2D Flood Modelling for Auckland Residential Development: A Practical Guide

Auckland Council's online flood viewer identifies your property as flood-prone, but the flood viewer uses regional-scale modelling that may not capture site-specific flow paths, overland flow, or the effect of proposed earthworks on your neighbours. For resource consent purposes, a 2D HEC-RAS model at 1 m x 1 m resolution often resolves flood extents that are materially different from the regional mapping. At Glendale Road, Henderson, site-specific 2D modelling revealed overland flow path conditions that the standard council maps had missed.

Why the Council Flood Viewer Is Not Enough

Auckland Council's GIS flood viewer displays flood extents derived from regional-scale hydrological and hydraulic models. These models are built at a resolution appropriate for regional planning: they identify broad flood-prone areas and inform the Auckland Unitary Plan's flood hazard overlays. They are not designed for site-specific engineering assessment.

The limitations of regional modelling become apparent at the site scale. A regional model might use a 5 m or 10 m terrain grid, which smooths out fences, walls, kerbs, building footprints, and minor topographic features that control how water moves across a residential site. An overland flow path that is directed between two buildings and across a neighbouring property at 1 m resolution may appear to flow through the middle of a house at 10 m resolution, simply because the model cannot resolve the building footprint.

For resource consent applications on sites within or adjacent to a flood hazard overlay, Auckland Council typically requires a site-specific flood assessment. On sites where overland flow paths are involved or where the proposed development could affect flood levels on neighbouring properties, a 2D hydraulic model is the appropriate tool.

What a 2D HEC-RAS Model Does

HEC-RAS (Hydrologic Engineering Center, River Analysis System) is a hydraulic modelling package developed by the US Army Corps of Engineers and widely used in New Zealand for flood assessment. The 2D solver in HEC-RAS models water flow across a two-dimensional terrain surface, solving the shallow water equations at each computational cell.

Unlike a 1D model (which assumes flow is confined within a defined channel and moves in one direction), a 2D model allows water to flow in any direction across the terrain. This is critical for urban flood assessment because floodwater in residential areas does not follow a single channel. It spills over banks, flows along roads, ponds behind fences, and splits around buildings. A 2D model captures all of these behaviours.

The key inputs to a 2D HEC-RAS flood model are:

• Terrain model. A high-resolution digital terrain model (DTM) derived from LiDAR data, typically at 1 m x 1 m resolution for site-specific urban modelling. The terrain model defines the ground surface over which the model routes water.
• Rainfall or inflow hydrograph. The design rainfall event (for pluvial flooding) or a river flow hydrograph (for fluvial flooding) that drives the model. The rainfall is typically derived from HIRDS (High Intensity Rainfall Design System) for the site's location.
• Manning's roughness values. Surface roughness parameters that represent the resistance to flow from different land covers (grass, pavement, buildings, fences). These values are assigned to the 2D mesh based on land use mapping.
• Boundary conditions. The downstream boundary condition (typically a normal depth or stage-discharge relationship) and any upstream inflow boundaries that represent flow entering the model domain from outside the site.

What Resolution Matters and Why

Resolution is the single most important parameter in a 2D flood model for urban residential sites. A 5 m mesh cannot resolve a 3 m wide driveway, a 1.2 m high fence, or a 200 mm kerb. These features control flood routing in residential areas, and if the model cannot see them, it cannot model their effect.

At Glendale Road in Henderson, the 2D model was built at 1 m x 1 m resolution using LiDAR-derived terrain data. At this resolution, the model resolved individual properties, fences, driveways, and the road kerb profile. The result was a flood extent map that showed overland flow concentrating along a specific path between existing buildings, a flow path that was not visible in the regional mapping.

The trade-off with resolution is computational time. A 1 m mesh over a 10-hectare domain produces one million computational cells, and the model must solve the shallow water equations at each cell for every time step across the full duration of the design storm. For a 100-year, 24-hour storm on a model of this size, the simulation can take several hours on a modern workstation. This is manageable for a site-specific assessment but would be impractical at a regional scale, which is why regional models use coarser resolution.

Modelling Pre-Development and Post-Development Scenarios

For resource consent purposes, the 2D model is typically run for at least two scenarios: the existing (pre-development) terrain and the proposed (post-development) terrain. The post-development scenario incorporates proposed earthworks, building platforms, retaining walls, and any changes to the stormwater conveyance system.

The comparison between these two scenarios is what the council assessor is most interested in. If the proposed development increases flood levels or flood extents on neighbouring properties, the consent will be either declined or conditioned to require mitigation measures. If the development reduces or maintains existing flood conditions, the flood assessment supports the consent.

At Glendale Road, the 2D model was run for both scenarios, and the results demonstrated that the proposed development, combined with the stormwater detention system, did not increase flood levels on adjacent properties during the 1% AEP (100-year) design storm. This was the evidence required to satisfy the flood hazard condition of the resource consent.

Common Issues in Auckland 2D Flood Models

Incorrect terrain data. The quality of the model output is entirely dependent on the quality of the terrain input. If the LiDAR data is outdated (captured before recent earthworks or building construction), the model will not reflect current ground conditions. For sites where significant changes have occurred since the last LiDAR capture, a site-specific topographic survey should be used to update the terrain model in the immediate vicinity of the development.

Inappropriate boundary conditions. The downstream boundary condition determines how water exits the model domain. If this boundary is set incorrectly (for example, a fixed water level that does not reflect actual tailwater conditions), the model will produce unreliable results near the downstream boundary. The boundary should be placed far enough downstream that it does not influence flood levels at the site.

Missing features in the terrain model. LiDAR captures the ground surface, but it does not capture all hydraulically significant features. Fences, retaining walls under vegetation, and underground culverts are not visible in LiDAR data. These features must be added to the model manually using survey data or site inspection records. A culvert that is not represented in the model will appear as a solid blockage, artificially raising upstream water levels.

Insufficient model domain. The model domain must extend far enough beyond the site to capture all flow paths that could affect the site. A model that clips the domain at the site boundary will miss overland flow entering the site from upstream properties. The domain should extend to the nearest ridgeline or catchment boundary upstream, and to the nearest natural drainage channel or stormwater main downstream.

When a 2D Model Is Required vs When It Is Not

Not every Auckland resource consent requires a 2D flood model. For sites that are clearly outside any flood hazard overlay and where no overland flow paths cross the property, a desktop assessment or a simple 1D analysis may be sufficient. The 2D model is typically required (or strongly recommended) in the following situations:

• The site is within or adjacent to a flood plain or overland flow path identified in the Auckland Unitary Plan
• The proposed development could affect flood levels on neighbouring properties (for example, by filling part of a floodplain)
• The site topography is complex and flow behaviour cannot be adequately represented in 1D
• The consent conditions explicitly require a 2D hydraulic assessment

For Auckland residential subdivisions, particularly in areas like Henderson, Massey, and the western suburbs where Waitemata Group clay soils produce rapid runoff and where overland flow paths are common, 2D modelling has become standard practice for any site with a flood hazard overlay.