Global Patterns, Trends, and Drivers of Fluvial Suspended Sediment Fluxes from 1985 to 2024

Fluvial suspended sediment is fundamental to channel morphology, delta formation, water quality, biogeochemical cycles and ecosystems. The Earth is currently experiencing unprecedented human impacts and climate change, which are profoundly altering suspended sediment dynamics in river systems. However, fluvial suspended sediment flux (SSF) remains poorly constrained at the global scale, and its response to environmental change is still not well understood. Here, we develop a machine-learning model (XGBoost) to estimate fluvial SSF using in situ observations from 445 gauging stations worldwide, combined with spectral bands and river widths derived from Landsat imagery, river slope from SWOT River Database (SWORD), and bankfull discharge from the Global River Bankfull Discharge (GQBF) dataset (Liu et al., 2024). The trained model is applied to estimate global monthly SSF from 1985 to 2024 for river segments wider than 90 m, covering a total river length of 1.08 × 10⁶ km extracted from the Global River Topology (GRIT) dataset (Wortmann et al., 2024). Independent validation indicates that the model achieves a relative error of 0.12 compared with in situ SSF observations. Our results show that the global average annual SSF is 112.8 kg s⁻¹, with a total sediment delivery from land to ocean of 3705.1 Mt yr⁻¹. From 1985 to 2024, global average annual SSF exhibits a significant decreasing trend (-0.81 kg s⁻¹ yr⁻¹), with an abrupt shift around 2005 from a non-significant to a significant decline (-0.83 kg s⁻¹ yr⁻¹). Despite this global decrease, a larger proportion of river segments show increasing SSF (32%) than decreasing SSF (26%). This apparent contradiction arises because river segments with decreasing SSF typically have higher fluxes (SSF > 50 kg s⁻¹), whereas increasing trends are concentrated in rivers with lower SSF (SSF < 50 kg s⁻¹). Finally, we investigate the dominant drivers and their compounded effects on global fluvial SSF dynamics using a Bayesian network framework. This study provides new insights into global patterns, trends, and controls of fluvial suspended sediment fluxes under ongoing environmental change.

Liu, Y., Wortmann, M., & Slater, L. (2024). Global River BankFull Discharge (GQBF) (0.1) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.13855371

Wortmann, M., Slater, L., Hawker, L., Liu, Y., & Neal, J. (2024). Global River Topology (GRIT) vector datasets (0.6) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.11219313