Evaluating fine-scale phenology from PlanetScope satellites with ground observations across temperate forests in eastern North America

Abstract

In temperate forests, leaf phenology is a sensitive indicator of climate change and a major regulator of seasonal carbon and water cycling. Many studies have documented large intra-site leaf phenology variability across individual trees but conventional approaches for monitoring individual tree-scale leaf phenology are often limited to a small spatial extent and sample size. Recent availability of PlanetScope satellite data with a 3 m spatial resolution, near-daily revisiting frequency, and global coverage provides opportunities to overcome this limitation. It also has the advantage of providing spatially explicit information across large spatial coverages compared with ground methods. However, comprehensive assessments of PlanetScope's capacity and scalability for individual tree-scale leaf phenology monitoring remain lacking. To address this knowledge gap, we propose an approach that integrates 0.1 m resolution airborne imagery and ground phenology records of individual trees with PlanetScope image time series, testing it at six NEON forest sites in eastern North America. We first extracted key phenological metrics at the individual tree scale from PlanetScope satellites and then evaluated the metrics with corresponding phenological metrics derived from ground observations over 2018 and 2019. Our results show that PlanetScope-derived fine-scale land surface phenology is able to 1) characterize significant leaf phenology variability at the individual tree scale across all forest sites and years, with r ranging from 0.21 to 0.42 when comparing PlanetScope-derived individual tree-scale phenological metrics with their ground correspondences. The accuracy is improved at the species level (r = 0.57–0.82) when more PlanetScope pixels are included; and 2) capture relatively more variations in fall phenology but also with larger uncertainties (e.g., r = 0.82 and RMSE = 2.14; species level) relative to spring phenology (r = 0.76 and RMSE = 0.72). Collectively, this study presents a comprehensive evaluation of PlanetScope's capacity for individual tree/species-scale leaf phenology monitoring and highlights the potential of PlanetScope to provide rich fine-scale phenology information to significantly advance the field of plant phenology research.