The delivery of remotely assessed pasture growth rate and feed on offer information to farmers in Western Australia.
1 Graduate School of Management, University of Western Australia, Crawley, Western Australia.
2 CSIRO, Livestock Industries, Wembley, Western Australia.
Phone: G.E.Donald, 08 9333 6617
Fax: G.E.Donald, 08 9387 8991
One of the major difficulties facing graziers in the Mediterranean environments with a high winter rainfall is managing feed demand to meet animal requirements throughout the year. It is becoming crucial to increase productivity, meet specific market demands and maximise profit; to do this farmers must utilise their feed base far more efficiently and strategically. In collaboration with Agriculture Western Australia and Department of Land Administration (DOLA), CSIRO has developed spatial pasture biomass models utilising satellite imagery to assess pasture growth rate and feed on offer. Pasture growth rate is estimated using NOAA satellite imagery in combination with spatial climate data and feed-on–offer is estimated using LandSat TM imagery. Both these models can be applied at the regional, farm and paddock scale. This paper details the progression from the research concept to a proposed pasture biomass delivery mechanism for farmers. Response from farmer groups and mail out surveys clearly indicated the need and urgency for this pasture information and helped defined the method and frequency of delivery.
pasture growth rate, feed on offer, biomass, satellite remote sensing.
In temperate and mediterranean regions of Australia, utilisation of pastures by grazing animals is often as low as thirty percent. Feed budgeting is a critical strategy for improving feed utilisation and there are now a number of pasture evaluation and monitoring programs available to farmers across Australia to enable them to estimate pasture growth rate (PGR) and feed availability or feed on offer (FOO). Unfortunately, many farmers do not have the confidence nor the time available to make regular accurate and regular estimates across many, sometimes large, paddocks. It is also extremely difficult to measure the spatial variation in these characters.
The WoolPro and ProGRAZE programs of Agriculture Western Australia (AgWA) recognise that the routine and accurate measurement of FOO and PGR are critical components for improved pasture management and may enable farmers to:
- graze pastures to an optimal level of feed on offer
- optimise pasture growth rate, quality and utilization
- better manage the whole farm system such as pasture-crop rotations
- optimise carrying capacity
- improve liveweight gain, wool production and quality
- more efficient feed conservation (eg hay and silage)
- better match land and soil types to management and selection of pasture cultivars.
CSIRO Livestock Industries, in collaboration with Agriculture Western Australia (AgWA) and the Department of Land Administration (DOLA) has developed technology to measure PGR and FOO using satellite imagery (Hill et al., 1998, Edirisinghe et al, 2000). This technology is currently being tested and validated throughout southern Australia and it is envisaged that it will be available this year to farmers via the Internet. The concept is to provide farmers with real-time, accurate estimates of PGR and FOO across their entire holding(s) to enable easy interpretation of spatial variation.
FOO and PGR estimation models utilizes the normalised difference vegetation index (NDVI) drived from respective satellite imagery. NDVI is also referred to as a ‘greenness’ index. Using three WoolPro Focus farms AgWA provided emperical data to validate FOO and PGR located in the south west of Western Australia from 1995 to 1998. Ground truthing data was obtained over a diverse range of pasture, soil and management types.
Assessment of FOO required high resolution (25m2) spatial imagery from the Landsat Thematic Mapper (Landsat TM) sensors which were available at approximately 15 day intervals. A novel method to calculate FOO using this satellite imagery was developed (Edirisinghe et al., 2000). The basis of this methodology is the establishment of correlationships between the greenness index and field measured FOO data of a target farm. Figure 1 shows the spatial distribution of FOO patterns in the target farm. Mean FOO value is then calculated for each paddock in the farm using this FOO map (Figure 2).
Pasture Growth Rate (PGR)
Estimation of pasture growth rate required low-resolution (1.1km2) imagery from the Advanced Very High Resolution Radiometer (AVHRR) of the National Oceanographic and Atmospheric Administration (NOAA). PGR is an estimate of net production of pasture growth (Edirisinghe et al., 2000). PGR was predicted by combining greenness index data from AVHRR imagery with light use, moisture and temperature indices as well as climatic information. PGR (kg/ha/day) can then be mapped at a resolution of 1.1km2 at the regional scale as shown in Figure 3. At a farm or paddock scale, this regional information can then be incorporated with land management units (LMU) or soil type in order to approximate paddock values (Figure 4).
Figure 1. FOO index map from satellite imagery.
Figure 2. Paddock estimate of FOO kg/ha.
Figure 3. Western Australian regional PGR rate kg/ha/d.
Figure 4 Paddock estimates of PGR kg/ha/day.
A pilot study was initiated in 2000. AgWA staff measured actual PGR and FOO levels on 11 farms while CSIRO predicted corresponding information from satellite imagery. These farms were located within the Kojonup/Katanning and Williams area of south-west Western Australia (WA) the region where the PGR and FOO models were originally derived. The aim of the pilot study was to test the model on farms other than those used to develop the model and to test any required modifications. Major problems facing researchers before any product can be delivered include identifying the potential target market, precise product description, market potential and the most suitable delivery mechanisms. To address these issues a two-stage market study was completed (Sneddon et al., 2000). In Stage 1, interviews with farmer focus groups were conducted. The aim of these focus groups was to the potential target markets, requirements and preferences for the product, possible delivery mechanisms for this type of remotely sensed pasture management information, and the development of strategies for the future commercialisation of the technology. Further investigation into the commercial feasibility of this technology was carried out through a survey sent to 250 farmers in south-west WA with the aim to exploring and validating the focus group findings (Stage 2).
