HEC-RAS 2D Flood Modelling: When Is It Required and What Does It Tell You?

A resource consent condition requiring a "2D flood hazard study using HEC-RAS" is increasingly common on Auckland development sites near overland flow paths. It is also more complex, more expensive, and more informative than the 1D approach most engineers default to. Knowing what it requires, what it produces, and when it is actually necessary will help you brief your engineer correctly - and avoid paying for a 2D model when a 1D would satisfy the council.

This post explains the technical difference between 1D and 2D flood modelling, the specific circumstances under which Auckland Council requires a 2D study, what those model outputs mean and how they are used, how climate change is incorporated, and what you should expect in terms of cost and programme.

1D vs 2D - What's the Difference?

HEC-RAS is the Hydrologic Engineering Center River Analysis System, developed by the US Army Corps of Engineers and now maintained as free software. It is the dominant hydraulic modelling tool in New Zealand flood assessment practice and the specific software Auckland Council and most other councils expect to see used for flood hazard reports submitted in support of resource consent applications.

The distinction between 1D and 2D modelling is fundamental. It affects what the model can represent, what inputs it needs, how long it takes to run, and what the outputs mean.

1D HEC-RAS models flow along a defined channel or overland flow path using cross-sections placed perpendicular to the expected flow direction. The software solves the one-dimensional energy equation (or the Saint-Venant equations in unsteady mode) between successive cross-sections. Water can only flow in the direction defined by the cross-section sequence - upstream or downstream along that single path. This works well for well-defined channels: rivers, culverts, bridges, and simple overland flow paths where the flow direction is clear and the channel geometry can be captured adequately by cross-sections placed at regular intervals.

The limitation becomes apparent when flow spreads laterally across a floodplain, divides into multiple flow paths, ponds and fills before overflowing in a different direction, or interacts with structures in ways that alter the flow direction. In these situations, a 1D model either fails to converge, produces unrealistic results, or requires the modeller to impose assumptions about flow distribution that cannot be independently verified.

2D HEC-RAS solves the full two-dimensional shallow water equations - the depth-averaged Navier-Stokes equations, simplified under the assumption that vertical accelerations are small relative to horizontal ones - on a computational mesh built from digital terrain data. Each cell in the mesh computes water depth, velocity, and surface elevation at every timestep. Water is free to flow in any direction across the terrain, following the gradient of the water surface at each moment. A 2D model therefore captures spreading, bifurcation (flow splitting into multiple paths and rejoining downstream), ponding and gradual filling, the influence of buildings and walls on routing, and the full spatial distribution of depth and velocity across the site.

Grid resolution is a critical design parameter for 2D models. Urban models are typically run on a 1 m x 1 m computational mesh where LiDAR resolution supports it. At this resolution the model can represent individual kerb lines, building footprints, and road crossfalls that meaningfully affect how water moves across an urban site. Coarser grids - 5 m x 5 m or 10 m x 10 m - are appropriate for rural or large-catchment applications where fine spatial detail is either not needed or not supported by the available terrain data.

LiDAR requirements: A 2D model on a 1 m grid requires LiDAR data at sufficient point density and vertical accuracy - typically point spacing of 1 m or better and a vertical accuracy of approximately ±50 mm or better. Auckland Council has a 2024 LINZ LiDAR dataset available for the Auckland region at this resolution. HBRC completed a region-wide LiDAR survey in 2020. Where LiDAR is unavailable, low-resolution, or was collected before significant earthworks, the grid must be coarsened or supplementary survey data incorporated before the model can be built.

The computational cost of a 2D model scales with the number of cells in the mesh and the simulation duration. A 1 m x 1 m urban model covering a ten-hectare site contains 100,000 cells. Running a 24-hour design storm through this mesh requires solving the governing equations at every cell at every timestep - a computation that takes hours on a modern workstation, compared with seconds for a 1D model of equivalent reach length.

When Does Auckland Council Require a 2D Study?

Auckland Council does not require a 2D flood hazard study for every resource consent application. The requirement is triggered by specific site characteristics, and understanding those characteristics helps applicants and their engineers assess what level of modelling is needed before consent conditions are issued.

