Alex B. McBratney and James A. Taylor
Australian Centre for Precision Agriculture, McMillan Bld A03, University
of Sydney, NSW, 2006
(Alex.McBratney@acss.usyd.edu.au
or j.taylor@agec.usyd.edu.au)
(c) copyright 1999 Australian Centre for Precision Agriculture
This paper outlines the theory and concepts of PA and how they relate
to the viniculture industry, particularly in terms of quality, environmental
and risk management. Some brief work on our experiences with Precision
Viniculture is presented. Areas where we believe future research should
be targeted are also discussed.
Precision Viniculture (PV) is a logical extension of Precision Agriculture
(PA) technology into the horticulture sector. But what exactly does the
term Precision Agriculture mean and imply? At the first workshop for PA
in viniculture in Australia it is perhaps fitting to take a step back and
evaluate the aims and misconceptions of Precision Agriculture before we
are swept away on a tide of technology and data sets.
What it is
In 1997 the U.S. Congress passed a Bill on Precision Agriculture which they defined as "an integrated and production based farming system that is designed to increase long term, site-specific and whole farm production efficiencies, productivity and profitability while minimizing unintended impacts on wildlife and the environment".
Simplified, PA is the use of new information technologies together with agronomic experience to site-specifically:
i) maximise production efficiency
ii) maximise quality
iii) minimise environmental impact
iv) minimise risk
Practically this is achieved by first recording environment parameters,
presenting the data in a form that is comprehensible, analysing these data
with data from other sources, e.g. market prices, in a Decision Support
System (DSS) and finally performing some differential management that can
be recorded the following year, restarting the cycle. This is made possible
by geo-referencing the data through the use of Global Positioning Systems.
This is the primary enabling technology of PA - the principle reason why
it has not been done before. The PA wheel is presented schematically in
Figure 1. It is important to realize that it is a wheel and without
one of the cogs it will not succeed.
Figure 1 – The PA Wheel
Also central to the PA philosophy are the concepts of Total Quality Management (TQM) and Vertical Integration (VI) in the agricultural sector. Traditionally farmers lost contact with their produce once it left the farm. Now with traceability of products, farmers are able to follow the movement of their produce into the market place. Nowadays a farmer is concerned not only with quality at the farm gate but also the quality at the point of sale and how his product meets consumer demands. This will bring premiums and also will probably be used for environmental auditing.
What it is not
There are several mistaken preconceptions about PA. The first is that PA is a cropping rather than an agricultural concept. This is due to cropping systems, in particular broad-acre cropping, being the face and driving force of PA technology. However PA concepts are applicable to all agricultural sectors from animals to fisheries to forestry. In fact it might be argued that PA concepts are more advanced in the dairy industry where the "site" becomes an individual animal which is recorded, traced and fed individually to optimize production. These industries are just as concerned with improved productivity and quality decreased environmental impact and better risk management as the cropping industry however PA concepts have yet to be applied on the same scale in these areas. For example a graziers use of advance warning meteorological data and market predictions to estimate fodder reserves and plan livestock numbers is a form of PA.
The second misconception is that PA in cropping equals yield mapping. Yield mapping is a crucial step and the wealth of information farmers are able to obtain from a yield map makes them very valuable. However they are only a stepping stone in a PA management system. The bigger agronomic hurdle lies in retrieving the information in the yield map and using it to improve the production system. The advance of PA adoption (usefulness) in this country is may soon be bottlenecked at this point due to the lack of decision support systems (DSS) to help agronomists and farmers understand their yield maps. Yield maps may not tell the whole story either with other data sources, e.g. crop quality and soil maps, economic indicators or weather predictions, proving further information necessary for correct agronomic interpretation.
The final misconception is that PA equals sustainable agriculture. PA
is a tool to help make agriculture more sustainable however it is not the
total answer. PA aims at maximum production efficiency with minimum environmental
impact. Currently it is the potential for improved productivity (and profitability)
that is driving PA rather than the more serious issue of long term sustainability.
PA will not fix problems such as erosion and salinity by itself although
it will help to reduce the risk of these problems occurring. Sensible sustainable
practices still need to be used in conjunction with PA.
