Novel integration of multi-scale remote sensing improves characterization of tropical phenology from individual tree-crowns to landscapes

Abstract

Increasing evidence from tower-mounted phenocams shows unusual leaf phenology patterns in tropical evergreen forests. The landscape is evergreen, but strong and highly seasonal leaf dynamics occur at the individual tree-crown level. This causes strong seasonal variation in leaf quality (i.e., photosynthetic capacity and optical properties) with the changing leaf age mix, which helps explain seasonality of tropical forest photosynthesis and of satellite-detected greenness. Accurate characterization and understanding of tropical leaf phenology at different spatial and temporal scales remains a central issue in tropical ecology and Earth system sciences. Phenocam observations may be the most accurate way to quantify tropical leaf phenology from the individual crown scale to the landscape scale. However, such observations are yet limited to only a few forest sites and time spans of a few years. Satellite observations might fill this gap. But the commonly-used MODIS satellite data with 250+ meter resolution are too coarse to detect crown-scale leaf phenology dynamics in hyper-diverse tropical forests. To resolve this observational challenge, we here propose a novel integration of satellite images of high spatial resolution (i.e., the 3m PlanetLabs Planetscope data) with MODIS observations. We also compare the results of such integration with local phenocam observations in the tropics. Our results demonstrate that the novel integration of Planetscope and MODIS data can not only accurately assess the landscape leaf phenology pattern, which is consistent with the literature, but can also provide greater details of leaf phenology dynamics at the individual crown scale. Further, by using spectral unmixing, we inferred seasonality of crown-level deciduousness from satellite-reflectance measurements. Such satellite-derived deciduousness seasonality also well agrees with local phenocam observations. Our novel integration of multi-scale satellite observations can significantly advance knowledge of leaf phenology in the tropics, ranging from improved characterization of spatial details to improved biophysical interpretation of leaf dynamics.