DRAFT Direct Rainfall (Rain on Grid) Modelling Guidance
Direct Rainfall
Rainfall depth can be applied directly to 2D cells across the entire catchment to simulate catchment runoff. This direct rainfall approach enables the integration of catchment hydrology and hydraulic modelling within a single framework.
Comprehensive detail is provided in the Rainfall Boundaries section of the Manual. https://docs.tuflow.com/classic-hpc/manual/latest, including input data requirements, available modelling approaches, methods for rainfall losses and adjustments, and other key considerations for producing a fit-for-purpose model.
Key Considerations for Model Configuration
The following list is not exhaustive, but provides key considerations for a basic direct rainfall model. See the Manual for further detail.
- It is highly recommended to use the TULFOW HPC Solver and sub-grid sampling (SGS)
- It is recommended to consider reducing Cell Wet/Dry Depth from the default 0.002m (or 0.007ft) to less than a mm (e.g. 0.0002m or 0.0007ft), due to substantial sheetflow. This applies for both Classic and HPC solution schemes, unless using sub-grid sampling (SGS). however, it should be reviewed and selected according to the magnitude of flooding depths. sensitivity test.
- For models with SGS turned off, a reduced cell wet/dry depth of 0.0002m (0.0007ft) is recommended due to the substantial amount of shallow sheet flow.
If SGS is turned on, a reduced cell wet/dry depth is not necessary because the cell wet/dry calculation conducted on the SGS storage curve captures a greater change in depth. In the case, the default of 0.002m (0.007ft) may be sufficient; however, it should be reviewed and selected according to the magnitude of flooding depths. - TUFLOW Executable (Model Precision):
- When using TUFLOW HPC, the single precision version of TUFLOW (iSP) is likely sufficient for direct rainfall models. Double precision version of TUFLOW (iDP) may be required particularly for simulations with very small inflow boundary values, long term simulations, fine scale processes such as groundwater, or 1D-2D linked models at high elevations.
- When using TUFLOW Classic, the double precision version of TUFLOW (iDP) is likely required for direct rainfall models.
- it is recommended to sensitivity test running the model with both single and double precision. If the results are unacceptably different or if the mass error is significantly lower in iDP, then iDP should be used.
- Note that the choice of single or double precision also impacts simulation times and memory allocation. Double precision (iDP) versions of TUFLOW Classic:
- Have approximately 25% longer simulation time for TUFLOW Classic, and for TUFLOW HPC can be up to four times slower than iSP on some devices.
- Require up to twice the memory usage which limits the ability to run concurrent simulations
- It is highly recommended to adjust the Manning’s n values to calibrate the hydraulic response of catchments. Application of depth varying roughness (see Section 7.2.7.3) or “Law of the Wall” approach (Section 7.2.7.2) can be also considered to take into account of the high roughness at shallow depth and/or to counter the overestimation of the pressure gradient in a direct rainfall model.
- Map outputs:
- Activate the following Map Output Data Types, as applicable: RFR, RFC, RFML, CI and IR.
- Apply Map Cutoff Depth and Map Cutoff Vector to facilitate fit-for-purpose mapping that limits the display of user defined shallow depths (does not affect hydraulic computations).
Useful Resources
Tutorial models
Example models
EG03_005 to EG03_014, 15 002 TUFLOW Example Models
The following links provide boundary condition guidance for applying direct rainfall:
Webinars
Please see Direct Rainfall (Rain on Grid) Webinar.
2023 11 Direct Rainfall Modelling in TUFLOW - YouTube https://www.youtube.com/watch?v=qDlrIsWlK1Q
Feb, 2021: Is Direct Rainfall Modelling Accurate? https://www.tuflow.com/library/webinars/#feb2021_direct_rainfall
The following links provide boundary condition guidance for applying direct rainfall:
Publications and media
2023-continuous-direct-rainfall-hydraulic-modelling-with-coupled-surface-ground-water-interaction-gao-et-al-hwrs.pdf
2022-direct-rainfall-hydraulic-model-validation-ryan-et-al-hwrs.pdf
Rainfall Losses and Soil Infiltration
TUFLOW supports a range of rainfall loss and soil infiltration options. These are split into two broad categories, defined by their respective calculation methods:
- Rainfall Excess Losses
- Soil infiltration Losses
The fundamental difference in calculation approach is described in the below Common Questions Answered (FAQ) section.
Rainfall Excess losses can be implement in a variety of ways, including
- Loss values specified within the TCF Read Materials File command, linking to TGC Read GIS Mat inputs;
- Loss values specified within the GIS layers associated with the TBC Read GIS SA RF command; or
- Global Rainfall Initial Loss and Global Rainfall Continuing Loss TBC commands.
