Evaluation of suspended particulate matter retrieval models for subtropical estuarine waters monitoring

Monitoring suspended particulate matter (SPM) concentration is essential for environmental protection and the management of aquatic ecosystem quality. Ocean colour remote sensing enables the monitoring of extensive coastal and marine areas with high spatial and temporal resolution. The challenge lies in selecting SPM retrieval models that represent local optical complexities and integrating them across multiple sensors. This study conducted a bio-optical characterization and evaluated 76 SPM retrieval models for multiple sensors, supporting monitoring in a subtropical estuary of environmental and socioeconomic importance in southern Brazil. Four distinct optical water types were identified, highlighting the optical complexity of the study area. The best-performing models for each sensor (with RMSE and absolute Bias ranging from 4.36 to 6.00 g m−3 and 0.01 to 0.95 g m−3, respectively) were applied to simultaneous images, and the derived concentrations were compared. The observed differences in concentrations retrieved by each sensor reflected the models’ characteristics and the distinct spatial, spectral, and radiometric resolutions of the sensors. These differences were resolved through data harmonization after model inversion. As a result, the time series becomes consistent, and integrated images from these constellations can be used for retrospective SPM monitoring in the study area, allowing OLI and MSI to replace the MODIS time series, which already show signs of degradation. The methodology applied in this study can serve as a guideline for implementing satellite-based SPM monitoring in other highly complex aquatic environments.