PA, and of course PV, is dependent on the existence of variability
in either or both product quantity and quality. If this variability
does not exist then a uniform management system is both the cheapest and
most effective management strategy and PA is redundant. Thus, in PA, "Variability
of production and quality equals Opportunity". Having said this the nature
of the variation is also important in determining the potential for PA
in a system. For example the magnitude of the variability may be too small
to be economically feasible to manage. Alternatively the variability may
be highly randomized across the production system making it impossible
to manage with current technology. Finally the variability may be due to
a constraint that is not manageable e.g. localized storms in large wheat
paddocks. Thus the implementation of PA is limited by the ability of current
variable rate technology (VRT - machinery/technology that allows for differential
management of a production system) to cope with highly variable sites and
the economic inability to produce returns from sites with low variability
using PA (VRT).
Figure 2 - The PA time line (adapted from Viscarra Rossel and McBratney, 1998)
Due to these constraints PA is at present operating on a zonal rather
than a completely site-specific basis. As VRT improves and the capital
cost of entering PA decreases, the minimum size of management zone needed
to effectively implement PA will decrease till eventually a truly site-specific
management regime is possible. Until this occurs there is a need to be
able to quantify both the variability of a production system and the size
of the minimum manageable zone (MMZ). If the variability in the production
system dictates management zones smaller than the MMZ than PA is not relevant
to the system at the present time (but may be in the future). It will be
interesting to see how the concept of the management zone develops and
to see how it compares with the concept of terroir.
In Australia several aspects of the winegrape industry lend themselves
to the adoption of PA technology. Viniculture is intensive, highly mechanized,
has high value adding potential and is dominated by large companies. Thus
the incentive, ability and capital is available. Viniculture is one of
the first horticultural crops in Australia to which PA methodology has
been applied. While many of the lessons learnt from broad-acre cropping
can be utilised, PV also offers new challenges.
Viniculture, and horticulture in general, has fixed perennial plants.
Thus
there is a long-term scale involved compared to the annual
nature of cropping. Plants are cloned eliminating within varietal differences.
This puts the emphasis on variability on the site
specific clone-environment-management interaction. The system is more intensively
managed allowing for more detailed ground truthing and data collection.
Management decisions are also capable of having a much larger impact on
yield in viniculture e.g. pruning strategies can affect yield by upwards
of 100%. The majority of Australian vineyards are irrigated, minimising
the impact of the biggest variable in crop production in Australia, and
giving further control to the grower in yield and quality production.
As discussed previously PV is only applicable to production systems
if variability is inherent in the system and while yield maps make pretty
pictures there is no simple quantitative measure of the variation present.
Fairfield Smith (1938) first proposed an empirical law for quantifying
yield variation that looked at the heterogeneity of the field. Recently
geo-statistics and in particular the variogram (McBratney and Pringle,
1999) have been used to describe variation of soil properties. Following
on from this McBratney et al (pers comm.) developed a method of
adapting Fairfield Smith’s work to PA and yield variograms to describe
variation. Variograms have proven very effective in describing spatial
variation as they model the semi-variance of the data with respect to distance.
An alternative method of estimating areal variability is the Opportunity
Index.
The Opportunity Index (O I) contains three terms. The first evaluates the area over which variation occurs, the second evaluates the magnitude of variation and the third term describes the economics of precision management. The OI may be interpreted according to Table 1.
| Opportunity Index (OI) | Potential for PA |
| <1 | Little to none |
| 1-2 | Small |
| 2-3 | Medium |
| >3 | Large |
Table 1 – The relationship between Opportunity Index and P.A. potential
(For further information on the derivation of the Opportunity Index please contact the authors)
It should be noted that the MMZ for a vineyard is considerably smaller than that required for broad-acre cropping. Viniculture tends to employ narrower applicators and travel at speeds slower than that in broad-acre situations. This means that PV has the ability to manage areas of high short-term variability that broad-acre PA cannot. For the calculations in this work the values in Table 2 have been assumed.
