscaled from composite temporal resolution NDVI) better represent the daily average activity of the canopy. These findings may enable us to retrieve the seasonal course of canopy physiological activity from widely available NDVI data series and, thus, to include it into carbon assimilation models. However, both smoothing methods for satellite-borne NDVI time series may generate incorrect estimates and must be treated with care.