Regional estimation of cadmium input to soil via hydrological pathways: Coupling process models and machine learning

Area of Science:

• Environmental Science
• Hydrology
• Soil Science

Background:

• Existing models struggle with spatiotemporal resolution for heavy metal fluxes, especially during floods.
• Accurately linking pollutant sources to sinks in complex hydrological systems remains a challenge.

Purpose of the Study:

• To develop an integrative model for quantifying cadmium (Cd) input fluxes via irrigation and flood pathways at a regional scale.
• To overcome limitations in existing models regarding spatiotemporal resolution and flood event dynamics.

Main Methods:

• Integrated SWAT and advection-diffusion equations for high-resolution Cd transport simulation.
• MIKE FLOOD for estimating Cd input during flood events.
• XGBoost-based machine learning (ML) correction using empirical soil Cd data.

Main Results:

• Estimated annual Cd input fluxes: 0.91 mg m⁻² (irrigation) and 4.63–7.95 mg m⁻² (floods).
• Sediment contributes significantly to Cd transport (16.82% irrigation, 32.17% flooding).
• Average corrected Cd input flux is 2.71 mg m⁻²·a⁻¹, with 28.08% of farmland affected by combined inputs. ML correction achieved R² = 0.89.

Conclusions:

• The novel integrative model provides a robust and scalable framework for assessing regional soil cadmium fluxes.
• This approach offers valuable insights for effective soil cadmium risk management and pollution control.
• The model enhances understanding of hydrological dynamics in heavy metal transport and deposition.