Location: Watershed Physical Processes Research
2025 Annual Report
Objectives
1. Develop physically based multidimensional numerical models and technology for watershed erosion and sediment transport processes, water quality issues due to agro-pollution, and surface-groundwater interactions in agriculture landscapes.
1.1. Expand CCHE1D model capabilities.
1.2. Simulate rain-induced watershed soil erosion over tilled and non-tilled agricultural land and sediment transport.
1.3. Develop a coupled surface-subsurface water model.
1.4. Improve CCHE water quality and pollutant transport models.
2. Enhance the Decision Support System for the web-based Agricultural Integrated Management System (AIMS) by integrating watershed and channel network technology with geospatial and remote sensing data for effective watershed runoff, sediment, and water quality management.
2.1. Improve AIMS and AnnAGNPS integration.
2.2. Integrate CCHE1D channel network delineation model and multiple-scaled channel networks and flow model into AIMS.
2.3. Develop assessments of the AIMS integrated technology with case studies.
2.4. Explore cloud computing capability for AIMS.
Approach
In order to better assess and predict the direct impacts of water availability and soil erosion from agricultural fields, gullies and channels, existing technology must be developed and upgraded to assess and improve predictions for more realistic conditions. This includes technology that is needed for decision making in identifying, targeting and implementing conservation practices that protect the landscape. The first objective of the project plan focuses on the development and improvement of numerical models associated with watershed erosion and sediment transport processes, water quality issues due to agro-pollution, and surface-groundwater interactions in agriculture landscapes. This objective relates to the ARS National Program Action 211, Component 1 (Effective Water Management in Agriculture) and Component 2 (Erosion, Sedimentation, and Water Quality Protection). The second objective focuses on the development and improvement of a watershed management tools to assess runoff, sediment, and water quality conservation practice impacts. This objective is related to NP 211 Component 1 and Component 4 (Watershed Management to Improve Agroecosystem Services).
Progress Report
Under Subobjectives 1.1.1–1.1.4 the CCHE1D looped channel network model has been enhanced for two-dimensional (2D) flow simulations. A novel approach was developed to generate unique channel network systems that cover 2D domains while avoiding overlaps and interceptions by adhering to hydrologic and geometric principles. These networks follow the steepest slope direction, distinguishing them from traditional 2D meshes.
The model now incorporates:
The Junction-Point Water Stage Prediction and Correction (JPWSPC) method for solving looped channel networks;
Support for Cross Sections with Multiple Channels (CMC), which differ from traditional compound channels and better represent complex 2D field conditions.
After validation using a benchmark looped network case, the model was applied to an urban flooding simulation involving 14 intersecting streets, and two meandering river reaches (Mississippi River and East Fork River), including point bars and islands. Results showed the 1D model produced outputs comparable to those of a 2D model. A journal article has been drafted.
The CCHE1D_WQ module was improved to simulate water quality constituents in channel networks, incorporating effects from upland watersheds. The updated CCHE1D model has been integrated with the AnnAGNPS model, enabling joint simulation of flow, sediment, and nutrient transport across upland and channel network systems. The linkage connects AnnAGNPS outputs (runoff, sediment, nutrients) to CCHE1D input nodes. The model was validated with data from the Goodwin Creek Experimental Watershed and the Upper Pearl River Basin. A conference paper was presented at the 2025 ASCE/EWRI World Environmental & Water Resources Congress.
The water quality module has also been embedded into the CCHE1D-GUI, making it significantly more accessible for users. Users can now define water quality parameters, initial/boundary conditions, and launch simulations through a graphical interface.
Under Subobjectives 1.2.1–1.2.3 the CCHE2D sediment transport model was accelerated using OpenMP for multithreaded CPU execution—an alternative for when GPU resources are unavailable. Application of this version in high-resolution laboratory and field scenarios demonstrated a 2× speedup on a standard laptop. A technical report detailing this implementation is in preparation.
The model was further applied to simulate sediment transport under unsteady flow conditions, using laboratory flume experiments with variable flow hydrographs. Calibration and validation were successfully achieved. Digital flumes with varying boundary/initial conditions and hydrograph characteristics (e.g., duration, sediment size, and rising limb slopes) were used to investigate sediment transport behaviors. Findings revealed:
A counterclockwise hysteresis pattern in sediment transport vs. flow rate;
Coarser sediments reduced transport rates, finer particles increased them, but hysteresis direction remained unchanged;
Limited sediment supply altered the hysteresis pattern to a figure-eight shape;
For longer hydrographs, irregular patterns emerged due to bedform evolution, which introduced uncertainty in model accuracy during peak flows.
A journal paper based on this study was submitted to the Journal of Hydraulic Engineering.