| Parameter | Broadacre | Viniculture |
| b (m) | 20 | 6 |
| n (ms-1) | 6 | 3 |
| t (s) | 3 | 3 |
Table 2 – Parameter values used for determination of MMZ
For this study the yield variograms and integral scales of the correlograms of various crops will be compared to winegrape yield data from the 1999 vintage at Richmond Grove Vineyard Cowra. The winegrape data was collected using a Harvestmaster Profile Grape Yield Monitor attached to a Gregoire G65 Grape Harvester. Data was collected for three varieties, Chardonnay, Cabernat Franc and Semillon. The semi-variance of yield is modelled using a double exponential function (McBratney et al pers. comm.). The variograms for winegrapes are shown in Figure 3 and parameters for all crops shown in Table 3. Yield maps are shown in Figure 4.
| Field | Crop | Location | Year | Mean yield
(m ) |
CVs (%) | CVv (%) | Variogram Parameters | Ja (ha) | MMZ
(ha) |
O I | ||||
| C0 | C1 | C2 | a1 (m) | a2 (m) | ||||||||||
| D3-4 | Cab. Franc | Cowra* | 1999 | 20.4 | 53 | 49 | 18 | 13 | 8611 | 6 | 100000# | 9216.0 | 0.005 | 6.6 |
| Horse | Wheat | Moree* | 1995 | 2.7 | 46 | 44 | 0.12 | 0.07 | 137.44 | 13 | 100000# | 9345.1 | 0.036 | 5.8 |
| Home 3 | Wheat | Wyalkatchem* | 1998 | 1.5 | 39 | 38 | 0.03 | 0.08 | 23.72 | 40 | 100000# | 9312.3 | 0.036 | 5.7 |
| w80 | Sorghum | Moree | 1996 | 4.2 | 32 | 31 | 0.18 | 0.14 | 154.7 | 8 | 100000# | 9216.0 | 0.036 | 5.6 |
| Blackies 6 | Lupins | Wyalkatchem | 1998 | 1.1 | 43 | 41 | 0.02 | 0.04 | 0.42 | 36 | 2061 | 709.4 | 0.036 | 4.6 |
| N3 | Wheat | Moree | 1995 | 2.2 | 59 | 53 | 0.32 | 0.47 | 1.26 | 392 | 784 | 82.8 | 0.036 | 3.8 |
| West Creek | Wheat | Moree | 1998 | 5.6 | 19 | 15 | 0.4 | 0.25 | 0.55 | 9 | 675 | 41.8 | 0.036 | 2.9 |
| B1-B2 | Wheat | Moree | 1995 | 1.4 | 66 | 63 | 0.08 | 0.15 | 0.62 | 155 | 200 | 6.7 | 0.036 | 2.8 |
| North | Chardonnay | Cowra | 1999 | 20.1 | 35 | 22 | 31 | 9 | 10.54 | 35 | 212 | 1.9 | 0.005 | 2.6 |
| B4 | Wheat | Moree | 1995 | 1.9 | 45 | 40 | 0.16 | 0.25 | 0.33 | 23 | 275 | 6.8 | 0.036 | 2.6 |
| West Creek | Wheat | Moree | 1997 | 3.7 | 29 | 25 | 0.29 | 0.53 | 0.35 | 210 | 213 | 7.0 | 0.036 | 2.4 |
| East Creek | Sorghum | Moree | 1996 | 7 | 15 | 12 | 0.49 | 0.23 | 0.44 | 18 | 134 | 1.4 | 0.036 | 1.4 |
| Rowlands 1 | Wheat | Wyalkatchem | 1995 | 1.5 | 33 | 31 | 0.03 | 0.09 | 0.13 | 50 | 50 | 0.4 | 0.036 | 1.3 |
| West Creek | Wheat | Moree | 1996 | 5.4 | 12 | 10 | 0.11 | 0.17 | 0.12 | 15 | 145 | 1.3 | 0.036 | 1.3 |
| Home 2 | Barley | Wyalkatchem | 1997 | 1.5 | 30 | 26 | 0.05 | 0.1 | 0.05 | 12 | 94 | 0.5 | 0.036 | 1.3 |
| Norwood 10 | Cotton | Moree | 1998 | 7.8 | 21 | 16 | 1.1 | 0.46 | 1.1 | 7 | 100 | 0.8 | 0.036 | 1.3 |
| Rowlands 4 | Lupins | Wyalkatchem | 1996 | 0.9 | 19 | 16 | 0.01 | 0.01 | 0.01 | 69 | 72 | 0.8 | 0.036 | 1.2 |
| Home 8 | Lupins | Wyalkatchem | 1997 | 0.5 | 35 | 28 | 0.01 | 0.01 | 0.01 | 39 | 35 | 0.2 | 0.036 | 0.9 |
| Oakville | Cotton | Narrabri* | 1999 | 6.4 | 16 | 14 | 0.36 | 0.31 | 0.44 | 10 | 218 | 0.4 | 0.036 | 0.9 |
| Telleraga 28 | Cotton | Moree | 1998 | 10.4 | 23 | 16 | 3.1 | 1 | 1.7 | 44 | 60 | 0.3 | 0.036 | 0.8 |
| C1-2 | Semillon | Cowra | 1999 | 23.9 | 21 | 8 | 21.5 | 3.22 | - | 7 | - | 0.0 | 0.005 | 0 |
* Wyalkatchem is located in the W.A. wheat belt. Moree and Narrabri are located in NW NSW. Cowra is located 300km west of Sydney.
# These data showed a large linear trend over the distance of the variogram fit (320m) resulting in large C2 and a2 values.
CVs is the coefficient of variation
standardised over a distance of 1000m 
Table 3 – Summary of yield variation measures for various crops. All
