Work

Consistent nutrient mapping from Sentinel-2 and Sentinel-3 in nearshore waters: A case study in Xiamen Bay

Year: 2025

Type: article

Abstract: High-resolution monitoring of nutrient concentrations is essential for assessing water quality in coastal and bay regions. This study proposes a two-step framework to map dissolved inorganic nitrogen (DIN) and phosphorus (DIP) concentrations at high spatial resolution in Xiamen Bay (XMB). First, the nutrient retrieval models, termed AutoGluon-DIN/DIP, were trained using matchups between radiometric measurements from Sentinel-3 Ocean and Land Color Instrument (OLCI, 300 m resolution) and in situ nutrient observations. The high revisit frequency of OLCI enabled the compilation of a large and representative training dataset. Then, a cross-sensor transfer model, termed AutoGluon-transfer, was developed to convert Sentinel-2 Multi Spectral Instrument (MSI, resampled to 10 m) reflectance data into OLCI-equivalent bands, enabling the implementation of OLCI-trained models on MSI imagery. The key inputs of AutoGluon-DIN/DIP were the Rayleigh scattering-corrected top-of-atmosphere reflectance ( ρ rc ( λ )) at eight bands common to both MSI and OLCI. Validation against in situ data showed that AutoGluon-DIN/DIP outperformed other machine learning models, with a root mean squared difference of 0.11 mg L −1 for DIN and 0.012 mg L −1 for DIP ( N = 636). The retrieved DIN/DIP values also aligned well with independent buoy measurements in both magnitude and temporal variation (coefficient of determination R 2 ~ 0.6, N = 382). The AutoGluon-transfer effectively converts MSI-measured ρ rc ( λ ) to OLCI-equivalent bands (R 2 > 0.8), yielding nutrient maps that are consistent with those from OLCI in both magnitude and spatial pattern. Overall, the proposed framework offers a promising solution for high-resolution nutrient monitoring in coastal waters. • Employed AutoGluon for nutrient retrievals and cross-sensor data fusion. • Developed cross-sensor fusion model for Sentinel-2 MSI and Sentinel-3 OLCI. • Achieved consistent nutrient mapping from MSI and OLCI. • Introduced Rayleigh-corrected TOA reflectance for improved nutrient retrieval.

Source: Ecological Informatics

DOI: 10.1016/j.ecoinf.2025.103470