Under Subobjectives 1.3.1–1.3.3 a 3D stable Radial Basis Function (RBF) model was developed for simulating groundwater (GW) flow induced by pumping/injection wells, including vadose zone effects. This meshfree model is computationally efficient and accurate. After validation, it was applied to a real-world case near a meandering river in Shellmound, Mississippi. This work has been published in Engineering Analysis with Boundary Elements.
A coupled surface water (SW) and groundwater (GW) model using the mimetic finite difference method was also developed. This method, capable of handling general polygonal meshes, was applied to simulate rainfall-induced infiltration in areas with complex topography (e.g., gullies). Validation and application were carried out in Jasper, Iowa, and presented at the 2024 Mississippi Water Resources Conference.
A 3D numerical model using the Hybrid High-Order (HHO) method was created to simulate GW pumping/injection. The method supports general polytope meshes, facilitating integration of high-resolution geophysical data. A novel well implementation approach avoided explicit smoothing operators, improving efficiency. The model was validated and applied to an agricultural case in Shellmound, Mississippi, and has been submitted to Advances in Water Resources.
A Lagrangian particle-based model for GW solute transport was developed using the Generalized Moving Least Squares (GMLS) method. This improved model demonstrated better performance than traditional MLS and SPH approaches, especially in handling anisotropic dispersion and advection-dominant flows (e.g., hyporheic zones). The manuscript is currently under internal review.
Under Subobjectives 1.4.1–1.4.5 the CCHE_WQ model was enhanced to simulate particulate and dissolved nutrients such as phosphate (PO4), organic phosphorus (OP), and organic nitrogen (ON). Processes including adsorption/desorption, bed release, and sediment exchange were incorporated. Model performance was validated with experimental data, and findings were published in the Journal of Environmental Engineering.
The CCHE_MIX_WQ model was linked with AnnAGNPS to evaluate lake water quality responses to agricultural practices. This system was applied to Beasley Lake and published in the Journal of Environmental Modelling & Software.
A pollutant transport model was also developed and will be validated using data from a flooding event.
Under Subobjectives 2.1.1–2.1.3 the AIMS team focused on strengthening and restructuring backend architecture using Proxmox, supporting high availability, virtualization, and containerization. A new architecture is being coordinated with the University of Mississippi IT department, enabling automated deployments, backups, and resource allocation using Ansible.
The web interface was modernized with Django, Tailwind CSS, and emerging JavaScript frameworks. A new dashboard includes project-level and scenario-level management with geospatial mapping.
Tile servers were deployed for faster and safer geospatial data display. This work was presented at AGU 2024.
A national-scale dataset of simulated water yield and sediment yield (2000–2022) using NLDAS-2 climate data is under development. Inputs are stored in PostgreSQL, with ancillary spatial data stored separately. A fallback to DAYMET is in place for missing data.
Efforts are also underway to use Crop Sequence Boundary datasets for refined management input and NHDPlus datasets for improved topological accuracy.
Under Subobjectives 2.2.1–2.3.1 a Goodwin Creek Watershed case study tested the integration of AnnAGNPS and CCHE1D, linking 2,271 AnnAGNPS cells to 912 reaches. Three routing approaches were evaluated, with the triangular hydrograph method yielding the best results. This was presented at the 2024 EWRI Conference.
A concurrent project using HydroSWOT data and ANN models is being developed to estimate river cross-sectional parameters for CCHE1D input.
A multi-watershed study was conducted to assess the effect of ET forcing on runoff predictions using MODIS, NOAH, and OpenET datasets, as well as PRISM precipitation. Results showed that ET forcing generally improves model performance, with location-dependent variability.
In the Upper Pearl River Watershed, a distributed simulation method was developed, dividing 61,000 AnnAGNPS cells into 3,048 groups for parallel execution. This reduced total simulation time from 781 to 6 hours. Model validation showed NNSE scores of 0.66–0.75, and findings were presented at the 2024 Mississippi Water Resources Conference.
Accomplishments
1. Model parallelization of CCHE2D sediment transport model. Accurate sediment transport modeling is often slowed down on standard computers when high-performance GPUs (Graphics Processing Units) are unavailable. To address this, an advanced two-dimensional model was developed to be accelerated using an application programming interface that supports multi-platform shared-memory multiprocessing (OpenMP) for parallel computing on multi-threaded CPUs (Central Processing Units). This adaptation improved simulation execution time by a factor of two on regular laptops, enabling faster modeling tests for field and lab applications. Researchers and engineers at Oxford, Mississippi working to find solutions to reducing sediment transport from agricultural watershed streams and watershed benefit from this more efficient modeling capability, particularly where GPU resources are limited.
