- English
Fluvial sediment fluxes are reshaping coastal water colour, geomorphology, and ecological processes, yet
schemes based on single-river loads or shoreline change may miss land-sea responses. We combine SSC records
from 19 river stations across nine Chinese basins (2004-2024) with MODIS reflectance to characterize clearing,
brightening, regional differentiation, and habitat reshaping. Clearing denotes a persistent reduction in suspended
sediment concentration, while brightening denotes the optical consequence of that reduction, namely enhanced
underwater light penetration. We classify optical water types (OWTs) using joint colour-brightness features and
train four machine-learning models per OWTs; LightGBM performs best in clearer regimes (max R 2
= 0.89), and CatBoost performs best in the most turbid regimes (R 2
= 0.83). Based on 21 years of retrieved SSC, we then
categorized long-term nearshore SSC trajectories into four directional change types: Further Clearing (FC),
Gradually Clearing (GC), Gradually Turbidifying (GT), and Further Turbidifying (FT), and identified heterogeneous
nearshore SSC trajectories: 5.0% and 41.2% of waters show FC and GC, 22.0% and 3.2% show GT and FT,
and the remaining 28.6% are relatively stable. The typical high-turbidity belt contracts from about 7.5 km to
about 6.0 km offshore, and mean SSC decreases from 58 to 48 g m 3 . We propose using muddy-water retreat
distance and extracting suspended sediment fronts to track onshore-offshore change. Brightening does not
necessarily imply greening, and SHAP attribution highlights reduced sediment input, offshore water intrusion,
and stable stratification as dominant drivers, with wind-wave resuspension emerging only beyond thresholds.
Our framework provides diagnostics for integrated land-sea management and nearshore restoration.
- English