Similarly, a variety of soil infiltration options are supported. The available options include, Initial / Continuing Infiltration, the Horton Infiltration method and the Green-Ampt Infiltration method. The following link provides further discussion on the Green-Ampt method.
Common Questions Answered (FAQ)
What is the difference between rainfall excess and soil infiltration?
Rainfall Excess Approach - Read Materials File and Read GIS SA RF continuing loss specification:
- This initial and continuing loss approach is a simplistic calculation method comparable to the loss methods included in traditional hydrology models (e.g. RORB, URBS, WBNM etc).
- The calculation approach is as follows:
- The user defines the rainfall hyetograph (time vs depth (mm)) boundary condition inputs.
- The rainfall value is reduced by the loss value (i.e. rainfall excess) before the boundary condition input is applied to the 2D cells.
- The GIS SA RF takes the rainfall excess (units = m) and multiples the value by the Area attribute (units = m2) in the GIS object to convert rainfall depth to a volume (m3).
- TUFLOW applies the calculated rainfall excess flow to the lowest cell in each GIS polygon during the first timestep that wetting occurs. Every timestep thereafter the inflow is distributed over the wet cells within the polygon.
Soil Infiltration Approach - TUFLOW Soils File (.tsoilf):
- This approach is a more realistic representation of the actual physics associated with water infiltration into the soil.
- The calculation approach is as follows:
- The user defines the rainfall hyetograph (time vs depth (mm)) boundary condition inputs.
- The total rainfall value is applied directly to every 2D cell within the 2d_rf polygon.
- When a 2D cell is wet the soil infiltration function subtracts the appropriate loss volume of water from it. Computationally this is referred to as a “Sink” term.
What is the best approach for modelling buildings in rain on grid model?
There are numerous different industry standard ways to represent buildings in a direct rainfall model. Australian Rainfall and Runoff Guideline, Project 15 (Representation of Buildings in 2D Numerical Flood Models discusses some of the available options. Common TUFLOW modelling approaches are summarised below:
- Using depth-varying Manning's n over the area of the building footprint, with the application of a lower value (n = 0.02) at shallow depths (d < 0.03m) and a higher value (n > 0.3) at a more significant depths (d > 0.1m):
- This is a very common and easy to implement option. The low Manning's n value aims to mimic the quick runoff response associated with drainage from the roof. The higher Manning's n value aims to represent the losses associated with deeper floodwater impacting the side of the building.
- Raise the building footprint elevation using TUFLOW's topographic update features (e.g. Read GIS Z Shape):
- Raising the model topography creates an obstruction to flow. It prevents floodwater from passing through buildings (as is the case with the Manning's n approach)
- The application of rainfall on top of the building can however produce some undesirable results that require further consideration. Those being, water falling from the building roof to the ground can require a reduced model timestep to maintain model stability which can slow the simulation speed. Also, depending on the Map Cutoff Depth assumptions, water may be present in the results on the building rooftops. This may not be desired for mapping purposes. There are a variety of options available to resolve these issues:
- Retain the model design described above, with subsequent post-processing of the mapped results before reporting. Either, delete the model result where there is overlap with the building footprint, or overlay the building footprint polygon objects over/above the result dataset in the GIS Map Layout (hiding the flood model result within the building footprints).
- Exclude buildings from the rainfall polygon: This removes the rainfall from the model that would otherwise fall on the buildings. This approach will under-estimate the amount of rainfall entering the model. If the collective building footprint area is negligible in comparison to the entire model, this approach may be acceptable.
- Exclude buildings from the rainfall polygon and represent the rainfall that would be falling on the building using Read GIS SA RF inflow boundaries. To do this, digitise a 2d_sa_rf polygon for each building (with a buffer of one of more 2D cells) where the building footprint has been excluded from the direct rainfall region. The 2d_sa_rf input will convert the input rainfall hyetograph to flow, deposited initially on the lowest 2D cell, then for subsequent timesteps distributed over all wet cells, within the 2d_sa_rf regions (i.e. on the ground surrounding the building). Refer to TUFLOW Example model EG03_014.tcf for a demonstration of this inflow boundary condition configuration.
- Exclude buildings from the rainfall polygon and represent the rainfall that would fall on the building using Read GIS SA RF PITS inflow boundaries. This approach is similar to the previous method, although instead of directing the inflow to the ground surrounding the building, it is directed into the sub-surface drainage (underground pipe) system. To implement this approach every 2d_sa_rf polygon must encompass at least one 1D pit. If multiple 1D pits are within a single 2d_sa_rf region, the flow from the polygon is split equally between the 1D pits. The pits can be snapped to a 1D node, connected via the pit-search distance or via an x-connector, to allow captured by the pit to enter a 1D representation of the sub-surface drainage system. Refer to TUFLOW Example model EG15_000.tcf for a demonstration of this inflow boundary condition option.
| Up |
|---|