Case Studies
Applications

_____________

Feed-on-Offer at the Paddock Level

The technology provides estimates of feed-on-offer (FOO) on individual properties at the paddock level for southwest Western Australia, measured using remote sensing methodology. We believe our technology is unique because we quantitatively measure FOO in kg DM/ha, rather than just providing a relative index which is of little value for on-farm decision making.

The technology to estimate FOO (kg DM/ha) has been developed and calibrated since 1995 in the southwest of Western Australia and is based on the relationship between normalized difference vegetation index (NDVI) or "index of greenness" and ground data to explain seasonal response patterns. 

Changes in NDVI are strongly correlated to the variation in leaf area index (LAI) of green herbage during the season, and therefore accumulation of FOO in the field. NDVI is derived from the red and near infrared (NIR) reflected solar radiation detected by relevant spatial bands of satellite sensors. The differential reflectance in these bands provides a means of monitoring density and vigour of green vegetation growth.  

We routinely use the Landsat TM or SPOT satellites to derive NDVI data. These satellites enable us to measure FOO at 20-30m spatial resolution which means we get 16 individual FOO values per hectare of pasture.

We will also be assessing the use of the information by the co-operators in the Precision Wool project.  If you are not a current co-operator in this project you can still view feed-on-offer data for the Badgingarra, and Avondale research stations in Western Australia by logging in as guest and guest2 respectively.  In both cases the password is the same as the login.

 

How good were we? - ground validation of satellite predictions in 2002.

From Mata et al 2003.

Figure 1. Selected zones in the south west of Western Australia where FOO estimations were carried out together with ground validation between May and October 2002. Figure 2. Relationship between FOO measured from the satellite and observed values from ground validation by measuring specified transects within collaborating properties.

Validation Method

The accuracy of remote sensed estimates of FOO was determined by comparing the average values estimated from satellite observations with those derived from visual estimate of standing pasture. During this project a total of 68 producers participated in the estimation of FOO, representing 9 zonal areas with distinct growing characteristics in the south west of Western Australia (Figure 1).  

In this report preliminary data analysis from 12 properties within Zones 1 and 4 are presented to validate the accuracy of the predictions.  Within each zone, field values were collected from 6 properties, 3 of which were classified as calibration farms 1 3 and were used for the calibration of the model.  The remaining 3 properties were classified as validation farms and values for those properties were used to test the application of the model.

Ground sampling was programmed every 34 days to coincide with the overpass of the Landsat TM satellite.  When Landsat TM images were not available due to cloud cover, acquisition of SPOT XS images was pre-programmed to provide an image within 5 days of the scheduled TM overpass. The field estimates of FOO were collected from May to September 2002 by officers of the Western Australia Department of Agriculture who conducted monthly calibration tests to ensure accuracy of field estimates (Thompson et. al. 1994). Two transects each 500m long and 30m apart were geo-referenced and FOO estimates obtained at 10m intervals. 

Accuracy

The accuracy of the FOO model was estimated at 10% for paddock estimates within the range of 400 and 5000 kg DM/ha (Edirisinghe et. al. 2000).  The accuracy deteriorates above 5000 kg DM/ha due to NDVI saturation (Hill et. al. 1998), however under normal practice levels above 5000 kg DM/ha are only likely to be found in paddocks closed for fodder conservation. 

Figure 2 shows the agreement between the observed field FOO (visual estimates along the transect) and the satellite estimated FOO for calibration and validation farms in zones 1 and 4. 

Practical implications

This range of FOO from 400 to 5000 Kg DM/ha encompasses the important changes in feed availability that needs to be managed in a sustainable system.  From an animal production perspective, the range also covers the levels of feed availability where changes in production responses are observed due to changes in efficiency of use.  For wethers, liveweight gain may be limited when FOO is between 1000 and 2000 kg DM/ha depending on age and condition score, above that level, grazing selection may take place potentially influencing botanical composition and paddock heterogeneity (Thompson et. al. 1994). 

 Managing pasture resources by maintaining grazing pressure within this range of FOO gives producers the opportunity to exploit the strong relationship between FOO and wool growth, fibre diameter and liveweight gain (Thompson et. al. 1994).  Manipulating stocking rate to maintain a desired level of FOO will lead to increased wool production per hectare while decreasing fibre diameter, the two main profit drivers for wool growers (Mata et. al. 2002 a &b). 

 

 

.

Copyright 2003 Pastures from Space
Last modified: October 01, 2003

Disclaimer of Liability