A Scale-Adaptive Framework for Modeling Critical Zone Processes and River Water Quality in the East River Watershed

Critical Zone processes encompass interactions among rock, soil, water, air, and living organisms, essential for quantifying water and nutrient fluxes and predicting downstream river water quality. High-fidelity reactive transport models (RTMs) are important for understanding Critical Zone processes but are typically computationally expensive, which limits their applicability across large catchments. To address these challenges, we developed a scale-adaptive reactive transport simulation framework that balances process fidelity with computational efficiency. We developed the RiverFlotran Module, which employs fully dynamic 1D shallow-water equations for river hydrodynamics, and integrated it into PFLOTRAN, a subsurface reactive transport model. This integration enables us to simulate bidirectional exchanges at the land-water interface. Subsequently, we developed a machine learning-based exchange function, trained on the simulated data, and tailored for the East River. This function allows us to predict river water quality along the river continuum. This framework was applied to the East River Mountainous Watershed in Colorado, a study site of Berkeley Lab's Watershed Function Scientific Focus Area, to demonstrate its effectiveness in capturing intricate Critical Zone interactions and predicting downstream river water quality. Our study of the East River Floodplain's alluvial aquifer revealed that prevailing anoxic conditions generate pronounced redox gradients, resulting in the downstream export of dissolved iron and nitrogen near meander bends. These bends consistently serve as nitrogen hotspots, irrespective of water levels, driven by variations in river stage, bathymetry, and meander geometry, such as sinuosity. This modeling framework provides a foundation for quantifying river water quality at the catchment scale.