ABSTRACT

Accurate satellite–in-situ collocation remains a key source of uncertainty in quantitative retrievals of coastal water turbidity, particularly in shallow and optically complex environments. This study presents a systematic assessment of how spatial, temporal, and depth-related collocation choices influence the performance of commonly used Sentinel–2 reflectance-based indices for turbidity monitoring. Three indices—the Normalized Difference Turbidity Index (NDTI), the Normalized Difference Water Index (NDWI), and the Normalized Difference Vegetation Index (NDVI)—were evaluated against an extensive in-situ dataset of CTD turbidity measurements aggregated into three near-surface depth layers (0–1 m, 1–2 m, and 2–3 m). Satellite–in-situ matching was conducted using spatial buffers ranging from 120 to 400 m, two processing resolutions (20 and 40 m), and a temporal matching window of up to ±48 h. Across all collocation scenarios, NDWI exhibited the strongest and most stable relationship with turbidity (FTU), with regression slopes and explained variance remaining consistent across depth layers, buffer sizes, and spatial resolutions. NDVI showed a weaker but coherent negative response, while NDTI demonstrated negligible predictive skill within the observed turbidity range. The limited sensitivity of retrieval performance to collocation parameter choices indicates that, for moderate turbidity conditions in shallow coastal waters, robust satellite-based estimates can be achieved without highly restrictive matching criteria. The proposed workflow provides a transferable and reproducible framework for evaluating satellite–in-situ collocation uncertainty and supports routine and operational coastal water-quality monitoring using Sentinel–2 data.