Research Programs

SAEcoMap

EO Analytics

The project aims to develop algorithms, evaluate the accuracy and produce two output processing pipelines to complete:

  • Native vegetation community and key species mapping needs for South Australian government using Kanyini Hyperscout 2 and compare these results to both ground-based (botanical survey) and drone assessment.
  • Key native, invasive and forestry vegetation type mapping for DEW and PIRSA using Kanyini Hyperscout 2 with an emphasis on determining l variations in the condition of vegetation types and determining relationship to a more variable and changing climate. Where possible, results will be compared and validated with drone assessments at PIRSA and partners’ trial sites.

Secondary project outputs are:

  • To validate Kanyini’s spectral response against a minimum of two other satellite instruments, one hyperspectral and one multispectral (e.g., Sentinel, Landsat, and hyperspectral missions such as EMIT, PRISMA, and PACE), for the purposes of cross-calibration that will form part of the Kanyini on-orbit calibration and validation plan.
  • Identify and use the images from these additional (two) satellite instruments to map native vegetation communities, the presence of invasives, and forestry plantation types and compare these to the results from both ground-based assessment and the mapping from Kanyini. This step will inform a data comparison to understand the suitability of Kanyini for vegetation community mapping.

The study will use airborne (including drones), ground-based fieldwork, and satellite hyperspectral imagery to focus primarily on sites such on Kangaroo Island and other strategic locations within SA. By comparing the Kanyini imagers’ capabilities with those of other instruments, including drones and existing satellite imagery, the project aims to advance the science of discriminating difficult-to-identify species and enhance ecological biodiversity mapping. As a preliminary project step, the project research team should identify the specific spectral bands useful for native vegetation community mapping. Wavebands known to be sensitive to detecting more vegetation stress/condition in the key vegetation types of interest to DEW and PIRSA will also be identified. This may be informed by consultation with experts within PIRSA and DEW already familiar with this problem area.

The outputs of the native vegetation mapping will seek to align with DEW’s Vegetation Community Mapping and PIRSA’s carbon sequestration monitoring objectives. The outputs of the early indicators of vegetation stress/condition relating to variations in climate will seek to align with PIRSA’s key vegetation identified as most sensitive to climate variations and be informed by DEW’s Sentinel-2
outputs identifying differences in key South Australian vegetation types.

P3.46

Project Leader:
Dr Sami Rifai, The University of Adelaide

Participants:

Developing capability to assess live coral cover and seagrass species using satellite based hyperspectral imagery

EO Analytics

Live coral cover and seagrass species composition are key indicators for scientists and managers to assess the health of coral reef habitats. The ability to differentiate between live and dead coral, seagrass species and other bottom features is driven by the density of cover and the spectral characteristics of the features. Research into the ability to detect live coral cover and differentiate seagrass species based on spectral reflectance properties has demonstrated that hyperspectral information (opposed to multispectral) is required.

Currently there have not been any successful attempts to map seagrass species composition across very large spatial extents (e.g. regional to global) nor live coral, with only a handful of examples at local scales (single reefs/meadows to small groups). As part of the Allen Coral Atlas, the UQ team mapped coral habitat globally but were not able to differentiate seagrass species and density, nor live coral cover. This mapping was constrained by the availability of only multispectral sensors (e.g. Sentinel 2, Landsat, Planet Dove) that cover large spatial extent and provide feasible complete global coverage of the world’s coral reef habitats.

The increasing numbers of hyperspectral satellites being launched since 2021, including public good and private-industry systems, capable of global coverage provides the opportunity to investigate these challenges and there is now a need to develop remote sensing frameworks to assess their capability for live coral and seagrass cover mapping. This work aims to acquire very high quality archived hyperspectral data for reef and seagrass environments, in conjunction with field data to test the coral and seagrass properties that can be differentiated from different spatial-spectral-radiometric dimensions to inform processing from future hyperspectral satellites, especially Australia’s Kanyini and the proposed AquaSat systems. Findings from this study may also provide the roadmap for a follow-up phase, to map live coral on shallow coral reefs and seagrass in Australia and globally, providing essential information to coral reef scientist and managers.

P6.05

Project Leader:
Associate Professor Chris Roelfsema, The University of Queensland

Participants:

Enhancing classification of farm-scale natural capital assets using remote sensing

EO Analytics

The demand for cost-effective information about property-scale natural capital is growing rapidly as farmers respond to opportunities and pressure from their supply chains and financial institutions to provide environmental performance information. We have developed proof-of-concept farm-scale natural capital accounts that measure and report on natural capital and environmental performance at farm-scale. Demand for farm-scale accounts is outstripping current capacity for generating them. This is largely due to the expense and technological constraints associated with classifying on-farm natural capital assets.

This project will address these constraints by harnessing the power of Machine Learning and Artificial Intelligence, state-of-art remote sensed products and our existing database of thousands of field-validated assessment points to expand and improve our approach and methods for classifying farm-scale natural capital assets. This research will underpin the further development and enhancement of
farm-scale natural capital accounts, increasing cost-efficiency and speed of generating accounts. This will broaden the capacity base for generating accounts and increase adoption of accounts for best-practice natural capital accounting and sustainability reporting in agriculture.

