Groundwater Modelling Advice Draft

Page Under Construction

Common Questions Answered (FAQ)

What are the benefits of integrating groundwater modelling into flood simulations?

The addition of a groundwater model accounts for the attenuation of the rainfall-runoff response and the discharge of soil water to creeks. This helps in long-term simulations by generating base flow to the surface runoff, and helps predict the infiltration capacity for subsequent rainfall events by tracking the long term change in soil moisture. For examples of real world applications, please see:

• Australian Water School Webinar on TUFLOW Groundwater Flow Modelling and its Application
• 2024 Enhancing Catchment Runoff Simulations Using Soil Moisture Dependent Hydraulic Conductivity, Gao et al, HWRS. link
• 2023 Continuous Direct Rainfall Hydraulic Modelling with Coupled Surface Ground Water Interaction, Gao et al, HWRS. link

How can a gravel trench be represented if it does not allow infiltration but connects two attenuation basins and intercepts runoff?

If the gravel trench does not allow infiltration but acts as a conveyance feature, it can be represented as a 1D channel or pipe to connect the two attenuation basins.

The groundwater functionality can be used by setting the surface infiltration rate as zero to model the lateral movement through the trench only. 2D cells with non-zero infiltration rate are still needed to infiltrate the surface water to the soil layer. However, soil water can discharge to the surface if the soil layer becomes full at the zero infiltration cells (TUFLOW does not put a 'lid' on soil surface). This approach is not recommended if the gravel trench is expected to become full.

How should the Initial Loss/Continuing Loss (ILCL) infiltration method be applied, and are there standard values for different land cover types?

To use ILCL infiltration, the Soil ID must be defined in the model, either globally or spatially using a 2d_soil layer, which is read into the tgc file with the Read GIS Soil command. The 2d_grd check file can be used to confirm that the correct Soil IDs have been applied across the model.

TUFLOW does not provide standard IL/CL values, but values may be estimated from databases such as the CSIRO Soil Atlas. ARR Book 9, Section 6.4.2 offers general guidance for rainfall loss values, but this is separate from ILCL soil infiltration.

Can water exfiltrate from subsurface layers other than the top layer?

Water can move both horizontally within a soil layer and vertically between layers. Downward flow is controlled by convective hydraulic conductivity (CO), while upward flow occurs through surcharging.

How can permeable pavements be modelled?

Permeable pavements can be represented using soil layers. The simplest approach is to define infiltration using a 2d_soil layer, assigning a Soil ID with infiltration properties matching permeable pavements. This method removes infiltrated water from the model. If groundwater movement is important, multiple soil layers can be used to model horizontal flow and subsurface drainage.

How should peat soils be represented in a direct rainfall model?

If observed flow data is available, calibrating the model to these measurements would be the best approach. If not, using ReFH2 as a comparison for flow estimates is recommended. Since peat is often saturated, infiltration rates may be low, but lateral water movement could still occur. In this case, using the interflow functionality in TUFLOW may help better represent water movement within the catchment.

How can a French drain (filter drain) be represented?

There is no direct method for modelling a French drain in TUFLOW, but there are a few possible approaches.

• One option is to drain water directly from surface cells using 1D pit.
• In TUFLOW 2025.2.0 version and later, 2D groundwater layers can also be connected to the 1D domain using SX BC.

In the 2 methods above, it's required to use Darcy’s law to estimate discharge rates and create a depth vs discharge curves.

• Alternatively, the model cell size can be reduced to the trench width to conduct localised modelling of the subsurface flow through the drain. The modelling result can be used to estimate infiltration rates to connect 1D and 2D (surface or subsurface) in a model with larger cell size and larger extent.

How do soil parameters like thickness and hydraulic conductivity impact groundwater modelling?

Soil thickness and hydraulic conductivity control how water moves through the ground in TUFLOW’s groundwater modelling.

In general, a thicker soil layer increases soil capacity and has a greater ability to attenuate surface runoff during floods. A thicker soil layer also contributes to the continuous release of baseflow when it isn’t raining.

Hydraulic conductivity controls how easily water flows due to the gravity drain. During a flood event, vertical hydraulic conductivity determines the rate of infiltration from soil surface or from the soil layer above. Horizontal hydraulic conductivity affects the lateral movement of groundwater towards the bottom of the hillslope. A higher horizontal hydraulic conductivity increases the groundwater discharge to creek and produces more baseflow, which also frees up the soil capacity faster for the next rainfall event.

These factors are investigated in detail by this HWRS paper Gao et al. 2023

How is total volume in and infiltration loss reported, and how can they be separated?

TUFLOW reports total volume in using multiple components, including rainfall, inflows, and other sources. The infiltration losses are reported in the HPC mass balance CSV file (MB_HPC.csv). However, it is subtracted from the "S/RF Vol In" value if the rainfall rate is larger than the infiltration rate, and is reported as "S/RF Vol Out" value if the infiltration rate is higher than rainfall rate. In addition, this value is combined with other outflow types, which can sometimes make it unclear how much water has actually entered the system versus how much has infiltrated.