data collected using yield monitors and GPS. The rows in boldface are from
our winegrape studies at Richmond Grove. The data from W.A. are courtesy
of Dr Simon Cook, CSIRO. The Moree and Narrabri data from the Australian
Centre for Precision Agriculture
Figure 4 – Yield maps of the three winegrape varieties used in this study
While the variogram tells us that there is a spatial dependence in yield it does not tell us the area over which yield varies. The areal scale indicates the area over which there is a correlation between yields. In the case of the Cab. Franc data the Ja value may not be entirely accurate due to a large linear trend over the distance of the modelled variogram (320m). This resulted in the fitting of large C2 and a2 values (also observed in the Horse, Home 3 and w80 data). To fit the variogram properly it may be necessary to model it over a larger distance. Both Chardonnay and Cab Franc have large Ja values indicating PV potential whilst Semillon has a very small Ja indicating little or no PA potential. When O I is calculated the Cab. Franc. block apparently has a large potential, Chardonnay a medium potential and Semillon no potential.
This preliminary study indicates that PV is applicable to parts of the
vineyard but not to all areas. In the case of the Semillon it highlights
a valuable lesson, if variability does not exists then PV is not necessary
and traditional uniform management is preferable. Having said this the
data should be interpreted loosely due to the lack of temporal data. For
an accurate calculation of O I the
analysis should be performed on data derived from several years’ harvesting.
The main objectives of PA have been listed previously, and before
we proceed any further it is pertinent to relate these to PV.
Maximising Yield and Quality:
In Viticulture quality is perhaps a more important parameter than yield in determining the value of the crop. There is generally considered to be a trade off between yield and quality in viniculture (and other crops). As noted above a viticulturist is able to exercise considerable control over the yield and quality of the crop. Heavy pruning and applying water stress can decrease yield but increase quality. However studies (Sinton et al) have shown that this trade off it not always necessary and both good quality and good yield can be achieved simultaneously. It is this scenario that is the objective for PV management of existing vineyards.
The first and potentially the biggest step in managing yield and quality and understanding the vine-environment interaction is the initial placement of vines. The long-term nature of a vineyard results in this becoming crucial for future management decisions. If vines can initially be planted in zones of similar environment or "terroir" it may reduce the need to differentially manage them later i.e. by differentially planting we can uniformly manage. This is much more economical than the reverse of uniformly planting and differentially managing. Unfortunately the latter is the more common situation facing existing growers entering into PV today. The benefit of planting varieties to soil type has already been recognised by the industry with soil surveys standard with new plantings. These surveys (usually on 75m grids) may not be detailed enough to provide the accuracy required for PA. Our work at Cowra shows a large proportion of variation at scales finer than this. The use of remotely and proximally sensed data may provide better information for more precise plantings and irrigation layout in the future.