The primary triggers for a 2D requirement in Auckland are:

• Sites crossed by mapped Overland Flow Paths (OLFPs): Auckland's Unitary Plan maps a network of overland flow paths across the urban area. These are routes along which stormwater flows overland during storms that exceed the capacity of the pipe network. Where a development site is crossed by or adjacent to a mapped OLFP, a flood hazard assessment will typically be required. Whether 1D or 2D is needed depends on the complexity of the OLFP at that site.
• Consent conditions explicitly requiring a flood hazard assessment: Where a pre-application meeting or a notified consent process results in conditions requiring a flood hazard study, the level of modelling is usually stated in the condition text. Conditions that specify "2D HEC-RAS" leave no room for a simpler approach.
• Sites where council GIS flood maps indicate significant flow depths or velocities: Auckland's GIS portal includes regional flood hazard layers showing approximate inundation extents, depths, and velocities for various design storms. Where these layers show significant hazard on or near a development site, the council will require the applicant's engineer to verify or update those levels through site-specific modelling.
• Development that alters, diverts, or obstructs an OLFP: Any proposal to fill, pipe, divert, or build within an OLFP requires demonstration that the works do not increase flood hazard upstream or adjacent to the site. For complex OLFPs with lateral spreading, this typically requires a 2D model of both the existing and proposed configurations.

A recent project illustrates this clearly. At Glendale Road, Auckland, the council issued a resource consent condition requiring a full 2D HEC-RAS flood hazard study. Two distinct overland flow paths crossed the site - immediately ruling out a single 1D model as an adequate representation. SAE built the model on a 1 m x 1 m computational mesh using the 2024 LINZ LiDAR dataset for Auckland. The results were specific: OLFP1 carried a peak flow of 0.469 m³/s at the 1% AEP (1-in-100-year) event; OLFP2 carried 4.28 m³/s - nearly ten times larger. Maximum flood depths across the site reached 1.2 m at the 1% AEP event. Pedestrian hazard was assessed against Auckland Council's depth-velocity product criteria across the full site area - an output that only a 2D model could produce.

What a 2D Model Actually Produces

The outputs of a 2D HEC-RAS study are substantially richer than those of a 1D model. Understanding what they are and what they are used for helps clients interpret reports and understand why each element of the study is required.

Depth rasters: For each design storm - 10-year, 100-year, and 100-year plus climate change as a minimum - the model produces a spatial grid of maximum flood depths across the site. Each cell in the grid carries a depth value. These are presented as colour-coded flood maps showing where water reaches and how deep it is at the peak of each event.

Velocity rasters: The same spatial grid is produced for flow velocity. Velocity maps identify where water moves quickly - in defined flow paths and channel constrictions - and where it moves slowly, indicating ponding or backwater zones where depth may be significant but hazard from flowing water is lower.

Depth-velocity product (hazard) maps: The product of depth and velocity, expressed in m²/s, is the standard metric for assessing pedestrian and vehicle hazard in New Zealand practice. Auckland Council uses specific threshold values for this product to define zones of low, medium, and high hazard. A person can typically maintain footing in flows with a depth-velocity product below approximately 0.3 m²/s; above 0.6 m²/s, even adults are at significant risk of being swept away. The hazard map is used to determine safe access routes, building setback requirements, and the need for flood-proofing measures within the development.

Water surface elevations at specific points: The model extracts water surface elevations at nominated locations - typically the proposed building platform, the site access point, and the property boundary. These elevations are used to set minimum finish floor levels for proposed buildings. The finish floor level recommendation in a flood hazard report is derived directly from the modelled water surface elevation at the most severe design event (100-year plus climate change), with a freeboard allowance added above that level.

Flow split analysis: Where OLFPs bifurcate or where multiple flow paths cross the site, the model quantifies how much flow is carried by each path and what the spatial distribution of that flow is across the site. This is critical for understanding which building lots are within which OLFP envelope and what level of hazard applies to each.

These outputs are delivered as GIS raster layers compatible with QGIS, ArcGIS, or the council's own GIS platform, together with PDF flood maps at publication quality and data tables embedded in the engineering report. The GIS layers allow the council's planning staff to overlay the modelled flood extents against the proposed site plan and verify that buildings and access routes are correctly positioned relative to hazard zones.

The Climate Change Component

New Zealand flood hazard studies are required to assess not only current flood conditions but projected future conditions accounting for climate change - specifically, increases in rainfall intensity driven by rising temperatures that increase atmospheric moisture-holding capacity.

NIWA's HIRDS V4 tool provides design rainfall depths for any location in New Zealand at a range of return periods and storm durations. For climate change assessment, HIRDS V4 incorporates temperature-scaling factors derived from New Zealand climate projections, allowing rainfall depths to be adjusted for a nominated temperature increase.

For Auckland projects, the standard practice is to use RCP 8.5 - the high-emissions Representative Concentration Pathway corresponding to approximately 3.8°C of warming by 2100. This is more conservative than the RCP 6.0 (mid-range emissions scenario) that is typically used for Hawke's Bay assessments under HBRC guidance, reflecting Auckland Council's policy position on which scenario should govern design for structures with long design lives.