Results from the Market Research and Pilot Study
The main features of a delivery system for remotely sensed pasture information was that it should be reliable, simple, and user-friendly, cost effective and accurate. The information provided should be in a map or text-format. From the 250 mailed survey forms in Stage 2 in the south-west portion of WA, 101 responses were received. The average respondent was male (92%), aged between 40-49 (33%), with a sheep farm (46%) of 500-1500 ha (40%) in size.
Three potential delivery models were proposed: (a) a basic model where simple pasture information was delivered to the market in a relatively simple and raw form, (b) a technology transfer model that supports a user and agronomic interface, and (c) a whole farm model where remotely sensed pasture information would be delivered along with grazing management options, cropping information, agronomic advice, market and financial information and so on. The respondents preferred a system of delivery providing PGR at weekly intervals and FOO at monthly intervals, both at a paddock scale. They indicated that they would be prepared to pay for the pasture map to maintain cost effectiveness, and requested that some agronomic support be provided (technology transfer model) and preferably by a government agency (agribusiness) to ensure that the service maintained its reliability.
Other requirements and suggestions included that the satellite images should have no more than a three day delivery time from overpass, and that the system be thoroughly evaluated, continuously monitored, and all technical difficulties overcome prior to commercialisation. It was also desirable for PGR to be adjusted for pasture quality and botanical composition at some stage of its development.
Key issues raised by the market report concerned the potential impact of the role of the female farming partner. It became apparent from Stage 1 and 2 that the female farmers filled a predominantly bookkeeping and administrative, and had a high level of use and understanding of information and communications technology. For potential instruction on the use of this and other computing technology, serious attention should be given to the farm partnership focusing on both male and female participation.
Current strategies for farmer feedback
A new CSIRO/AgWEST/DOLA consortium has commenced a commercialisation project for PGR and FOO. Because it is critical that cloud free satellite data be provided as promptly as possible after overpass, DOLA will be providing real time imagery for a trial region in WA via a functional Internet based server system. ArcIMS (ESRI, Australia) was the suggested software to handle the complexities of functionality, input and outputs of vector (cadastral maps) and raster (imagery) data. In 2001 DOLA will provide PGR at a regional scale and FOO and PGR at a sub-paddock scale for selected focus farms. Researchers from the University of Western Australia will be monitoring the impact of this remote sensed technology on farm performance and providing input into product design and commercialisation strategies.
Currently the PGR model uses NOAA NDVI data with a resolution of 1.1 km2 and soil maps are used to interpolate the information to paddock level. Obviously some sensitivity is lost during this process, for this year LandSat TM information will be incorporated to assist with this interpolation. DOLA is currently making provision to obtain imagery from Moderate Resolution Imaging Spectroradiometer (MODIS) with a resolution of 250m2 (10ha). With imagery at the sub-paddock level interpolation using soil/LMU information should no longer be necessary. Further to this, there is a desire to test both models outside the region of development. Therefore farms have been targeted in the Geraldton, Esperance and Bunbury regions of Western Australia, and south South Australia, south-western Victoria and south-west of NSW to provide ground truthing of FOO and PGR to test the potential migration of these models to these areas.
With the continuous supply of weekly PGR and monthly FOO, a profile could be produced for farmers to assess paddock growth potential over different land management units and paddocks across a relatively long time frame. Operational use of satellite based images to determine pasture biomass, growth rates and quality information is becoming a reality. With the instability of agricultural commodity prices it is crucial to maintain sustainable and profitable productivity. The formulation of a proper and formal link between the acquisition of remote sensed data, climatic data and other cadastral information and the farms is of a high priority. The delivery model will need to migrate from a simple technology transfer model to a whole farm model as farmer’s confidence and skills improve.
We wish to acknowledge the contribution made by Dr. Brian Warren, Mr. Michael Hyder (AgWEST), Dr Richard Smith (DOLA) and Dr Michael Hill (Bureau of Rural Sciences, Canberra) for their vision and determination to initialise this research. Also, Dr. Rob Kelly of CSIRO and Dr. Tim Mazzarol of University of Western Australia for introducing Student Industrial Traineeships that provided a mechanism to define the target market and substantiate the need to commercialise FOO and PGR delivery.
Edirisinghe,A.,Hill,M.J. and Donald,G.E.,(2000) Estimating Feed-On-Offer and Pasture Growth Rate using remote sensing. Proceedings of the 10th Australasian Remote Sensing and Photogrammetry Conference, Adelaide, August, 2000.
Hill, M.J., Donald, G.E., Wheaton, G.A., Hyder, M. and Smith, R.C.G.,(1998). Remote sensing for precision pasture management in South Western Australia. Proceedings of the 9th Australasian Remote Sensing and Photogrammetry Conference, Sydney, July, 1998.
Sneddon, J.N., Mazzarol, T., and Soutar, G.N., (2000). Farm Management in the 21st Century: A Feasibility Study of the Commercail Delivery of Remotely Sensed Pasture Management Information (Stage 1 and Stage 2 report): 62. Perth, WA: CSIRO and Curtin Business School.