2. Enhancement of CCHE1D looped channel network model. Simulating complex river networks where flows loop completely around landscapes within rivers can be difficult using traditional one-dimensional (1D) modeling tools. A suite of enhancements was developed for the one-dimensional channel simulation tool, CCHE1D, to overcome this, including a unique and creative algorithm for generating 1D networks in which individual segments do not share or duplicate the same spatial locations, a solver for looped networks, and a method for modeling compound cross-sections. These enhancements to the 1D model allow for better simulations of real-world river systems. Water resource planners and hydrologists can now better understand and manage complex water networks impacting agricultural watershed systems.
3. Application of a numerical model for sediment transport in laboratory studies. Sediment transport must be considered in the management of rivers and streams; however, the expense and difficulty of measurements mean that computer models are used to predict sediment transport rates based on channel and flow conditions. Most rivers and streams have conditions that change with time due to factors such as rainfall and runoff events and hydropower plant operations. Predictive methods that were developed for river conditions that do not change over time will result in greater errors when applied to streams with flow rates that change with time. Hence, there is a need for computer models that can more accurately predict sediment transport during changing flow conditions. In this study, we used data from experiments where sediment transport and the shape of the sand bed were carefully measured while the flow rate was changed over time to develop a method for predicting the transport using a computer model. The method was developed using just one of the experiments at a USDA’s National Sedimentation Laboratory flume, and it was shown to perform well by predicting transport rates for other experiments. This model can eventually be used model transport rates in rivers and streams, including prediction of transport rates during unusually large flows, such as those caused by intense rainfall events. The results will be used by river managers, model developers, and other researchers studying river engineering and sediment transport within agricultural water systems.
4. Development of a groundwater model using a hybrid high-order method for flow induced by pumping and injection wells. Efficient use of water is needed to sustain irrigated croplands. Farmers and water managers often pump water from wells to irrigate agricultural fields to maximize cropland yields. Sustainable yields require that groundwater aquifers be replenished, with one option to return any available water back into the aquifers via injection wells for later use. Managing how to withdraw and replenish water in aquifers requires advanced modeling technology to accurately show how using these wells affect groundwater supplies. A simple and unique modeling approach was adopted to describe the complex system of groundwater flow that includes determining the impact of extracting and injecting water through wells on groundwater movement. This was accomplished by incorporating advanced geophysical techniques that measure complex underground systems into efficient numerical methods to describe extraction, injection, and water movement processes. The new model was tested with real-world data from a Mississippi Delta farming area where water was extracted from a plentiful groundwater source and transported nearly two miles for injection via wells into a depleted groundwater aquifer for eventual use as cropland irrigation water. The study showed that having detailed information from local farmers is key to making these models accurate. This work supports smarter farming practices and better water management, which benefits both agriculture and society.
5. Development of a coupled surface and groundwater flow model. Understanding how rainfall interacts with groundwater is critical in areas with varied topography. Conventional numerical models typically have strict requirements for the computational grids used in spatial discretization, making them less efficient for modeling problems with complex topographies. However, surface and groundwater interactions related to agriculture often occur in areas with intricate gullies and preferential flow paths. A coupled numerical surface and groundwater flow model was therefore developed using a novel numerical technique that preserves important physical rules from the real world and works effectively on complex, irregular grid shapes. With this newly developed model, dynamic soil moisture data with high spatial and temporal resolution can be generated through numerical simulations, making the coupled modeling system a powerful tool within a decision support framework for evaluating farming strategies and guiding soil conservation practices.
6. Integration of CCHE1D and AnnAGNPS for watershed modeling. Accurate modeling of flow, sediment and nutrient transport across watersheds is complex due to the need to link upland and stream processes. Simulating hydrologic, sediment and nutrient transport from upland areas into stream networks requires seamless integration between different watershed and hydrodynamic models. The enhanced on-dimensional channel model, CCHE1D, was integrated with the USDA watershed conservation planning tool, AnnAGNPS, to simulate flow, sediment, and nutrient transport across upland watersheds and river reaches. To enable this, a method was developed for converting AnnAGNPS reach and cell outputs into boundary conditions compatible with CCHE1D compute nodes. Three boundary condition approaches were tested: (a) using AnnAGNPS cell outputs directly as boundary conditions without stream routing, (b) applying daily averaged discharges with routing to account for stream travel time, and (c) using triangular hydrographs derived from AnnAGNPS outputs to represent more detailed flow dynamics. The system was validated using field data from the Goodwin Creek Experimental Watershed and the Upper Pearl River Basin. This integrated modeling approach benefits agricultural watershed planners by providing scalable tools for simulating detailed watershed-scale hydrology and water quality.
7. Enhancement of CCHE_WQ water quality model for nutrient simulations. Water quality models often lack the ability to simulate both particulate and dissolved nutrient dynamics. The numerical water quality model of CCHE_WQ was enhanced to include key processes like adsorption, sediment exchange, and bed release. Initial testing showed promising alignment with experimental data, with field-scale validation ongoing. These improvements provide a more complete tool for studying nutrient pollution and benefit agricultural watershed and lake management efforts.
8. Linking watershed and hydrodynamic water quality models for lake water quality assessment. Understanding how agricultural practices impact lake water quality requires integrated modeling systems that can simulate both water flow and pollutant movement. The CCHE_MIX_WQ hydrodynamic water quality model simulates lake and river water quality by tracking pollutants like nutrients and sediments in water bodies, while the AnnAGNPS watershed model estimates how agricultural runoff—including fertilizers, pesticides, and soil erosion—moves across the land into streams and lakes. A link between these two models was created to combine their strengths and better evaluate how farming activities affect water quality downstream. This combined model was applied to the rural agricultural Beasley Lake Watershed in the Mississippi Delta to assess the effects of different cropping and tillage practices on lake pollution levels. This advanced decision-support technology provides agricultural stakeholders with a powerful tool to test and apply conservation management practices aimed at improving lake health and protecting water resources.
9. Integrating default watershed modeling simulations into the web-based decision support system of agricultural integrated management system (AIMS). AIMS is a web-based decision support tool designed to help users run watershed and channel models across the United States using automated input data preparation and a seamless map server. Integration of the Annualized Agricultural Non-Point Source (AnnAGNPS) model from USDA-ARS for watershed simulation, and the CCHE1D channel network model from the University of Mississippi for water routing, is currently underway within AIMS. In addition, long-term default hydrologic simulations for all U.S. catchments are being incorporated into the system. As part of this effort, 20-year pre-runs of erosion and water yield data have been initiated for every catchment in the continental U.S. (CONUS) using AnnAGNPS. These pre-computed results are intended to give users quick access to key data without the need to set up or run their own simulations. For the agricultural community, having access to this pre-calculated erosion and water yield information allows farmers and land managers to better understand water availability and soil loss risks. This supports more informed decision-making around conservation practices and promotes more sustainable and resilient farming systems.
10. Impact of evapotranspiration datasets on runoff prediction. Runoff predictions are highly sensitive to the evapotranspiration (ET) data used. Evaluations of spatially and temporally continuous ET datasets from NASA’s MODIS (Moderate Resolution Imaging Spectroradiometer) satellite sensor and the NLDAS (North American Land Data Assimilation System) land surface models were conducted using the USDA watershed model AnnAGNPS across various U.S. watersheds. Results showed that incorporating ET data generally improved runoff predictions, with differences depending on land use and watershed size. This helps modelers choose more suitable datasets, improving the accuracy of runoff forecasting. For agriculture, more accurate runoff predictions are essential for managing irrigation, reducing soil erosion, and planning conservation practices. By better understanding how much water remains available in the soil or runs off fields, farmers and land managers can make smarter decisions to protect crops and conserve water resources.
11. Accelerating large-scale watershed modeling. Modeling large watersheds is often limited by memory and computation time. To address this, a distributed simulation method was developed using the USDA watershed model AnnAGNPS with parallelization, enabling faster and more scalable runs. This approach was tested on several large agricultural watersheds in Mississippi, including the 7,124 km² Upper Pearl River Basin. The accuracy of model outputs for water and sediment yield was maintained while runtime was significantly reduced. This approach allows for comprehensive modeling of large watersheds, benefiting researchers and agricultural land managers by providing timely, accurate data to support conservation planning, reduce soil erosion, improve water resource management, and enhance the long-term sustainability of farming operations.
Review Publications
Chao, X., Zhang, Z., Zhang, H. 2025. Advancing surface water quality modeling for TMDL implementation: enhancing sediment processes, atmospheric reaeration, and bed layer interactions. Journal of Environmental Engineering. 151(7). https://doi.org/10.1061/JOEEDU.EEENG-7888.
Fang, J., Al-Hamdan, M.Z., O'Reilly, A.M., Ozeren, Y., 2024. A stable localized weak strong form radial basis function method for modelling variably saturated groundwater flow induced by pumping and injection. Engineering Analysis with Boundary Elements 168: 105922. https://doi.org/10.1016/j.enganabound.2024.105922
Zhang, Y., Al-Hamdan, M., Chao, X. 2024. Parallel implicit solvers for 2D numerical models on structured meshes. Mathematics. 12(14):2184. https://doi.org/10.3390/math12142184.