Over the past few decades there has been an increase in consumer awareness of quality and government legislation on quality assurance. This has forced farmers to produce within defined accreditation standards and at a consistent quality. To help regulate this on a global scale the International Organisation for Standardization (ISO) has developed a set of quality management standards (ISO 9000) and environmental management standards (ISO 14000 -discussed below) for a wide variety of industries. (These are not product standards but management standards and are often incorporated into national standards). ISO 9000 has been developed to meet customer quality requirements thus an accredited company is tailoring the quality of their product/service to the customer and gaining an advantage over their competition. In terms of quality product standards many wineries are now using HACCP (Hazard Analysis Critical Control Points). Under new legislation this will become mandatory for all food business, including wineries, in 2000. Currrently HCCAP is not applied to a vineyard situation (Small, 1999).
The quality issue is especially pertinent in the viniculture industry where inconsistency in grape quality will degrade wines even if average quality is good. Vignerons are also able to produce higher quality wines if the higher quality grapes can be segregated (Johnson et al. 1997). With variable grape yield and quality the norm in most production systems and harvesting done uniformly, viticulture is currently not taking advantage of the variation in quality across a vineyard. If, through remote or proximal sensing, quality can be mapped just prior to or at harvest the opportunity is there to segregate grapes to produce better and more profitable wines.
Minimising Environment Impact
Vineyards have two main environmental impact concerns, irrigation and the use of chemical fungicides. Irrigation and salinity is currently one of the biggest concerns in Australian agriculture and as a major user of irrigation water the viticultural industries need to be aware of the potential dangers of over irrigation. A general movement to drip rather than broadcast sprays will help but there is a need to continuously monitor water table levels and adjust management accordingly. There may also be an opportunity for vineyards to employ differential watering regime to further maximise the irrigation efficiency and minimise loss to ground water.
Vineyards are big applicators of chemicals, using upwards of 10 sprays a season to combat fungal and insect pressure on the grapes. A better understanding of the areas most prone to outbreak may allow for a differential application of chemical that is more cost effective and less environmentally damaging.
As mentioned above ISO 14000 standards have been developed for environmental management however the adoption and adaptation of these standards to agriculture is very limited. (For example Denmark has some 50 accredited farms (Langkilde 1999) while there is only one accredited cotton farm worldwide, which is in Australia). The ISO 14000 was developed in response to a need for sustainable development thus PA should be an integral part the guidelines. By adopting these standards produce can be targeted to the environmentally conscious and sold at a premium like free range eggs and dolphin free tuna currently are.
Minimising Risk
Risk management is a common practice today for most farmers. Economically many farmers hedge on the stock market to ensure a minimum price for their product. Others insure to avoid acts of God. With improved communication and information transfer, farmers in the future will hopefully have more data and a better chance of optimizing the use of these economic risk management options. Physically farmers practice risk management by erring on the side of extra inputs. Thus a farmer may put an extra spray on, add extra fertilizer, buy more machinery or hire extra labour to ensure that the produce is harvested/sold on time thereby guaranteeing a return. This is contrary to the concept of PA. PA needs to provide a better management system, to aid in risk management, to substitute for these extra physical inputs (Harris, 1997). This better management strategy will come about through a better understanding of the environment-crop interaction and a more detailed use of emerging and existing information technologies, such as overseas crop reports, short and long term weather predictions and agroeconomic modeling.
Incorporated also into the concept of risk management in PV are TQM
and VI. Obviously the quality of grapes is vital in determining the quality
of the wine as is expressed in the adage "wine is made in the vineyard".
The concept of TQM is schematically described by the Deming Wheel (Figure
5). A TQM approach (using ISO 9000 guidelines) aims to increase quality
by firstly decreasing the variability of the system then secondly improving
the system (Bishop 1998). By decreasing the variability of the system the
risk of the system failing is reduced.
