Divergence in Autumn Phenology Extracted From Different Satellite Proxies Reveals the Timetable of Leaf Senescence Over Deciduous Forests

Remote sensing detection of autumn phenology is challenging and highly uncertain, as exemplified by the observed divergence in autumn phenology extracted from different proxies. Here, we compared the autumn phenology derived from Solar-Induced chlorophyll Fluorescence (SIF), Chlorophyll/Carotenoid Index (CCI), Enhanced Vegetation Index (EVI), and Normalized Difference Vegetation Index (NDVI) over deciduous forest sites. We observed a clear temporal sequence in the derived autumn phenology from various proxies: SIF < CCI < EVI < NDVI. Comparison with field measurements supported that SIF, EVI, and NDVI can successfully capture the attenuation of photosynthetic activity, leaf coloration, and leaf fall, respectively. The sequence among the autumn phenology derived from those proxies was also consistent with their responses to climate cues, where SIF had the highest partial correlation coefficient to solar radiation in autumn, followed by CCI, EVI, and NDVI, while NDVI was more correlated with temperature, followed by EVI, CCI, and SIF.