Accurate ground cover assessment is paramount for sustainable land management and minimizing soil erosion in Australia and New Zealand. It is also critical to assess the green to dry fraction of grasslands as an important predictor of pasture quality. Our study explores the feasibility of combining hyperspectral and multispectral satellite data to improve pasture condition products, such as fractional cover for Australia and nutrient content for New Zealand.
By enabling better land condition assessments and management practices, this work will support the improvement of sustainable land use, mitigation of erosion risks, and protection of vital soil and water resources. This benefits government, industry, and NGOs in both Australia and New Zealand.
A fractional cover product quantifying the proportion of green vegetation, non-photosynthetic vegetation, and bare soil from satellite imagery supports diverse applications in agriculture, forestry, and environmental conservation. These applications require information on vegetation dynamics, soil erosion, or land degradation. Existing fractional cover models, primarily developed for Australian rangelands using multispectral moderate spatial resolution data such as Landsat and Sentinel-2, have been proven valuable for many applications. However, the current model lacks the precision needed to map the variability of New Zealand’s diverse grasslands and pastures accurately.
To improve the fractional cover model and extend its applicability beyond Australia to New Zealand’s ecosystems and for global application, we require a large amount of quality training and validation data to capture the diversity of the landscape and ground conditions. Relying solely on field surveys is impractical due to the data volume required and the associated costs and time. Our project aims to develop a cost-effective and robust method to improve the fractional cover model by leveraging hyperspectral data as a more effective, efficient, and scalable option than traditional field capture.
With the increasing availability of high-quality hyperspectral satellite data from missions such as EMIT and ENMAP, methods integrating the strengths of hyperspectral data and high-temporal-resolution multispectral data hold promise for achieving accurate and cost-effective large-scale vegetation cover characterization and monitoring.
- Validate satellite hyperspectral data in Australia
- Build approaches to apply spectral unmixing to generate fractional cover
- Use these data to improve a Sentinel 2 fractional cover model in Australia and its application in New Zealand.
- Use Hyperspectral data in conjuntion with physical based modeling to retrieve pasture condition assessments in New Zealand.