Figure 5 – The Deming Wheel of TQM (adapted from Hutchins, 1992) illustrating
the similarities with the PA wheel
Viniculture is a very vertically integrated industry in Australian agriculture
with many companies running both commercial vineyards and wineries. This
provides the opportunity for the company to value add to their product
prior to sale. To do this efficiently there must be constant communication
between vineyard and winery to ensure the correct product is produced.
For example wineries will be able to look at overseas or local vintages,
weather and market predictions and determine (or predict) what sort of
wines are (or will be) limited. With an understanding of the plant-environment
interaction, management practices in the vineyard can be applied to produce
these wines allowing wineries to target particular markets with the right
product. It is a challenge for PV that efficiency of production is maximised
in the winery as well as the vineyard.
PV in Australia has barely learnt to crawl yet alone walk yet. At
such an early stage it is important that the concept of PA is not misunderstood
as it has been in other industries. PV is not a case of whacking a yield
monitor onto a harvester and taking off at 100 miles an hour. To make PV
work all areas of the PA wheel (Figure 1) need to be addressed. Currently
most of the research is directed at data acquisition, environmental monitoring
and attribute mapping to quantify variability in the system and identify
MMZ's to determine if PV is applicable. If viniculture is not to fall into
the same trap as the grains industry there is a need to formulate a PV
approach that encompasses all aspects of the PA wheel.
Geo-referencing
Differential Global Positioning Systems are now common place on many farms and the technology is adequate for use in viniculture. One aspect that does need further refinement is the accuracy of DGPS in the z (or elevation) plane. While x, y data (latitude and longitude) is accurate to <1m, elevation is accurate to ± 3m. The use of Digital Elevation Models (DEM) in farm situations is increasing and therefore so is the value of this z information. The commercial potential for this information will result in this improvement coming mainly from private industries.
Crop, Soil and Climate Monitoring
Many sensors and monitors already exist for in situ recording. The challenge for PA and PV is too make these real-time on-the-go sensors. While the commercial potential of these sensors will mean that basic R&D will be done by private industry, research bodies have an important role to play in the development of the science behind the sensors. Market concerns will lead private industry to sell sensors prematurely to ensure market share. This may lead to substandard sensors and a failure to adequately realize the potential of the sensor. Scientists also need to determine what and how multiple indicators can be measured. For example a NIR baume sensor is currently being developed for commercial release. However NIR may also be used to measure other important must characteristics e.g. terpenes, or further characterise sugar content into sugar types. It is also important to utilize other sensors, e.g. ion-selective field effect transistors, to simultaneously measure other must characteristics, e.g. pH and K. The use of multiple sensors also creates new problems in the area of data fusion and decision making, an area which has had little research done on it.
Attribute Mapping
For several decades geostatisticians and pedometricians have been researching ways of describing and representing spatial data that accurately interprets the raw data. Most of this has been done with point data and low data densities. While PA and PV can utilize this previous work it offers new problems. Yield data is often convoluted in the harvester and needs to be post-processed (deconvoluted) before it can be used. PA also produces large dense data sets that are producing new challenges for interpretation and mapping. One of the largest problems is the determination of initial and future sampling schemes to ensure that the variability of the system is properly characterised. These challenges have seen many geo-statisticians and pedometricians move into the area of PA. PV can benefit from the work already done however differences in the production system between vinicultural and broad-acre crops means some research will be needed to adapt and expand these methods.
The other challenge is to bring together data from different sources and present it on a common platform. The development of Geographical Information Systems (GIS) is allowing this to occur however the adaptation of this technology to farm scales is still in its infancy.
Decision Support Systems
Techniques for data presentation and storage, e.g. GIS, developed in other industries are also applicable with some modification to viniculture. However DSS are not so flexible and it is in this area that real research needs to be done. The majority of engineering companies currently supplying PA technology are not interested in and are unable to produce DSS. Thus the onus will fall on the industry and to a lesser extent the government to fill the gap. Initially it may be sufficient to adapt an existing DSS such as AUSVIT to site-specific situations. In the long run a viniculture DSS that is able to site-specifically model vine-environment interactions in terms of yield and quality will be needed. This will need to be flexible enough to incorporate all aspects of the new information technologies, accept feedback from other parts of the PA cycle and be able to conform to standards such as ISO 9000/14000.
