Methodology

Built the way a CPEng would defend it.

Every overland flow path and flood depth layer is derived from current LiDAR, processed with open and auditable methods, and aligned with the standards New Zealand councils use for consent assessment.

Overview

FloodMap.nz publishes a consistent nationwide overland flow path dataset derived from the LINZ 1m LiDAR composite. Flow paths are classified into six tiers by upstream catchment area. Site-specific flood assessments — adding HEC-RAS 2D hydraulic modelling, depth and velocity outputs — are scoped per site as bespoke engineering through SAE Ltd; the methodology for those is documented below for reference.

The dataset is produced from the DEM up using open-source geospatial tools. No single council's model is embedded in the output — every tile is reprocessed when new LiDAR is released.

Data sources

Digital Elevation ModelLINZ New Zealand LiDAR 1m DEM composite (layer 121859), 2018–2024 surveys
Land coverLCDB v5 (Landcare Research, 2018) — 33 classes mapped to Manning's n
SoilsFundamental Soils Layer (FSL) + S-Map where published; mapped to NRCS Hydrologic Soil Group (A/B/C/D)
RainfallNIWA HIRDS v4 (High Intensity Rainfall Design System); TP108 regionalised for Auckland
BasemapsLINZ Basemaps (aerial); OpenStreetMap (streets)
AddressesLINZ NZ Street Address (for geocoding + property reports)

All data sources are publicly available under CC BY 4.0 or equivalent open licences.

6-tier catchment classification

Flow paths are classified by upstream contributing catchment area. Tier thresholds align with typical council mapping conventions so the layers map cleanly onto consent requirements.

TierCatchment (ha)Typical useVisible zoom
T1>500Regional rivers — use NIWA REC2 for authoritativez7+ (Phase 2)
T2100–500Major streams, district plan hazard zonesz7+ (Phase 2)
T325–500Streams (current published tier)z7+
T42–25Minor watercourses, subdivision designz11.5+
T50.4–2Fine OFPs (AUP E36 trigger = 4,000 m²)z13+
T6<0.4Property-scale OFPs (on-site drainage)Phase 2

Published today: T3, T4, T5 nationwide. T1/T2 are reserved for authoritative sources (NIWA REC2) because 1m DEM block-boundary accumulation errors make large-catchment upstream areas unreliable. T6 is under development.

Processing pipeline

For each region (defined by LINZ survey block extent), the pipeline runs:

  1. VRT mosaic — tiles assembled into a region-wide virtual raster
  2. Depression handling — per-block using WhiteboxTools FillDepressions with MINSLOPE=0.0001 for ponding detection and MINSLOPE=0.01 for stream delineation
  3. Flow accumulation — D8 algorithm via WhiteboxTools D8FlowAccumulation
  4. Stream extraction — GRASS r.stream.extract with tier-specific thresholds (4,000 m² / 20,000 m² / 250,000 m²)
  5. Vectorisation — GRASS v.to.points to extract outlet points, then joined back with MAX(ABS(upstream_cells)) for robustness across multipart lines
  6. Classification — each line segment tagged with its maximum upstream catchment, binned into tiers
  7. Tile packaging — output vector-tiled to PMTiles for hosted delivery on Cloudflare R2

Blocks are processed at 12 km with 4 km overlap on the North Island, 10 km blocks on the South Island (alpine terrain hang-risk). Overlap regions are merged with COALESCE(ABS(upstream_cells), 1000) to recover any block-boundary NULLs.

Hydraulic modelling (site-specific assessments)

Site-specific flood assessments — commissioned through SAE Ltd as bespoke engineering — add a HEC-RAS 2D layer on top of the published OFP network:

  • Grid resolution: 2–5 m mesh (user-configured per catchment area)
  • Rainfall: TP108 design storm, 10%, 2% and 1% AEP, 24 hr duration (or regional equivalent via HIRDS v4)
  • Impervious surfaces: derived from aerial imagery via ExG (Excess Green) threshold, masked by LCDB urban classes
  • Curve numbers: FSL Hydrologic Soil Group + LCDB land cover mapped through TR-55 AMC II table
  • Manning's n: LCDB v5 classes mapped per SWCoP v4 (buildings 0.150, sealed 0.015, grass 0.030, forest 0.100)
  • Solver: HEC-RAS 6.7 implicit 2D, unsteady
  • Output: maximum depth raster, velocity raster, depth × velocity product, hazard grid

For before/after earthworks modelling, the client's proposed contour is burned into the DEM and the catchment is re-run. Output includes depth-difference rasters and a summary of AEP depth at the property boundary.

Standards alignment

TP108 (Auckland)Rainfall design storm, runoff coefficients
SWCoP v4 (Auckland Council Stormwater Code of Practice)Manning's n tables, impervious surface definitions, OFP requirements
NZS4404:2010Land development and subdivision engineering
AUP E36Overland flow path regulation — 4,000 m² catchment trigger
HEC-RAS 2D User Guide (USACE)Hydraulic modelling conventions
NRCS TR-55Curve number method, AMC II assumption

Where a regional council publishes its own stormwater code (e.g. Christchurch Waterways Wetlands and Drainage Guide), site-specific assessments for that region default to the local standard.

Limitations

What the OFP layer does not represent:

  • Underground stormwater pipe networks, culverts, and constructed drainage are not modelled — flow paths are surface-terrain only
  • The layer is a modelled terrain estimate, not a field-verified watercourse or an authoritative flood record
  • Minor artefacts may occur at LiDAR tile boundaries, especially in flat or heavily developed areas
  • Buildings and vegetation are represented by the LiDAR surface, which can produce spurious flow paths around large structures
  • T1 / T2 (large catchment) tiers are not yet published — for authoritative large-river hazard, use NIWA REC2 or council maps
  • Wellington region has partial coverage pending LINZ LiDAR infill for the southern area

The free map layer is a general reference tool. Decisions on resource consent, building consent, property purchase, insurance, or infrastructure design should be based on a site-specific flood assessment, a LIM from your council, or advice from a qualified engineer.

Versioning & updates

Each published dataset carries a version tag showing the LiDAR survey vintage and pipeline version. The update cycle is:

  • Quarterly: affected regions reprocessed as LINZ publishes new survey blocks
  • Annual: full nationwide rebuild to capture DEM composite updates and any methodology changes
  • Major revisions: versioned with changelog announced via the contact channel; council partners are notified directly in advance

Current published version: v2.0 (April 2026) — LINZ L121859 composite, 6-tier classification, T3/T4/T5 publishing.

Full methodology PDF

CPEng-signed 12-page methodology statement for council submission. Coming soon.

Request a copy →