Publications

Liu, Y; Wu, CY; Tian, F; Wang, XY; Gamon, JA; Wong, CYS; Zhang, XY; Gonsamo, A; Jassal, RS (2022). Modeling plant phenology by MODIS derived photochemical reflectance index (PRI). AGRICULTURAL AND FOREST METEOROLOGY, 324, 109095.

Abstract
Vegetation phenology is a sensitive indicator of ecosystem responses to climate change, and thus the accurate estimation of vegetation phenology is critical to evaluate the impact of climate change on terrestrial ecosystems. Common structural vegetation indices (VIs) such as the Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), Near-infrared Reflectance of Vegetation (NIRv) and Plant Phenology Index (PPI), are the most widely used indicators of phenology, but they have limited potential in tracking autumn phenology, especially for evergreen species with low seasonal variability of canopy greenness. Given the important role of carotenoid pigments in regulating photosynthetic activity and plant phenology, we hypothesize that satellite-based indicators of leaf pigments derived from MODIS ocean bands could be useful for phenology modeling. Using 624 site-years of flux data at 84 FLUXNET sites and 9979 ground observations at 138 PEP725 sites, we first explored the potential of different forms of scaled photochemical reflectance index (sPRI(ref)) in monitoring photosynthetic activity, and found that band 10 and band 13 were more suitable for tracking gross primary productivity (GPP) than other reference bands. By comparing with canopy photosynthetic phenology, sPRI(10) and sPRI(13) showed improved representation of phenological transitions (the start and end of growing season, SOS and EOS, respectively) than structural VIs. In spring, all VIs exhibited comparable performances for estimating SOS at deciduous broadleaf forests (DBF) and grasslands (GRA) sites; however, sPRI(10) and sPRI(13) were better predictors of SOS than structural VIs at evergreen needleleaf forests (ENF) and mixed forests (MF) sites. In autumn, sPRI(10) and sPRI(13) showed improved predictive strength of EOS than structural VIs for ENF, MF and GRA sites. Further investigations using the ground observed phenological records also confirmed the improved performances of sPRI(10) and sPRI(13) for both SOS and EOS estimation. We also investigated the spatial patterns of sPRI(10)-derived SOS and EOS over the Northern Hemisphere with respect to different plant functional types. We showed that sPRI(10) reliably tracked plant phenology with 83.0% and 78.8% success in detecting SOS and EOS, respectively. Spatial patterns of SOS exhibited obvious latitudinal gradients, while EOS showed a strong regional heterogeneity. In addition, sPRI(10) predicted an overall earlier SOS (61.8%) and later EOS (51.2%) than the MODIS phenology product (VNP22Q2 v001) estimated from structural VI, suggesting the latter underestimated the greening potential of the Northern Hemisphere. Our results suggest that MODIS PRI could be useful to monitor vegetation phenology, and further reveal the importance of underappreciated carotenoid pigments in tracking plant seasonal changes, particularly in autumn months.

DOI:
10.1016/j.agrformet.2022.109095

ISSN:
1873-2240