Differential Action
The production of VRT is essentially an engineering problem. Due to the commercial potential of VRT much of this engineering development will again be driven by the private sector. The main input from an agronomic point of view is the provision of accurate information on application rates (derived in the DSS) and interpretation of the results of the differential action for feedback into the DSS.
Vertical Integration and Total Quality Management
The other great challenge for PV that is unique in Australian agriculture is the successful implementation of a vertically integrated PV system. For this to succeed the PA wheel needs to be effective at the vineyard level and then brought into the winery. Existing industry standards and guidelines eg ISO 9000/1400, HACCP and Australian Standards need to updated and combined in the context of PA, particularly in areas of quality and environmental management and assurance.
It is inevitable that PV will become the dominant production system
for winegrape production in Australia. The real question is how long will
it be before this situation is reached. Before we can run we must be able
to walk and before we can walk we must be able to crawl. By correctly identifying
and targeting the major obstacles to PV implementation we shall facilitate
its adoption in Australia.
Acknowledgments
We would like to thank Orlando-Wyndham and the staff at Richmond Grove,
Cowra for the research site and their co-operation. Thanks also to Ron
Campbell and HarvestMaster US. Pty. Ltd. for the supply of a yield monitor.
Finally thanks to Dr Simon Cooke (CSIRO Land & Water WA) for letting
us use some of his yield data.
References
Bishop, T. F. A. (1998) Quality and Precision Agriculture. Leading Edge Vol. I No. I (Berekua Pty. Ltd.) pp 33-34
Harris, D. (1997) Risk Management in Precision Farming. Precision Agriculture '97 -Volume II: Technology, IT and Mangement. Proceedings of the 1st European Conference on Precision Agriculture, Warwick University Conference Centre, UK, 7-10 September. (BIOS Scientific Publishers) pp 949-956
Hutchins, D. C. (1992). Achieving Total Quality. Cambridge, England
Johnson, L., Lobitz, B., Wiechers, S., Willimas, D. and Skinner, P. (1998) Of Pixels and Palates: Can geospatial technologies help produce better wine? Proceedings, 1st Int'l Conference. Geospatial Information in Agriculture and Forestry. Lake Buena Vista FL, 1-3 June 1998 (http://geo.arc.nasa.gov/sge/crush/ERIM.html)
Langkilde, N. (1999) Practical experiences in precision agriculture. Proceedings of the 2nd European Conference on Precision Agriculture. Odense Congress Centre, Denmark, 11-15 July. (Sheffield Academic Press) pp 35-46
McBratney, A.B. and Pringle, M.J. (1999) Estimating average and proportional variograms of soil properties and their potential use in Precision Agriculture. Precison Agriculture 1(2), pp 125-152
McBratney, A.B., Whelan, B. and Pringle, M.J. A development on Fairfield Smith’s law of yield heterogeneity for use in Precision Ageiculture. (in preparation).
Russo D. and Bresler E. (1981) Effect of field variability in soil hydraulic properties on solutions of unsaturated water and salt flows. Soil Science Society of America Journal 45(4) pp 675-681
Sinton, T.H., Ough, C.S., Kissler, J.J. and Ksimatis, A.N. (1978) Grape Juice Indicators for prediction ofpotential wine quality: I. Relationship between crop level, juice and wine composition, and wine sensory ratings and scores. Am. J. Enol. Vitic., 29(40) pp 267-271
Small, G. (1999) HACCP: What is it and how can it help me in my vineyard? The Australian Grapegrower and Winemaker, 430, pp 13-15
US House of Representatives, 1997. Bill No. H.R.2534. An Act to reform, extend and repeal certain agricultural research, extension and education programs and for other purposes. Title IV - New research, extension and education initiatives. Subtitle B -Precision Agriculture (http://thomas.loc.gov/cgi-bin/query/z?c105:H.R.2534:)
Viscarra Rossel R. and McBratney A.B. (1998) Soil chemical analytical
accuracy and costs: implications from precision agriculture.
Australian.
Journal of Experimental Agriculture, 38, pp 765-75
THE UNIVERSITY OF SYDNEY
© 2003 - Australian Centre for Precision Agriculture