To separate infiltration losses:

• Remove negative rainfall or negative SA boundaries if possible.
• The rainfall volume can be calculated manually based on the rainfall rate and the rainfall area, or by running the model without soil infiltration to obtain the "S/RF Vol In" value in the absence of any soil infiltration.
• Run the model with soil infiltration. The difference in "S/RF Vol In" value is the volume of water infiltrated by soil layer.
• If "S/RF Vol Out" value are reported, the total of "S/RF Vol Out" and the "S/RF Vol In" difference is the volume of water infiltrated by soil layer.

Adding 2D PO lines across boundaries or 2D PO GWVol polygon can also help track how much water leaves the system via different pathways. TUFLOW does not currently provide a simple breakdown of water volumes, but understanding how mass balance values are calculated can help extract the needed information.

Why is water not leaving the system post-peak in a groundwater model?

If water remains in the system after the peak flow, review the following parameters:

Groundwater related:

• Check groundwater depth output (GWd) to see if the soil layer became saturated. Increasing soil layer thickness can increase the storage and sustain infiltration loss.
• Horizontal hydraulic conductivity: Low values may restrict lateral groundwater movement, limiting drainage, and keeping the soil layer full.
• However, if the horizontal hydraulic conductivity is high, the soil layer may generate continuous baseflow after the rainfall event, which can sustain higher water levels in creeks.

Surface water related:

• Downstream boundary conditions: An HT boundary may hinder efficient water exit. Consider testing with a QH boundary.

Running sensitivity tests by adjusting horizontal conductivity, initial moisture, and soil thickness can help identify key factors affecting outflow rates.

Why does groundwater exfiltrate immediately after the simulation starts?

If the initial soil moisture is saturated or near saturation, groundwater can immediately discharge to surface at low lying areas, causing rapid surface expression even before rainfall. This type of model setup is sometimes applied if the initial groundwater condition is unknown. It is suggested to try the following to reduce the impact of the initial condition:

• If initial groundwater condition is known (e.g. groundwater level monitoring data, soil moisture focusing data), set the initial condition using Set/Read GRID/Read GIS IGW Depth/Elevation commands.
• If initial groundwater condition is unknown, conduct a warm-up run to drain groundwater to a certain level before applying rainfall.
• Conduct sensitivity tests by adjusting initial groundwater depth/level and horizontal hydraulic conductivity.
• Find observation data (e.g. surface water flux, water level) to calibrate the initial groundwater condition and the model parameters.

How can interflow behaviour be better aligned with observed conditions?

• Ensuring groundwater inputs and outputs are balanced is key.
• Comparing groundwater accumulation areas with surface water flooding can help verify results.
• Using the groundwater XDMF output can assist in visualising flow behaviour and refining parameter selection.

Fine-tuning soil properties, hydraulic conductivity, and boundary conditions will improve interflow simulation accuracy.

Why does the groundwater lateral flux calculation include porosity, and what changes are planned?

Benchmarking tests have identified that the current sub surface flow equation in TUFLOW underestimates steady state discharge rates due to the inclusion of porosity. While the steady state water level gradient is correct, transient state simulations show discrepancies.

To address this, Groundwater Horizontal Flux Include Porosity command can be used to turn the porosity term on or off in the equation.

• OFF for correct modelling.

• ON for backward compatibility with previous versions.

Why is hydraulic conductivity measured in mm/hr instead of m/d?

TUFLOW currently uses mm/hr for hydraulic conductivity to align with the ILCL units and Green Ampt infiltration rate, which is commonly used in surface water modelling. Please convert the value if the reference uses different unit of hydraulic conductivity.

Why is the groundwater model affecting areas beyond the expected flow path?

Groundwater movement in TUFLOW is influenced by topography, soil properties, and groundwater parameters. If groundwater is introduced, infiltrated water may reappear on the surface depending on these factors.

Key considerations:

• Without horizontal groundwater movement, infiltrated water is lost from the system. When groundwater movement is enabled, water can accumulate in low-lying areas or flow towards the downstream boundary.
• Unexpected groundwater presence in areas without assigned soil depths may be due to groundwater behaviour at boundaries or model-wide settings.
• The 2d_po regional outputs introduced in version AF can assist with analysing groundwater movement and verifying model behaviour.

Should soil layers be assigned across the entire model?

If groundwater is being simulated, defining soil layers across the full model domain can provide more control over groundwater behaviour.

Options include:

• Assigning a depth of zero in areas where groundwater should not be present.
• Setting a large depth in areas where groundwater storage is needed but horizontal transmission is not desired.

Model validation using observed data is recommended to confirm that groundwater interactions align with real-world conditions.

How can a lined filtration trench be modelled?

A lined filtration trench can be represented using one of the following methods:

1D Channel Approach

• Model the trench as a 1d_nwk “Q” type structure, which uses a depth-discharge relationship.
• Connect the trench to multiple 2D cells using a 2d_bc SX line.
• If the trench is long, divide it into multiple sections to improve accuracy.

