Tracking Intra-annual plasticulture covering cycles using dense Sentinel-2 time-series

• First attempt to characterize plasticulture covering cycles using dense time series. • Enables seasonal plastic cover extraction amid varied application timing and spectra. • HSL transformation enhances plastic cover features, improving detection robustness. Plasticulture, a widespread practice using plastic covers (PCs) to enhance crop yields, has raised environmental concerns. Existing studies capture static PC distributions, leaving temporal dynamics uncharacterized and often missing seasonal PCs applied outside the mapping window. To address this gap, this study proposes Plastic Cover Tracker (PCT), a rule-based approach that utilizes dense Sentinel-2 time series to track the intra-annual covering cycle of PCs. PCT extracts key temporal metrics, including the start date, end date, and duration of PC coverage, and classifies each pixel as permanent, seasonal or non-PC. Specifically, PCT first applies a Hue–Saturation–Lightness (HSL) transformation to improve spectral separability between PCs and the background. A single-date PC classification is then performed using the low-saturation and high-lightness of PCs, followed by temporal filtering to remove noise and ensure temporal continuity. An NDVI-based confirmation is subsequently applied to retain only agricultural PC applications. Results demonstrate the effectiveness of PCT: the HSL transformation significantly improved feature separability between PCs and non-PCs, with the Jeffries-Matusita (JM) distance increasing from ≤ 0.50 using a single RGB band to 1.50 using the saturation-lightness combination. For temporal metrics, PCT achieved accuracies of ± 1.83 days for start dates, ±1.74 days for end dates, and ± 3.27 days for durations using Sentinel-2 imagery. For pixel categorization into permanent, seasonal, and non-PC types, PCT achieved F1 scores of 0.79, 0.93, and 0.98, respectively. Comparison with existing PC datasets showed high consistency in permanent PC identification, supporting the reliability of PCT. Moreover, PCT successfully detected seasonal PCs that were overlooked in these datasets. In conclusion, PCT fills a critical methodological gap in tracking PC temporal dynamics and detecting seasonal PCs, with strong potential to support environmental risk assessments and promote sustainable plasticulture management.