NatureScan's drone preparing for take-off. Credit: UTAS.
Taking to the skies. Credit: UTAS.
A view of our first restoration area. Credit: UTAS.
This image was taken from a DJI Mavic 3M flight at 110 metres and has a pixel resolution of 3cm. Credit: UTAS.
NDVI (Normalised Difference Vegetation Index) is a plant health visualisation technique. In the overlay image, higher NDVI is shown in red, highlighting new growth, with lower values in yellow to green, picking up on the less photosynthetically active grasses. Credit: UTAS.
This image is a false colour composite set up to identify vegetation that strongly reflects near infrared light. In this image, the trees and the younger and/or healthier grass is bright red. Credit: UTAS.
Date
29 May 2024 (and ongoing)
Investigators
Alice Robbins
Professor Arko Lucieer
Associate Professor Ben Sparrow
Dr Juan Carlos Montes Herrera
Partners
University of Tasmania
The University of Adelaide
Terrestrial Ecosystem Research Network
Department of Climate Change Energy Environment and Water (DCCEEW)
Aim
To use consumer-grade drone data to map and monitor how landscapes respond to management practices and restoration projects.
Background
To monitor how a landscape is changing, you need to map it at recurring intervals, but to date, mapping techniques have been limited.
Field surveys, for example, tend to be labour- and time-intensive. They generally cover a small area (about 1 hectare) and take about a day to complete (depending on the sampling scheme). Field surveys capture data at a very fine scale, but this data usually lacks spatial coordinates.
Satellite data has spatial coordinates, but it often has coarse resolution. For publicly available satellite imagery, the finest resolution is 10 metres, so you lose a lot of fine detail about vegetation composition and structure. Satellite platforms are also impacted by cloud coverage, and you have no control over when they capture imagery over an area.
Drones can capture fine-resolution data in short flight times, can be flown on demand and are less impacted by clouds. However, most mapping is done with research-grade drones, making this technique inaccessible to land managers.
The NatureScan project is solving this inaccessibility problem by using consumer-grade drones to map landscapes.
In a recent ~40-minute flight NatureScan captured 50 hectares at ~1 cm spatial resolution (the pixel size).
NatureScan uses consumer-grade drones because they require far less knowledge to operate and are more affordable. The project aims to accelerate the uptake of these kinds of drones by land managers by developing data collection and processing protocols and harnessing AI to derive essential biodiversity variables.
The NatureScan project builds upon previous and current work in the Terrestrial Ecosystem Research Network (TERN), addressing a critical need for rapid, cost-effective and accurate biodiversity monitoring.
The Quoin as a case study
The first stage of the NatureScan project is to collect data from a range of ecosystems and develop appropriate protocols to enable consistent data collection and the processing of analysis-ready RGB and multispectral datasets from raw imagery. This data will provide a foundation for developing AI techniques to derive essential biodiversity variables from drone imagery.
The Quoin is one of these ecosystems.
Drone imagery was collected at The Quoin’s Honeysuckle Hill restoration site using an affordable DJI Mavic 3M drone. The site will be remapped after significant restoration and regeneration works, serving as a key case study for the NatureScan project.
Data was also collected using a DJI Matrice 350 RTK drone equipped with more research-grade sensors including the MicaSense RedEdge-MX Dual Camera system, Zenmuse P1 RGB mapping camera and the Zenmuse L2 LiDAR sensor.
Using consumer- and research-grade sensors will enable a comparison of data quality, particularly the point cloud data derived from photogrammetry and LiDAR.
Significance
The NatureScan project aims to advance biodiversity monitoring in Australia by leveraging low-cost drone technology, cutting-edge artificial intelligence and expertise in field ecology.
The project will develop drone remote sensing techniques for national-scale biodiversity monitoring and contribute accurate baseline information and detailed spatio-temporal data to inform the Federal Government’s Nature Positive Plan and Nature Repair Market.