Pastures from space - prediction of feed-on-offer in intensively-grazed dairy pastures

The accurate assessment of available pasture is critical in the dairy industry to permit producers to allocate feed on a per head basis as part of the intensive management of grazing rotations. Fulkerson et al. (2005) showed that precise daily allocation of feed to milking cows grazing perennial pastures could lead to about a 10% increase in milk production and deliver significant financial benefits to producers. However, very few producers invest the time or resources necessary to monitor pastures on a regular basis in a way that would allow them to accurately budget feed resources. Edirisinghe et al. (2000) described the Feed-On-Offer (FOO) model for predicting pasture biomass of extensively-grazed annual pastures from satellite-derived Normalised Difference Vegetation Index (NDVI) data. Edirisinghe et al. (2004) reviewed the accuracy of the predictions from the FOO model and concluded that paddocks-scale predictions of FOO for annual pastures in Western Australia were of sufficient accuracy to meet farmers’ requirement s for general feed budgeting decision-making. This paper presents preliminary results of analyses of the accuracy of FOO predicted for intensively-grazed pastures. Validation of FOO predictions was carried out at 3 sites between July and September 2005; Vasse RS, W.A. (annual ryegrass), Ellinbank RS, Victoria (perennial ryegrass) and Camden, N.S.W. (perennial ryegrass). Satellite NDVI data were obtained from 2 high-resolution sensors; Ikonos (4m pixels) and SPOT-5 (10m pixels). NDVI is calculated as ((red band – NIR band) / (red band + NIR band)), where ‘red band’ and ‘NIR band’ are the reflectance values of the red and near-infrared (NIR) spectral bands of the satellite imagery. Ground data were collected using a rising plate meter (RPM) and pasture cuts to ground level, and the data were analysed in a GIS environment.