Grazing Intelligence

“This project has delivered unparalleled predictive accuracy that dynamically accounts for environmental variability, populating producers’ devices with actionable insights.”

‍Ollie Roberts, Pasture.io

‍

⬇️Download the project explainer (.pdf)

‍

The Challenge

Farmers generally make subjective decisions about grazing management and supplementary feeding requirements. There is a high degree of uncertainty about these decisions in extensive sheep and cattle systems, particularly as farmers are heading into dry conditions, as there are limited options to get real-time feedback on decisions. There is also a lot of complexity due to variation in landscapes and seasonal conditions.

Furthermore, animal weight and production data is collected infrequently and is not linked to pasture observations. This leaves long lead times for issues in animal performance to arise. It also makes it difficult for producers to link day-to-day decisions with livestock performance. This separation between data and decision can impact on the business bottom line. 

The Solution

Data on pasture biomass and composition and liveweight change of animals was collected on four NSW DPI research stations at Orange, Cowra, Trangie and Glen Innes and on commercial farms. This data was then aligned with other data sources, such as mob movements, Sentinel 2 imagery, weather, and soil data for analysis of grazing events in individual paddocks.

Machine learning methods (Xgboost) were used to develop pasture predictions of green and brown dry matter (DM) and plant functional categories (based on taxonomic or selective grazing criteria), lumpiness or height using relationships between bare ground and DM, and growth of individual species and functional categories from remote sensing, weather, and mob data.

Delivered through the Pasture.io software, farmers can now predict feed availability and estimate animal performance and make confident decisions regarding purchasing and de-stocking.

Outcomes

• Improved livestock performance through better feed allocation and early identification of deficits
• Reduced Greenhouse Gas emissions through improving livestock production efficiency, such as selling young stock earlier or at a heavier weight and maintaining less breeders
• Reduce the environmental impact of livestock grazing through projecting pasture DM, and key components of pasture forward to guide proactive management.

Project Updates

December 2023 - Final Report: Grazing Intelligence

December 2022 - Research Symposium: Warwick Badgery, NSW DPI

‍