Tracking photosynthetic phenology using spectral indices at the leaf and canopy scales in temperate evergreen and deciduous trees

Abstract

Vegetation photosynthetic phenology is an important indicator to characterize the biological responses of terrestrial carbon cycle to climate change. Remote sensing-derived spectral indices have been used to estimate photosynthetic phenology of terrestrial ecosystems, yet there are large uncertainties due to differences in leaf structure and canopy function between evergreen and deciduous trees. In this study, we used in-situ measured meteorology, the maximum rate of leaf-level CO2 assimilation (Amax), chlorophyll fluorescence and canopy color changes in evergreen and deciduous species to evaluate and compare the abilities of spectral indices for photosynthetic phenology modeling. Seasonal variation in Amax is used as a proxy for photosynthetic phenology. Several indices were used, including the normalized difference vegetation index (NDVI), near-infrared reflectance vegetation index (NIRv), photochemical reflectance index (PRI), chlorophyll/carotenoid index (CCI), green-red vegetation index (GRVI), and green chromaticity coordinates (Gcc). We found that NIRv and PRI estimated phenological periods for Cedrus deodara and three deciduous trees all differed from Amax by less than 5 days, and all were linearly related to Amax (R2 > 0.6). CCI had less potential for Amax modeling, while NDVI had the largest error in predicting Amax. Gcc was a more accurate proxy for photosynthetic phenology than GRVI at the canopy scale. In addition, we found that the variability of different spectral indices in tracking photosynthetic phenology of evergreen and deciduous species was related to non-photochemical quenching processes. Especially for the fraction of absorbed light quenched by dynamic NPQ (ɸNPQ) and the fraction of absorbed light quenched by sustained NPQ (ɸf,D), they contribute more than 40 % to the variations of the NIRv and PRI spectral signals. Therefore, NIRv and PRI provide powerful tools for monitoring photosynthetic phenology, and future exploration of NIRv and PRI at larger spatial scales will favor the global plant phenology and carbon uptake modeling.