Goncalves, NB; Lopes, AP; Dalagnol, R; Wu, J; Pinho, DM; Nelson, BW (2020). Both near-surface and satellite remote sensing confirm drought legacy effect on tropical forest leaf phenology after 2015/2016 ENSO drought. REMOTE SENSING OF ENVIRONMENT, 237, 111489.

Amazon forest leaf phenology patterns have often been inferred from the Moderate Resolution Imaging Spectroradiometer (MODIS) Enhanced Vegetation Index (EVI). But reliable MODIS detection of seasonal and interannual leaf phenology patterns has also been questioned and is generally not validated with field observation. Here we compare inter-annual patterns of local-scale upper canopy leaf phenology and demography derived from tower-mounted phenocams at two upland forest sites in the Central Amazon, to corresponding satellite vegetation indices retrieved from MODIS-MAIAC (Multi-Angle Implementation of Atmospheric Correction). We focus on forest response to an unprecedented drought caused by the El Nino of 2015-16. At both sites, multi-year phenocam data showed post-drought shifts in leaf demography. These were consistent with MODIS-MAIAC anomalies in two vegetation indices. Specifically, a precocious leaf flush at both sites during the first two post-drought months, Feb-Mar 2016, caused (1) an anomalous decrease in flushing trees in Jun Jul of 2016 and (2) an increase of trees with early mature stage leaves (2-4 mo age) in Apr-May-Jun of 2016. At both sites, these two phenological anomalies showed up in MODIS-MAIAC as, respectively, (1) a strong negative anomaly in Gcc (Green chromatic coordinate), which prior work has shown to be sensitive to the abundance of leaves 0-1 mo old, and (2) a strong positive anomaly in EVI, which is sensitive to abundance of leaves 2-4 mo age. A shift to sub-optimal seasonal leaf age mix is expected to change the ecosystem-scale intrinsic photosynthetic capacity for similar to 18 month after the drought.