Evaluating Sentinel-2 and PlanetScope Imagery for Detecting Ice Phenology in Alaskan Lakes and Ponds
Abstract. Lake ice serves as both an indicator and driver of climate change, yet past studies have often overlooked ice phenology in small lakes and ponds due to the spatiotemporal limitations of existing satellite missions. Here we produce annual ice cover time series for 15,885 Alaskan lakes using high-resolution Sentinel-2 and PlanetScope imagery for the years 2019–2025. We assess the relative performance of both missions, then use our time series data to analyze the controls on lake ice timing across Alaska. We find that Sentinel-2 outperforms PlanetScope for ice phenology monitoring, especially at high latitudes, due to its superior revisit timing and improved classification performance. Planet’s underperformance stemmed from its lower spectral resolution and overarching issues with its cloud mask. Our analysis of ice-on and ice-off timing demonstrates that large lakes maintain ice cover for, on average, 7 days longer than small ponds. This difference is primarily driven by delayed ice-off timing in large lakes, as ice-on took place synchronously across lake sizes. Climatologically, we confirm that the seasonal 0 °C isotherm date is highly correlated with both ice-off and ice-on timing. We also find that lake ice cover responds more slowly to earlier warming in spring (and, conversely, faster to later cooling in fall), a moderate negative feedback which could dampen the number of lost ice cover days due to climate change. Overall, our findings reveal the value of high-resolution remote sensing data for studying lake ice dynamics in small lakes and emphasize Sentinel-2’s particular utility for high-latitude observation.