Enhanced fractional cover models using hyperspectral data for improved pastoral condition assessment in Australia & New Zealand
The Challenge
Accurate assessment of pasture quality and condition is essential for sustainable land management across Australia and New Zealand. However, several key challenges limit current capabilities.
Traditional approaches for measuring pasture quality, such as Near Infrared Spectroscopy and metabolisable energy assessments, are not scalable or timely for operational use at national or farm levels. This has created a critical information gap, particularly for farmers seeking data to inform grazing decisions, feed budgeting, or land condition assessments.
Existing fractional cover models that estimate green/dry vegetation and bare soil are widely used as a key indicator of total standing dry matter. These models do not transfer well to environments outside of the Australian rangelands such as New Zealand’s diverse grasslands and intensively managed pastures. In both countries, the collection of high-quality, standardised pasture quality data or calibration field data is limited by the high cost, the logistical complexity of field sampling and laboratory analysis.
Despite the emergence of powerful new spaceborne hyperspectral sensors (e.g. EMIT and EnMAP), their potential to enhance pasture monitoring had not been fully tested. There was a need to evaluate whether these data sources could support improved fractional cover modelling and estimation of pasture quality metrics.
The project also addressed the challenge of developing transferable, interoperable tools that could be applied across contrasting landscapes in Australia and New Zealand.
Partners
This project was a collaboration between FrontierSI and Cibo Labs in Australia and Manaaki Whenua – Landcare Research and Lincoln Agritech in New Zealand.
Funding for this project was provided by SmartSat Cooperative Research Centre, Ministry of Business, Innovation & Employment and New Zealand Space Agency Catalyst Fund.
The Solution
This trans-Tasman project developed innovative, space-based solutions to improve the monitoring of pasture condition and quality in New Zealand and Australia.
New Zealand Team – Pasture Quality Mapping
Landcare Research with support from Lincoln Agritech focused on quantifying pasture quality from spaceborne data, targeting metrics such as protein content, metabolisable energy (ME), and digestibility. Using EnMAP hyperspectral imagery and the Prospect-Pro radiative transfer model, they demonstrated that high-quality pastures could be characterised based on reflectance signatures between 1,200 – 2,500 nm. These spectral profiles enabled the estimation of key leaf biochemical properties, including protein and water content. In parallel, they developed linear regression models linking Sentinel-2 spectral bands to historical ME and digestibility measurements from over 800 pasture sites. These models provide a practical basis for operational mapping of pasture quality at national scale.
Australian Team – Enhanced Fractional Cover Modelling
FrontierSI and Cibo Labs focused on improving Sentinel-2-based fractional cover models using hyperspectral data. By applying spectral unmixing techniques to EMIT imagery, they extracted high-quality endmembers representing bare soil, green vegetation, and dry vegetation that improved the accuracy of fractional cover predictions. Their method integrated hyperspectral data with Sentinel-2’s to support frequent, detailed mapping of total standing dry matter which is critical for feed budgeting and land condition assessments.
Together, the teams laid the foundation for a pasture monitoring system supported by free satellite data that is scalable for diverse farming systems.
Impact
This project demonstrated that satellite-based hyperspectral and Sentinel-2 imagery can deliver accurate, timely, and scalable assessments of pasture quality and condition across Australia and New Zealand.
In New Zealand, key pasture quality metrics such as protein content, metabolisable energy (ME), and digestibility were successfully inferred from EnMAP data and Sentinel-2, enabling near real-time paddock-level insights to support better grazing and feed decisions.
In Australia, EMIT hyperspectral data was used to enhance a Sentinel-2 fractional cover model through improved estimation of green vegetation, dry vegetation, and bare ground fractions. This supports more accurate, high-frequency monitoring of total standing dry matter, especially in areas with limited field data.
Together, the teams developed methods that reduce reliance on field sampling, lower monitoring costs, and support data-driven decision-making. The collaboration lays the foundation for a unified, operational pasture monitoring system that can be trialled in a future phase and adapted to diverse farming systems.
Contact
To learn more, contact FrontierSI at contact@frontiersi.com.au.