1D Pit and Culvert Approach

• Use multiple 1d_nwk Q pits connected via a 1d_nwk culvert.
• The connected 2D cells are automatically selected using the sag or on-grade method (see section 5.12.3.3 of the TUFLOW manual).
• To manually control cell connections, set the 1d_nwk Conn_No attribute to a negative value. For example, Conn_No = -1 ensures each pit connects to only one 2D cell.
• This method allows the trench to be represented as multiple pit points, each selecting one 2D cell and linking to the 1D node using X connectors.

Additional Considerations

• Pit Inlet Discharge Curve: If using a 1D network, pre-compute the discharge for various depths to define a suitable pit inlet discharge curve.
• Interflow Functionality: The use of interflow depends on the cell size relative to the trench feature.
• Two-Stage Modelling Approach: One method involves running a 2D infiltration simulation first to determine infiltration rates. The infiltration can then be applied as a 1D boundary condition in a second simulation to represent flow into the drainage system.

TUFLOW is currently developing functionality to dynamically link interflow to a 1D network, which will allow infiltrated subsurface flows to connect directly to 1D nodes or pipes in future releases.

What methods and result outputs can be used to quantify infiltration losses over a given area?

To assess infiltration losses, the following outputs can be used:

Map Output Data Types

• CI (Cumulative Infiltration): Displays the total infiltration over time.

• IR (Infiltration Rate): Shows the infiltration rate at each timestep.

• In TUFLOW 2025.2.0 version and later, QZ (average vertical flux over the past output interval) and QZI (cumulative vertical flux) can be applied.

Point Output (2d_po) Approach

• If infiltration data is needed for a specific area, a 2d_po region can be set up.

Both methods help understand how infiltration occurs across different areas.

Why does changing the initial soil moisture in the Green-Ampt (GA) infiltration method not affect infiltration rates as expected?

In the 2023 release of TUFLOW, the treatment of initial soil moisture in the GA method was updated to accommodate horizontal soil water movement. This change allows soil moisture to flow between sub-surface cells and be removed via evapotranspiration, making it more dynamic over time. Instead of remaining static in the infiltration equation, the initial soil moisture is now used as the cumulative infiltration (F) at the first timestep of the simulation. This means:

• The infiltration rate at the start of the simulation is influenced by the initial soil moisture, but subsequent infiltration behaviour depends on the balance of infiltration, drainage, and evapotranspiration.

• If no soil thickness is defined, the model assumes dZ = 0, meaning the initial moisture value is ignored altogether.

Suggested Workarounds

1. Reverting to the pre-2023 method.

Use the backward compatibility switch:

Defaults == Pre 2023

Note: This changes multiple default settings, not just the GA method.

2. Adjusting the soil porosity in the .tsoil file.

Instead of defining initial soil moisture separately set: Adjusted Porosity = Porosity - Initial Moisture

Example: If soil porosity is 0.385 and initial soil moisture is 0.200, set the porosity to 0.185.

TUFLOW is updating documentation to clarify this change and is developing a dedicated command to revert to the original GA method without affecting other default settings. Further refinements are also being considered for long-term simulations involving multiple wet/dry cycles.

Feedback or specific examples can be provided to TUFLOW Support via support@tuflow.com.

Can groundwater be trapped beneath impervious areas to prevent unrealistic exfiltration?

Currently, TUFLOW does not directly put a 'lid' on soil surface, even if the surface imperviousness is set to 1.0. The existing groundwater model calculates exfiltration (upward flux) based on mass balance only, meaning that groundwater can still migrate upward even in areas where the surface is defined as impervious.

Recognising the importance of this feature for improving groundwater representation in urban modelling, consideration is being given to incorporating a method to account for trapped groundwater as part of future TUFLOW release.

In the meantime, the issue may be mitigated by:

• Remove the soil layer under the impervious surface.
• Adjusting the soil thickness and properties to reduce unrealistic seepage effects, e.g. set horizontal hydraulic conductivity to zero for soil polygons beneath impervious areas. Note this stops groundwater from flowing to this area through lateral flux. However, water can still surge from the soil layer beneath, and then to the surface if multiple soil layers are used.

If this issue is recurring or if there are specific use cases where this feature would be beneficial, feedback is encouraged via support@tuflow.com.

What is the recommended method for representing a railway ballast area?

Traditionally, a railway ballast is modelled using layered flow constriction with high blockage and form loss value.

With the groundwater feature, it can also be represented as a soil layer (specific Soil ID) with high infiltration, suitable porosity, and high hydraulic conductivity in both horizontal and vertical directions. If the railway ballast area is the only area in the model that a soil layer is applied, the soil layers do not need to be applied across the entire model.

Why does setting up a second soil layer for a model require using the ‘CO’ type?

When defining Soil Layer 2, the convective (CO) layer type must be used. This soil type requires fewer inputs compared to other soil types that require parameters used to calculate surface infiltration. If another type is assigned, TUFLOW returns ERROR 2314 (invalid Soil ID).

To set up CO soil type:

• Define a new Soil ID for Layer 2 using the CO type in tsoil file.
• Use Set/Read GRID/Read GIS Soil Layer X commands to update soil layer under the 1st layer the new CO type.