Remote sensing inversion of water quality parameters (TSM, Chl-a, and CDOM) in subtidal seaweed beds and surrounding waters

Abstract

Due to environmental factors such as water transparency, subtidal seaweed beds are often challenging to observe directly via satellite. However, the presence of seaweed beds can lead to variations in the concentrations of total suspended matter (TSM), chlorophyll-a (Chl-a), and chromophoric dissolved organic matter (CDOM) in the surrounding waters. This study focuses on the seaweed beds around Gouqi Island, Zhejiang, integrating several months of in-situ water quality sampling data with PlanetScope satellite imagery to develop inversion models for water quality parameters using Random Forest (RF), Gradient Boosting Decision Tree (GBDT), and Support Vector Regression (SVR) algorithms. By analyzing the differences in water quality parameters between areas with seaweed beds and those without, we explored the underlying causes of these variations and proposed an indirect method for estimating the distribution range of underwater seaweed. This research not only provides a new perspective and technical approach for marine resource management but also contributes significant foundational data and scientific evidence for the conservation of coastal zone ecosystems.

Keywords

• Seaweed bed
• Water quality inversion
• Machine learning regression
• SHAP algorithm
• Indirect estimation method