Extract from the Executive summary

This project uses high spatial and temporal resolution satellite imagery alongside machine learning techniques to identify “bare earth” and unsealed roads across the Auckland region, specifically using Maxar 0.3m RGBI imagery from March/April 2023 (post cyclone Gabrielle). The model analysis defines areas of bare earth as gravel, exposed soil, and sand. 

GIS (Geographic Information System) processing techniques were used to consider key features, in conjunction with bare earth and unsealed roads, including intersections with overland flow pathways, river networks, and steep slopes. 

The recently developed Sediment Sources Rapid Identification and Management System (SSRIMS) method, piloted on the Rodney region in 2023 (this is basically a set of scripts and GIS tools customised for Auckland Council), was used to score and rank unsealed roads for purposes of management. This has the potential to inform compliance, management, and monitoring across Auckland, utilising near real-time satellite imagery.

These resulting datasets can be used to provide further understanding and monitor the complex interactions of sediment within sensitive receiving environments and waterways. Additionally, the identification of unsealed roads, within high-risk environments, can be used to prioritise sealing works and road improvement works, such as through the Unsealed Roads Improvement Framework that Auckland Transport leads or to support the strategic management of potential sediment generating areas, across the entire Auckland region.

The 778 kilometres of unsealed roads, in the sample data set for the Auckland Region, are represented by 79,205 segments. Each segment has its own score and contributes to the overall score of the road. The total of this score is referred to as the High Sediment Potential (HSP) score. This is composed of weighted values for road slope, overland flow path proximity, distance to bare earth, proximity to Ecosystem Current Extent, and intersection with rivers.

The geospatial scoring method identified all roads (made up of segments) considered to have an HSP score of eight or more. Of the 809 unsealed roads, 267 (sub samples) had lengths/ segments that were scored eight or more, this represents approximately 33% of all the unsealed roads in the region. ...

Lynker Analytics, Auckland Council, March 2024.

See also

Mapping bare earth areas from Cyclone Gabrielle 2023. Data quality, accuracy and development report