Preview

PDF Pending copyright assessment - Request a copy 98Kb

For effective pasture utilisation, livestock grazing systems require real time pasture monitoring and measurement. Manual techniques are laborious and have limited large-scale application due to the area of the paddocks required to be monitored. The recent inception of large, freely available daily volumes of data from earth observation sources have potential to aid real time management of pastures remotely, saving livestock managers both time and money. The aim of the present research is to use daily high resolution multispectral (RGM and NIR) images from Planet Labs to determine whether such imagery can be used in real-time for monitoring to aid grazing management decisions for a wool producing enterprise. The study area located in Okehampton, approximately 10km north of Triabunna on Tasmania’s east coast, with Merino sheep rotationally grazing paddocks over almost 1000 ha. We used machine learning to estimate pasture biomass within each of the 32 paddocks. Destructive Field samples of pasture biomass were undertaken with samples collated and used to validate the model. This is a work in progress. If the validation is satisfactory we will assess the capability of Planet Labs imagery to be used to determine how often sheep should be moved from one paddock to another based on a minimum pasture biomass threshold.

Item Type:	Conference Extract
Keywords:	satellite imagery, pasture biomass, planet lab, machine learning, grazing management
Research Division:	Agricultural and Veterinary Sciences
Research Group:	Agriculture, Land and Farm Management
Research Field:	Agricultural Spatial Analysis and Modelling
Objective Division:	Animal Production and Animal Primary Products
Objective Group:	Livestock Raising
Objective Field:	Sheep - Wool
UTAS Author:	Ara, I (Dr Iffat Ara)
ID Code:	135028
Year Published:	2019
Deposited By:	TIA - Research Institute
Deposited On:	2019-09-23
Last Modified:	2019-10-04
Downloads:	0