Applying RCP 8.5 to Auckland design rainfalls typically results in a 10–15% increase in rainfall depth for a given return period and duration. This translates to a broadly similar increase in peak flow - the exact relationship depends on the catchment's runoff characteristics and the SCS curve numbers used in the hydrological model. The 100-year plus climate change scenario therefore represents meaningfully more severe loading than the current 100-year event, and flood depths and velocities under this scenario are correspondingly larger.

The implication for floor level setting is direct. Finish floor level recommendations in Auckland flood hazard reports are governed by the 100-year plus climate change water surface elevation plus freeboard, not by the current 100-year level. On sites where the climate change uplift adds meaningful depth - which on smaller urban catchments is typically 100–200 mm - this can translate to a floor level recommendation that is materially higher than a report based only on current climate conditions would produce. A building consented now may stand for 80 years; the climate change scenario ensures the design accounts for conditions across that full design life.

What It Costs and How Long It Takes

A 2D flood hazard study is not a rapid or inexpensive exercise, and understanding the reasons for this helps clients plan their project programme and budget accordingly.

Data collection and hydrology: Before the model is built, the catchment must be delineated, NIWA HIRDS V4 rainfall data extracted for the relevant design durations and return periods, and design hydrographs computed using HEC-HMS or equivalent. LiDAR terrain data must be sourced from the LINZ or HBRC portals and processed into bare-earth digital elevation models compatible with HEC-RAS. This phase is broadly similar in effort regardless of whether the final hydraulic model will be 1D or 2D - the divergence in effort comes at the model-build stage.

Model build: A 2D model build on a 1 m grid for an urban site is significantly more labour-intensive than a 1D model covering the same reach. The computational mesh must be constructed, refined at critical locations such as structures, road crossings, and flow splits, and verified against the terrain. Boundary conditions - inflow hydrographs at the upstream end, tailwater conditions at the downstream end - must be specified and checked. Where flow enters the 2D domain through culverts or stormwater pipes, those structures must be represented using HEC-RAS's inline structure tools. For a site with two or three OLFPs and several structures, model build takes significantly longer than the equivalent 1D setup.

Run time and scenario set: Each design scenario requires a separate model run. A minimum set for Auckland Council submissions typically includes the 10-year, 100-year, and 100-year plus climate change events - three full runs. A 2D model on a 1 m x 1 m urban grid running a 24-hour storm may take several hours per scenario. Across three to five scenarios, model run time alone can consume a full working day.

Post-processing and reporting: Raster outputs must be processed in GIS, flood maps generated at appropriate scales for the report and council review, depth-velocity products computed and mapped, and the results written up in a form that a planning officer can review without specialist hydraulic knowledge. This phase takes as long as the model build for most projects.

Total timeframe: A standalone 2D flood hazard study - data collection, model build, scenario runs, post-processing, and report - typically takes 3–5 weeks from receipt of LiDAR data and site information. Where LiDAR must be sourced, checked against recent earthworks, and reprocessed before the model can be built, allow additional time. Where the flood study is one component of a larger infrastructure report, the overall report programme governs the timeline.

Cost: A 2D study costs more than a 1D study. SAE quotes on enquiry based on site complexity, the number of OLFPs to be modelled, whether HEC-HMS hydrology needs to be developed from scratch or can draw on prior work, and the extent of post-processing required. Projects with a clear single OLFP and available catchment data cost less than projects with multiple interacting flow paths, significant in-channel structures, and no prior hydrological analysis.

When 1D Is Sufficient

Not every flood hazard assessment requires a 2D model, and it is worth being clear about the circumstances in which 1D remains the appropriate and council-acceptable approach.

• Single, well-defined OLFP with no bifurcation or significant ponding: If the overland flow path on a site is clearly defined - a single linear path following a valley or swale with no evidence of lateral spreading - a 1D model adequately represents the hydraulics. Cross-sections are placed along the path, the energy equation solved between them, and the result gives a conservative flood level envelope. This is appropriate and sufficient for many straightforward residential sites.
• Rural culvert sizing: Culvert hydraulics are inherently one-dimensional - the flow path is the pipe, and the key outputs are headwater and tailwater levels. HEC-RAS 1D with culvert routines is the standard and appropriate tool for this application.
• Sites where council maps show a well-defined OLFP with no spread risk: If Auckland Council's GIS layers show a narrow, clearly mapped OLFP crossing the site and consent conditions do not specifically require 2D modelling, a 1D approach may be accepted. Confirm this with the council at the pre-application stage rather than assuming.
• Pre-development due diligence: Where a client needs an approximate assessment of flood risk before committing to a site purchase or development concept, a 1D model built from available GIS cross-section data can provide a conservative flood level envelope quickly and at lower cost than a full 2D study. This is not a substitute for a consented 2D study where one is subsequently required, but it serves a different purpose earlier in the project timeline.