P3.45

Project Leader:
Associate Professor Jim Radford, La Trobe University

Participants:

SAEcoMap: Vegetation Mapping with Hyperspectral Imagery

Next Generation Earth Observation Data Services

The project aims to develop algorithms, evaluate the accuracy and produce two output processing pipelines to complete:

  1. Native vegetation community and key species mapping needs for South Australian government using Kanyini Hyperscout 2 and compare these results to both ground-based (botanical survey) and drone assessment.
  2. Key native, invasive and forestry vegetation type mapping for DEW and PIRSA using Kanyini Hyperscout 2 with an emphasis on determining l variations in the condition of vegetation types and determining relationship to a more variable and changing climate. Where possible, results will be compared and validated with drone assessments at PIRSA and partners’ trial sites.Secondary project outputs are:
  3. To validate Kanyini’s spectral response against a minimum of two other satellite instruments, one hyperspectral and one multispectral (e.g., Sentinel, Landsat, and hyperspectral missions such as EMIT, PRISMA, and PACE), for the purposes of cross-calibration that will form part of the Kanyini on-orbit calibration and validation plan.
  4. Identify and use the images from these additional (two) satellite instruments to map native vegetation communities, the presence of invasives, and forestry plantation types and compare these to the results from both ground-based assessment and the mapping from Kanyini. This step will inform a data comparison to understand the suitability of Kanyini for vegetation community mapping.

The study will use airborne (including drones), ground-based fieldwork, and satellite hyperspectral imagery to focus primarily on sites such on Kangaroo Island and other strategic locations within SA. By comparing the Kanyini imagers’ capabilities with those of other instruments, including drones and existing satellite imagery, the project aims to advance the science of discriminating difficult-to-identify species and enhance ecological biodiversity mapping.

As a preliminary project step, the project research team should identify the specific spectral bands useful for native vegetation community mapping. Wavebands known to be sensitive to detecting more vegetation stress / condition in the key vegetation types of interest to DEW and PIRSA will also be identified. This may be informed by consultation with experts within PIRSA and DEW already familiar with this problem area.

The outputs of the native vegetation mapping will seek to align with DEW’s Vegetation Community Mapping and PIRSA’s carbon sequestration monitoring objectives. The outputs of the early indicators of vegetation stress / condition relating to variations in climate will seek to align with PIRSA’s key vegetation identified as most sensitive to climate variations and be informed by DEW’s Sentinel-2 outputs identifying differences in key South Australian vegetation types.

P3.46

Project Leader:
Dr Sami Rifai, The University of Adelaide

Participants:

Kanyini Waru: Utilising Satellite Imagery to Map Land Surface Temperatures in Metropolitan and Rural South Australia During Heatwaves

Data Service Testbed

The project aims to develop a best practice satellite-based monitoring system utilising thermal imagery to measure land surface temperatures (LST), identify urban heat islands, and establish connections to human thermal comfort metrics. This initiative will concentrate on the Adelaide metropolitan region and select regional areas of South Australia (for example Port Augusta or Whyalla), investigating the potential of satellite thermal imagery to supplement or even replace the currently more costly and infrequently captured airborne thermal imagery.

A key component of the project is evaluating how the Kanyini satellite’s thermal imagery can be integrated with data from other satellite sources. The ultimate goal is to provide regular data capture, especially during heatwave events, to deliver a more efficient and timely relevant monitoring solution. The final system may integrate data from multiple satellite sources. The use of Kanyini is to highlight any benefits of having a dedicated space asset with controllable image tasking. Through this project, potential across government end users will be consulted to determine the feasibility of using satellite-based heat mapping for a range of applications, including policymaking and on-ground action.

P3.47

Project Leader:
Associate Professor David Bruce, Flinders University

Participants:

From Space to Ground: Modelling Australia’s Water Dynamics through Remote Sensing and Land Surface Modelling

EO Analytics

This work is being undertaken to address a significant gap in Australia’s water resource management by enhancing the Bureau of Meteorology’s (BoM) capability to model groundwater as part of their water accounting obligations under the Water Act 2007. Currently, BoM’s land surface model, lacks a dedicated groundwater component, leading to inaccuracies in national water accounting. Additionally, the model does not incorporate real-time observational data from satellite missions, further limiting its accuracy.

This project will develop a comprehensive groundwater model that integrates advanced data assimilation techniques and satellite-based observations, such as gravitational changes and surface water heights, to improve hydrological predictions. By creating a robust, scalable system for real-time water resource management, this project will not only meet statutory requirements but also provide valuable tools for government departments, private industry, and environmental sectors in addressing water availability, compliance, and environmental flow management. The collaboration with BoM and the use of cutting-edge satellite technologies will revolutionise water monitoring in Australia and potentially serve as a model for global water resource management.

P3.49

Project Leader:
Professor Paul Tregoning, The Australian National University

Participants: