Precision Ag News Volume 3 Issue 1 Spring/Summer 2006 pages 1-24 © SPAA 2006 ISSN 1449-3705

THE MAGAZINE OF THE SOUTHERN PRECISION ASSOCIATION

Features SPAA research update Broadacre experience with sensors Selective grape harvesting Virtual fencing

The Southern Precision Agricultural Association (SPAA), exists to promote the development and adoption of precision agriculture technologies in the: grains - wine grapes - horticulture - other Ag industries Learning from shared experience “I believe PA is the way forward. Joining SPAA is helping me to make sense of the yield data and to use PA resources to manage highly variable soil types,” Randall Wilksch, Yeelana SA.

SPAA supporting growers, advisors and researchers in the development and adoption of PA technologies. Annual membership FULL $55, Associate (for those who trade under the same business name as a full member) $27.50 To join SPAA please contact Pam Pilkington [email protected] or visit the website www.spaa.com.au

 Precision Ag News Contents

Looking back and forward . . . . . 4 Research program expands . . . . 6 In- sensing ...... 7 Guided by the crop ...... 8 Spot spraying in broadacre cropping systems ...... 11 Virtual grazing systems ...... 13 Mapping actual fertiliser distribution ...... 15 Data layers for soilborne disease ...... 16 Practical applications of precision viticulture ...... 18

– SPAA Sponsors –

Precision Ag News is published by Southern Precision Agricultural Association SPAA DISCLAIMER SPAA, SPAA have prepared this publication, on the basis of information available at the time of publication without any independent PO Box 31, CRYSTAL BROOK SA 5523 verification. Neither the Corporation and its editors nor any Ph 08 86362649, Fax 08 86362960 contributor to this publication represent that the contents of this publication are accurate or complete; nor do we accept any [email protected] omissions in the contents, however they may arise. Readers who www.spaa.com.au act on the information in this publication do so at their risk. The contributors may identify particular types of products. We do Precision Ag News is written and produced for SPAA by not endorse or recommend the products of any manufacturer referred to. Other products may perform as well or better than AgriKnowHow with design by Lightening Designs. those specifically referred to. Newsletter and Magazine copy deadlines 2006/07 Publication Copy deadline Advertising bookings Distribution Precision Ag News 12th February 5th February March 28th September 20th September October SPAA Newsletter 11th December No advertising December 22nd January February 30th April May 1st July July

Precision Ag News  From the President

Looking back and forward Malcolm Sargent, SPAA President

elcome to the first issue monitor in disgust; let’s salvage The key goals for the organisation of the new style Precision something useful. are to: WAg News. I hope you will It has been a busy year for SPAA spot that it is not just the design that increase grower adoption of and with an estimated 70% of has changed. precision agriculture across Australia; producers using some form of For those of you who did not make guidance the opportunities for become a significant source of it to the AGM, no you are not the future are expanding. precision agriculture information experiencing ‘déjà vu’, I am back in in Australia; In August the SPAA committee held the chairman’s seat. a strategic planning day to review maintain and develop a stable and Although 2006 is a harvest most the future direction for SPAA. professional organisation. Some proposals in the plan are grain growers would rather forget, From a technical perspective, SPAA listed below. The committee is I do not think we should walk will address three key areas: away from it too quickly as there is meeting regularly to establish and probably much to be learnt. With implement a plan of action. - provide an independent source very little rainfall since the break of of advice on new concepts and SPAA will become the driving force the season, have survived on equipment; for precision agriculture in Australia. stored water. This year’s yield maps will be a strong indication of areas SPAA will be a broad based - lobby equipment manufacturers with better plant available water, organisation that covers a range of for greater compatibility between so don’t switch off the yield primary industries. systems and components;

 Precision Ag News - facilitate research in innovative a big role to fill. The position of day, a PA training day at Birchip areas such as variable rate Executive Officer will be advertised and an owner training day for technologies. in the near future (for more details those using variable rate with please contact the SPAA office). One Kee Technology equipment. None of this will be achieved without of the tasks of the new Executive SPAA also supported the 10th a capable and dedicated committee Officer will be to expand the work of Annual Symposium on Precision and executive team. SPAA into other industry sectors for Agricultural Research and Many thanks to Brian Tiller for his example potato and cane industries. Application in Australasia. great contribution as SPAA chairman SPAA is currently involved with four over the past two years. Brian and this year’s yield maps research trials (See page 6) and I have agreed that we don’t intend will be a strong has made submissions to National to rotate as chair and are delighted indication of areas Landcare Program and GRDC for to welcome some new faces to this with better plant two projects that relate to the year’s committee. available water implementation of variable rate A complete list of the committee, technologies in broadacre cropping. their positions and contacts is found Thanks must go to our Treasure Peter Many thanks to Allan Mayfield for all below. Many thanks to retiring Cousins and Administration Officer his work on these submissions. committee members: Allen Buckley, Pam Pilkington for taking on many As a membership based organisation Brendon Frischke, SARDI and of the tasks previously managed we want to hear from you. If you viticulturist Grant Yates, for their by Rohan. have any suggestions regarding contribution to SPAA. Training and research are a central research or training needs or are Following four and half years of part of SPAA activities. During willing to share your PA experiences dedicated service to SPAA, Rohan 2006 SPAA organised the second we would be delighted to hear Rainbow’s resignation left SPAA with successful autosteer demonstration from you. SPAA Committee 2006/07

Allan Mayfield, John Heap, Executive SPAA Research Coordinator Senior Research Scientist - SARDI, Malcolm Sargent, Agricultural Consultant, Clare SA. Adelaide SA. SPAA President E [email protected] E [email protected] Farmer, Crystal Brook SA. P 08 8842 3230, M 0418 818 569 P 08 8303 9444, M 0428 833 119 E [email protected] Matt McCallum, P 08 8636 2565, M 0407 395 075 Agricultural Consultant, Committee Ardrossan SA. Mark Branson, E [email protected] SPAA Vice President members: P 0438 895 167 Farmer, Stockport SA. E [email protected] Rob Bramley, Richard Turner, P 08 8528 2412, M 0417 832 776 Principal Research Scientist - CSIRO, Farmer, Snowtown SA. Adelaide SA. E [email protected] Brian Tiller, E [email protected] P 08 8865 2167, M 0428 652 167 SPAA Immediate Past President P 08 8303 8594, M 0417 875 803 Farmer, Warnertown SA. Ashley Wakefield, E [email protected] Randall Wilksch, Farmer, Urania SA. P 08 8634 3053, M 0408 810 068 Farmer, Yeelanna SA. E [email protected] E [email protected] P 08 8834 1219, M 0408 606 177 Peter Cousins, P 08 8676 5050, M 0427 86 5051 SPAA Treasurer Colin Hinze, Agricultural Consultant, Craig Humphris, Viticulturist, Taylors Wines, Crystal Brook SA. Farmer, Jamestown SA. Auburn SA. E [email protected] E [email protected] E [email protected] P 08 8636 2649, M 0408 210 893 P 08 8665 3264, M 0407 586 640 P 08 8849 1171, M 0404 819 330

Precision Ag News  SPAA Research Research program expands Dr Allan Mayfield, Allan Mayfield Consulting Research Coordinator for SPAA

PAA is currently involved with Treatment assessments completed under partnership with KEE Technologies, four research projects; two the first two projects during 2006 include: Nufarm and Croplands. Sfinish in 2007 and two finish Research crops were scanned with • SPAA’s role is to establish five in 2009. The following information five different sensing systems: variable rate pesticide application N-Sensor, GreenSeeker Crop Circle, gives a brief summary of the trial experiments/year for three years, using a hyperspectral camera and a simple precision agriculture and crop sensor objectives and activities during 2006. ground cover imaging system. The technologies. It hardly needs to be said, that as value of each system to asses crop with farming, 2006 has been a biomass and N status between Outcomes in 2006 challenging year for research. zones and treatments was compared. In 2006 two mapping and variable rate Crops were located at Hart, SA and trials looking at pasture desiccation Rupanyup, Vic. Zones were monitored and leaf disease control were PA 1/04 - using the Yield Prophet program. This established on Mark Branson’s, property finishes data is being compared to establish at Tarlee, SA. 30/6/07 whether one data source is more The N-sensor and GreenSeeker were robust than another. used to map ryegrass in peas. This Leaf tissue was sampled from treated information will be used in establishing Project Aims areas. Leaf tissue nutrient analysis was variable rate herbicide treatments at • To increase the effective use of done to compare a range of nutrient sowing () next year. precision agriculture systems by grain concentrations between zones and growers in South Australia. treatment blocks, to assess crop Actions for 2007 • To test the economic value of using nutrient status across the range of We will continue to use different variable rate inputs to grain crops fertiliser treatments. scanners to assess variations in crop to manage both stable and At St Arnaud Vic, the value of different biomass and weeds, and to test variable unstable variables. ryegrass control options to target the rate herbicide and fungicide treatments. • To broaden the application of areas of high ryegrass density, were A further aspect of this research is to precision agriculture to include tested again in 2006. The crops were map and manage higher weed levels lower rainfall areas. scanned in October and results should be that may arise from paddock heaps available for the next magazine in March. dumped by chaff carts at harvest. In addition the N-sensor was used to SPAA00003 map ryegrass in canola crops. finishes Grain samples are being collected from 30/6/07 each treatment area and at fixed points across the nutrition trial paddocks. These Project Aims & Outputs samples will be analysed for nutrient To increase returns to growers by content, including protein and for PA project in Mallee & improved understanding and harvest index. This data will be used to Southeast with Advisory management of within-paddock assess if increased nutrient inputs, such as phosphorus (P), flow through to grain Board of Agriculture variability, and validation of PA systems finishes 28/2/09 under southern Australian conditions nutrient levels. Results will also be used using large-scale trials. in calculating P fertiliser rates for next Project Aims year based on P removal in the grain. • Workshops, field days and printed Seven objectives on increasing the newsletters for farmers and advisors to uptake of PA technologies to improve improve skills in PA and to identify key land and resource management. PA technologies for their businesses. Outcomes in 2006 • Results of large scale field experiments to assess the economic value of Two paddocks were scanned with the variable rate inputs. N-Sensor to map the variation in crop • Assess the practical application of biomass and nitrogen status. the N-Sensor for variable rate N Conservations Agriculture In October Allan Mayfield presented application in cereal and canola crops. project with SANTFA, outcomes from SPAA research at a yield mapping field day at Coomandook, SA. Outcomes in 2006 finishes 30/6/09. Results for the first two years of both Project Aims Actions for 2007 of these trials have been reported in • Release a proven, commercially viable, SPAA will assist with trial design and previous issues of Precision Ag News variable rate, direct chemical injection assessment as well as assist with (www.spaa.com.au/publicaitons) system and application protocol in field days.

 Precision Ag News SPAA Research Research program In-crop sensing expands n-crop applications of nutrients, crop nutrient requirements based on the following information as a particularly urea are becoming crop reflectance. summary of the attributes of each widely accepted as a more of these systems. These help set I This year, Sam Trengove, Allan productive and profitable option. the scene for the three reports Mayfield onsulting,C has been Traditionally, such applications have from growers on their experience been determined by pre-seeding soil running trials using and/or with in-crop sensing as a basis for tests. However, SPAA and several comparing three different sensor nutrient decisions. In future issues growers have been experimenting systems: the N-Sensor, GreenSeeker results from Sam’s research will be with systems that help determine in and Crop Circle. Sam has compiled presented.

Attribute N-Sensor GreenSeeker Crop Circle Height to canopy Height of tractor cab roof 0.8-1.2m 0.25-2.2m (optimal 0.5-1m) Field of view Approx 50m2 (changes with Width = 0.6m (does not Width (m) = 0.57 x height sensor height) change with sensor height) Frequency of measurement 1Hz Configured between 0.33 Configured between 1 to 20Hz (Hz = readings per second) to 50Hz Light source Old model – no Yes (Light emitting diodes) Yes (PolysourceTM) New model (ALS 2005) – yes (Xenon flash lamp) Wavebands emitted from Not specified Red (660nm) and NIR (770nm) Red (650nm) and NIR (880nm) or light source Amber (590nm) and NIR (880nm) Outputs Applied N rate varying according N rate based on N-rich strip. Vegetative indices NDVI, SRI,

to indicative crop N status. Also Also outputs vegetative indices RNIR, RVIS, WDRVI records a biomass index NDVI, SAVI, WDRVI, SRI and IRVI. Mount Tractor cab Handheld, boom or vehicle Handheld, boom or vehicle Weight 15kg 800g 385g Indicies and equations

NDVI = Normalised Difference Vegetation Index = (RNIR-RRed)/(RNIR+RRed) IRVI = Inverse Ratio Vegetation Index = RRed/RNIR

WDRVI = Wide Dynamic Range Vegetation Index = (RNIR-RRed)/(RNIR+RRed) SAVI = Soil Adjusted Vegetation Index = ((RNIR-RRed)/(RNIR+RRed+L))*(1+L)

WDRVI reverts to NDVI if a =1, a typical value for a is 0.1. SAVI reverts to NDVI if L = 0, a typical value for L is 0.5. RNIR = Near Infrared band reflectance

SRI = Simple Ratio Index = RNIR/RRed RVIS = Visible (red) band reflectance

Sensor model/ Variable rate Sensor interface/ Approx Set-up package control software price N-Sensor One sensor, cab mounted Yes N-Sensor terminal $29,500 (UK price) N-Sensor ALS (active One sensor, cab mounted Yes N-Sensor terminal $51,700 (UK price) light source) GreenSeeker Handheld One sensor, mounted on a No, scouting and Ntech Capture and RTMapper $4,640 + $670 for handheld pole or vehicle mount mapping only loaded onto a pocket PC RTMapper software GreenSeeker RT100 One sensor, tractor mounted No, scouting and Ntech Capture and RTMapper, $3,800 + $670 for mapping only. loaded onto a pocket PC RTMapper software GreenSeeker RT200 Six sensors, boom mounted Yes RTCommander $19,870 Crop Circle One sensor, mounted on a No, scouting and GeoSCOUT GLS 400, or data Handheld/ mapping handheld pole or mapping only captured using a laptop PC, PDA vehicle mounted or other data acquisition devices Crop Circle Mapping/ VRA Incorporates one to eight Yes GeoSCOUT GLS 420 sensors, boom mounted The sensors with variable rate control capability are incompatible with some spreaders/controllers. Information on compatibility is available from the sensor manufacturers and dealers. For more information National Distributor N-Sensor None http://fert.yara.co.uk/en/crop_fertilization/advice_tools_and_services/n_sensor/index.html Greenseeker Fairport www.ntechindustries.com www.fairport.com.au/greenseeker/#3 Crop Circle gps-Ag www.hollandscientific.com www.gps-ag.com.au

Precision Ag News  Grower Experience

Guided by the crop

Mark Branson – Stockport, SA In August this year, I sponsored I used GreenSeeker to indicate Mark and Nola Branson together Brenda Tubana from Oklahoma State the crop’s nitrogen status, and to with parents Deane and Jennifer University to run a GreenSeeker calculate its nitrogen requirement. workshop at my farm. The hand farm 1200 hectares near The GreenSeeker calculates the crops held GreenSeeker was provided as Stockport, in the mid North of NDVI and infers that low values are part of that package. The workshop SA. Each year approximately due to N deficiency and high values was a great opportunity for other 80% of the land is cropped to due to sufficiency. canola, wheat, faba beans, peas, growers to gain first hand experience durum wheat and malting barley. of GreenSeeker. I am very pleased with the results The remainder is sown to pasture from the sensor. Clear differences for the self-replacing Merino I used GreenSeeker were recorded between the N-rich and prime lamb flocks. Average strip and the 30m by 100m area of growing season rainfall is 350mm to indicate the crop scanned adjacent to the N-rich and the average wheat yield is strip (See Table 1). These differences 4t/ha. But Mark tries to avoid crop’s nitrogen where not obvious when I visually talking about averages as soils status, and to inspected the crop at late tillering, range from red brown earths to calculate its nitrogen except in Jumbo’s where the crop heavy cracking clays, topography was showing early signs of nitrogen is undulating and rainfall across requirement. deficiency, when compared to the the farm can vary significantly. nitrogen rich strip. Performance The N recommendations generated y first on farm experience Before the workshop I scanned all by the GreenSeeker software were with GreenSeeker was this generally small and due to the dry year, 2006, when I used a my wheat, barley and canola crops. M I wanted to see what differences in season no nitrogen applications hand held GreenSeeker RT100. soil nitrogen uptake existed between were made. During my Nuffield Scholarship in a 30m by 100m nitrogen rich area, 2005, I looked at several different created in each paddock by applying Hitches and glitches scanning options but GreenSeeker 200kg N/ha on the 9th of July, and Initially, I had problems trying to use was the most commercially the rest of the paddock. the sensor and its software as the advanced; I was keen to experiment with it on my own crops. Table 1 Mark Branson’s wheat crops scanned on the 3rd August, only the crop in Jumbo’s showed any visual signs of N deficiency. My interest in understanding the real rather than predicted relationship Days from N Rich Standard Max yield N rate Paddock between soil available nitrogen (N) sowing NVDI NVDI (kg/ha) rec. and crop nitrogen status was triggered by my frustration with Jumbos 65 0.474 0.377 4200 42 nitrogen calculators. The nitrogen Barr 75 0.76 0.65 4800 35 mineralisation calculations used in Barn 73 0.637 0.592 4000 14 these tend to over or underestimate my nitrogen fertiliser requirement. TamFm 72 0.68 0.66 4100 5

 Precision Ag News manual was not written very well; Phil Longmire – Esperance, WA Grower Experience after a few brief phone calls these Phil and Belinda Longmire problems were overcome. There crop 4500 hectares of their was a problem at the start with the 8500 hectare farm northeast handheld computer not reading the of Esperance, WA, with the software but after rebooting the remainder under pasture. The handheld computer this has not Longmires have farmed the been a problem. property for 18 years, as part of I am using the software that comes a family operation with Phil’s with the GreenSeeker and have not parents, Ian and Chris. The had any problems, so far. Next year I family crops barley, wheat, will attempt to use it with the ZYNX pulses and canola, grow pastures computer for variable rate work and and run 10,000 Dohne and will see if there are any compatibility Merino sheep. This is medium problems. At the moment I am rainfall cropping country with using NTech Capture for collecting an annual rainfall of 425mm basic NVDI data and RT Mapper for and a growing season rainfall of Geo-referenced NVDI mapping. This 325mm, resulting in an average data is able to be transferred into wheat yield of 2.6t/ha. Fairport’s Farmstar software. wanted to be able to identify Where to from here? variability that I could not see Even with close inspection it is unlikely that Guided by the crop I am keen to use crop sensing to I by looking at the crop and then nutrient deficiency will be visually identified identify areas of high biomass, tissue testing to establish if there until it is severe. Sensors such as GreenSeeker to map these and then carryout was a nutrient deficiency in these can help provide an early warning. paddock checks to see if areas of areas. Having seen the GreenSeeker herbicide, at rates of up to 3.5L to high biomass are in reality high system being used effectively on these areas, with a constant base of density pockets of ryegrass. If this a farm in North Dakota, USA, for 1L, this will be cost effective. Our works, this information would VRT nitrogen application and crop trial work with high rates of foliar provide the basis for variable mapping, I was keen to test the manganese this season is showing rate herbicide application and system on our own crops. good responses. sowing rates. After the test paddocks have been I believe the collection of NDVI data From the NDVI map harvested we will compare the NDVI could be of great benefit in the we calculated that map with yield data and analyse the paddock choice for and nutrient effect of variation and the management of malting barley, and dry season. for early post nitrogen applications approximately 15% according to how the canopy is of the paddock is Hitches and glitches developing during tillering. highly manganese To date we have not really put the I am definitely in the market for this software interfaces to the test. type of sensor and am looking for deficient. Apparently, compatibility with our the one with the best optics and that ZYNX system is only a formality will work with my fertiliser boom. Performance according to both parties. We are From my experience, the Crop Circle In 2006 a GreenSeeker scanner was hoping to analyse the data through sensor seems to be a better sensor made available for use in our area SMS Advanced. than the GreenSeeker but at this through Leighton Wilksch, Landmark, Where to from here stage does not have the agronomic SA. It consisted of a single sensor, development of the GreenSeeker. I am excited about the potential of fitted to the back of a ute and driven gathering biomass data at every pass I am investigating the use of satellite across the whole paddock following of the boom-spray for both crops and aerial NVDI collection across a our autosteer lines. From the data and pastures. whole paddock. My aim is to see a NDVI map was created. We have if there is potential to use these also been looking at the Crop These data layers will be analysed to relatively cheap sources of NVDI data Circle sensor from Holland try and find the key limiting factor as a basis for calculating variable rate Scientific, Nebraska. - be it rainfall, compaction, nutrition nitrogen. I will use the handheld or disease. Our first priority is to We scanned 180 hectares of hard supply high rates of nutrient to the sensor to verify the data. To use wheat at mid tillering. Tissue testing these other sources of NDVI data small proportion of cropped area plants from areas of low NDVI, as that requires a nutrient boost. Our with the GreenSeeker software I will indicated by the GreenSeeker, has need to create baseline nitrogen aim is to make nutrient application highlighted that manganese is the cost effective and highly responsive. data for the soils and environments most limiting nutrient. From the on my farm. Hopefully the layers of data will NDVI map we have calculated that eventually paint a consistent For more information approximately 15% of the paddock pattern to help indicate more http://nue.okstate.edu/CCA_2005/ is highly manganese deficient. If we options for effective and affordable Australia_Workshop2006.htm can apply foliar manganese with our cost applications.

Precision Ag News  Grower Experience ince 2003 we have used areas of high potential being 50kgN/ high nitrogen (N) strips ha, the same as the soil test. in-crop and found these to be Currently we are looking at S We used the GreenSeeker data to much more accurate than soil tests purchasing two Crop Circle sensors create a variable rate application for determining the need for top (from our experience this system has map and urea was side dressed dressed nitrogen. given us a more accurate footprint) in alternate strips at 40, 60 and and locating them across our 36m In 2003 and 2004 the rate of 80 kg/ha. boom-spray. This will give us an nitrogen recommended from soil No economic response to N was average NDVI across the width of the tests for a 5t/ha crop was 100kgN/ recorded at any rate. However, boom, based on two 18m passes. ha. N-rich strips were created and if we had used the GreenSeeker As we can only vary inputs in the visually compared to the crop; these recommendation rather than N direction of travel and not across indicated no fertiliser was required. requirement generated by the soil the width of the boom, we feel two Comparisons were also made using test we would have saved 40kg/ha sensors will be sufficient. tiller counts, tissue tests, an N- of urea across the paddock. To apply the liquid nutrients we will Sensor, aerial and satellite imagery. use an additional tank, controller All methods indicated that the crop This year all our winter crop nitrogen and pumping system and plumb did not require nitrogen fertiliser. applications were applied as UAN, on-the- go by the GreenSeeker the boom into our second spray Despite all this data we lacked using a Hardi controlled on Hardi line. This will enable us to apply confidence and applied fertiliser to Commander boomspray fitted with nutrients at the same time as a proportion of the crop, leaving fertiliser nozzles. There were no broadleaf herbicides. enough strips without fertiliser to compatibility problems between the allow us to measure a range of programs and the rate response was treatments. While there was a yield generally very good. benefit from applying N it was not enough to cover the cost of the I like the fact that the GreenSeeker in-crop N and its application. software logs the recommended N and the actual N rate applied. This is In winter 2005, a net benefit of important as the actual rate depends $20/ha, over normal practice, was on rate of flow response by the achieved by top dressing nitrogen control applicator. at rates determined by visual observation of nitrogen rich strips. The benefit was gained mainly ‘benefit was gained from savings in fertiliser rather mainly from savings than extra yield. in fertiliser‘ Performance Hitches and glitches In October 2005 we had our first on-farm use of GreenSeeker RT200, We have been happy with the although we had tested a hand held system. Data is imported from the GreenSeeker software into

Photo Grain Business RT100 in 2004. We opted for the GreenSeeker as unlike the N-Sensor FarmWorks, Farmsite and this NSW cropper Richard Heath has been using works well. N-rich strips as a method of judging crop it has its own light source, allowing it nitrogen requirement for the past four years. to be used anytime of day or night. The biggest challenge is to know In 2005 he started using a GreenSeeker to We also liked GreenSeeker because where to locate the N-rich strips. calculate in-crop N requirements. it is supported by software based Our aim is to place them in a part of on a robust agronomy package. the paddock where N is anticipated Richard Heath – The GreenSeeker software gives a to be the only limiting factor. We Gunnedah, NSW recommendation based on the crop use a combination of knowledge and Pine Cliff is a mixed cropping data not by varying a predetermined information from PA data to help and grazing property on the nitrogen rate. locate the strips. In a paddock over Liverpool Plains. Richard with In the first year we used the 100 hectares we use two his father and three brothers GreenSeeker to map and N-rich strips. crop 3300 hectares of winter and recommend rates for side dressing summer crops including wheat, Where to from here? urea on corn and . corn and sorghum. Annual As the GreenSeeker is permanently rainfall is 636mm (summer In the corn paddock soil tests on the boom-spray it is used to dominant) and average wheat recommended a rate of 50kgN/ record NDVI every time a spray yields are 4.5t/ha. In this area ha. N-rich strips were established application occurs. This data is the predominant method by applying 60kgN/ha pre sowing mapped giving us a picture of of nitrogen fertilisation is and these strips where visually how biomass is changing across a incorporation of anhydrous evident by side dressing, six-weeks paddock during the year. In future ammonia or urea either in the after planting. The GreenSeeker we may use this information for months before seeding or during recommendations for the block planning variable rate use of plant a one-pass seeding operation. averaged 25kgN/ha (64kg/ha urea), growth regulants and possibly Top dressing is uncommon. with the recommended rate in the fungicides.

10 Precision Ag News Grower Experience

Spot spraying in broadacre cropping systems

David Brownhill farms on the Liverpool Plains, NSW. He was so delighted with the 89% reduction in fallow herbicide achieved using a WeedSeeker that he secured its Australian distribution licence. David Brownhill, Crop Optics Australia

rop and weed sensing commercial release of WeedSeeker herbicides, insecticides, fertilizers and technology has existed as selective application equipment by fungicides to plants in a wide range Ca research tool in Australia US based company N-Tech Industries. of agricultural situations including since 1984. The concept was Selective spot spraying technology cotton, broadacre, horticulture originally developed by researcher was commercialised in the USA and viticulture. Warwick Felton at the DPI Tamworth. seven years ago and has found Anywhere a product can be However, the reliable application wide commercial application targeted to the plant and not of automated spot spraying in around the globe in all types of bare soil, the WeedsSeeker can broadacre agriculture has only agricultural environments. Sensors provide large savings and big can be used to selectively apply recently been possible with the environmental benefits. Table 2 Examples of herbicide savings when using WeedSeeker for Following my Nuffield Scholarship, weed control in fallow. I imported the first WeedSeeker sensors for broadacre use into Jamie Grant Peter Farrell Dalby QLD Moree NSW Australia in 2002 using matching grant money provided by AFFA Field area 246 hectares 120 hectares under the “Farm Innovation Weeds Peachvine, milk thistle, Volunteer cotton Program”. Sixty four sensors were fleabane, volunteer cotton 30cm high purchased and set-up on our 24m Herbicide 2.6L/ha Round up + 1L/ha Starane® + wide boomspray. The sprayer was 4L/ha Surpass® 1L/ha MCPA a tow behind set-up and tramlines Actual area sprayed 11.88ha (4.5%) 18ha (15%) were the only guidance; an 89% Cost of blanket spray $7840 $3360 saving in herbicide was achieved Cost with WeedSeeker $353 $504 across our fallow paddocks. This level of saving paid for our Herbicide saving per field $7487 $2865 investment in WeedSeeker within Herbicide saving per hectare $30.43 $23.80 two years.

Precision Ag News 11 We now run WeedSeeker sensors expensive and environmentally across a 27m, self propelled spray-rig unstainable. with 2cm accuracy GPS guidance; There are now 13 commercial spray- with this we achieve the same rigs across NSW and Qld operated savings in herbicide. by farmers and contractors using DPI research in Northern NSW has WeedSeeker. These are achieving shown that average weed cover in large reductions in fallow herbicide fallow paddocks is as low as 20% of use with savings of 80% and better. the paddock area. Therefore, 80% Cost examples of the possible Consisting of a sensor and spray nozzle with of the herbicide applied falls on bare savings using WeedSeeker for fallow fast-fire solenoid valve the WeedSeeker soil and is wasted. This is inefficient, weed control are found in Table 2. can accurately spot spray weeds reducing herbicide use and environmental impact. How WeedSeeker average weed cover from only applying herbicide to the weed targets. Others relate works in fallow paddocks is to management of herbicide WeedSeeker consists of an resistance and reduced herbicide as low as 20% integrated light emitting diodes inputs with no-till, not to mention (LED), light detector and a spray I believe there is a range of the environmental benefits of less nozzle controlled by a fast-fire applications for WeedSeeker. spray drift and reduced potential solenoid valve. for off-target impacts from drift and 1) Broadacre fallow spraying herbicide leaching. 1. The LED produces a combination of 2) Shielded spraying in row crops For example, in relation to managing invisible infrared 3) Application of fungicides, herbicide resistance WeedSeeker and visible red insecticides and fertiliser in allows us to use mixtures of different light, this is horticultural crops herbicide groups eg Select® to projected into control ryegrass in the fallow, 4) Spot spraying on irrigation the target weeds, Garlon® for melons etc, which may channels approximately be currently too expensive to apply 60cm below the sensor. 5) Viticulture and tree crops in a blanket application. This will help reduce resistance in weed 2. The light reflected from the 6) Industrial uses including weeds on populations and prolong the life of target is captured by the roads, railways, airports etc existing herbicides, greatly improving detector at the front of each sustainability of cropping systems. WeedSeeker sensor. For WeedSeeker benefits 100% ground coverage The reduction in herbicide cost For more information sensors are spaced at 38cm due to target spot spraying is David Brownhill 0427 473725, across the boom. just one of the benefits gained [email protected] 3. Sophisticated circuits inside the sensor analyse the reflected light and determine when it matches the light reflected by green plants. The WeedSeeker circuitry is not able to differentiate between the intensity of green nor estimate biomass as is done by the GreenSeeker sensor. 4. When reflectance from green plants is identified, the sensor waits until the plant is under the spray nozzle and then triggers a fast-fire solenoid valve that sprays the plant. In the cab WeedSeeker is controlled by a simple analogue control box. Controls include on/off, adjustment for weed size and calibration for herbicide rate and speed of travel. Anywhere a product can be targeted to the plant and not bare soil, the WeedsSeeker can provide large savings and big environmental benefits.

12 Precision Ag News Research -Pasture Virtual grazing systems The summation of 30 years of research on animal behaviour and the greater availability and affordability of satellite based navigation systems has enabled Dr Dean Anderson to develop the ultimate fencing solution – Directional Virtual Fencing (DVF™). During a three month sabbatical with CSIRO Dr Anderson explained his system to Australian livestock managers. Emma Leonard ntensive rotational grazing is By combining physics, electronics, the GPS identifies an animal is too widely accepted as a system that biology and ecology Dr Anderson close to the virtual line the cues will Iincreases pasture productivity and has developed a locator/controller be triggered until the animal has herd performance. However, this collar that produces the audio and responded appropriately. electric signals. But how does the system can be expensive to establish Animals have been found to collar know when to emit and time consuming to manage. condition rapidly to these cues and the signals? Dr Dean Anderson, has developed a the process is consistent with low system of virtual boundaries that is Global Positioning Satellites (GPS) stress animal handling procedures. able to manage animal movement are used to determine an animal’s in both intensive and extensive Using Dr Anderson’s system farmers location. This location data is will be able to program stock livestock operations. transmitted from the collar to a Dr Anderson is a Research Animal movement from their computer. Geographic Information System (GIS) This could be moving fences hourly Scientist with the US Department of that determines whether the animal Agriculture based at the Journada in intensive grazing situations to is in the correct location range. droving cattle across range lands Experimental Range in New If not or if too close to the boundary Mexico. The mission at Jourdana by regularly moving a virtual sends cues to the collar to emit back fence. is to develop new knowledge of signals to the right or left side of ecosystem processes as a basis the animal. The introduction of stock to graze for management and remediation stubbles in unfenced paddocks could of desert rangelands. These also be possible with this system. rangelands, like those in Australia, The virtual line or Dr Anderson believes that farmers are fragile areas carrying large fence is determined need to change their current numbers of stock at low stocking thinking about fencing and rates, where grazing conditions can by the manager embrace the opportunities that new change considerably depending on submitting technologies offer; he suggest many patterns of rainfall. coordinates into have yet to be imagined. Research and practice has shown that grazing management can the GIS system. “By 2050 internal conventional yield improved productivity and fences will be obsolete with be less damaging to these fragile Audio cues are the least severe cue conventional fences only being environments. Dr Anderson has while a combination of audio and required around the boundary of the focused nearly 30 years of research electric shock are the most severe. property,” said Dean Anderson. on developing animal management Even at the more sever levels the Research has proved that virtual systems to control the movement of system causes the animal minimum fences work and have a range of grazing animals by using their innate stress. The cues are applied either practical applications but the product behaviour, and precision guidance to the right or left side of the has yet to be commercialised. software. Earlier this year Dr animal depending on the angle of Dr Anderson anticipates this will Anderson presented his Directional the animals head with respect to a occur in the foreseeable future. Virtual Fencing™ system to livestock virtual line. producers in SA and WA. He is continuing to work on the The virtual line or fence is concept and is currently investigating From his research Dr Anderson has determined by the manager the identification and use of collars established that applying audio or submitting coordinates into the only on lead animals in a herd electric shock cues to an animal’s GIS system. The coordinates can or flock. right or left side controls animal relate to a single boundary that location and direction of movement; joins physical fences or can consist For more information the stronger the cue the greater of multiple virtual boundaries Dean Anderson the response. producing a virtual enclosure. When [email protected]

Precision Ag News 13 An exciting new reference to all aspects of PA is available on disc from GRDC, and what is more PA on disc it’s free. The GRDC Precision Agriculture Manual is a step by step guide to the background and implementation of PA. With contributions from many leading researchers, agronomists and farmers involved in the GRDC PA research initiative, the Manual provides a unique resource for anyone who wants to learn, research or implement PA. The Manual has been produced on CD and the documents have numerous active links so they can be jointly searched, and are referenced to websites with additional information.This is the first time all this information has been collated one place. It is hoped that the PA Manual will be used by many audiences: • educators • researchers • advisers & growers The CD can be ordered online through the GRDC website ‘Grainzone’ http://www.grdc.com.au/bookshop/free.htm#new or by contacting Canprint on freecall 1800 11 00 44 Free Fax: 1800 00 99 98

An introduction to, and discussion of, the various tools and technologies associated with Precision Viticulture. How does someone interested in using these tools make a start in adopting Precision Viticulture practices? Learn how spatial information should be acquired, delivered and used and arm yourself with the questions that need to be asked of a service provider. The practical application of Precision Viticulture is a central theme to this book and a number of commercial case studies are presented.

Phone Winetitles on (08) 8292 0888 or visit our website www.winebiz.com.au

14 Precision Ag News Research -Pasture

Mapping actual fertiliser distribution Dr Ian Yule and Haden Lawrence, Massey University, Palmerston North

ver four million tonnes was perfectly controlled by the paddock. Without the use of of fertiliser are spread in on-board computer. GPS guidance a field CV of 90% New Zealand each year. was recorded. The use of GPS O Actual average application rates In pastoral production, covering varied from 51.8 to 106.7kg/ha of guidance reduced this to 60%, 38% NZ, the majority is applied urea (46%N) spread on individual still considerably higher than from as granular fertiliser distributed paddocks. The typical field CV was ground spreading without GPS by on-ground or aerial spreading found to be 37.9% across the four guidance. Further tests on a 2,200 without GPS guidance. The recent farms. hectare property produced a CV of introduction of environmental 72%. The main reason for the legislation in relation to fertiliser Variation due to driving accuracy high CV with aerial spreading was application has increased the need and driving method was calculated due to the lack of a flow control for a better understanding of current to be 22.9%, considerably greater system to adjust for fluctuations in application performance levels and than the 15% calculated from the aircraft speed. the economic benefit that may be transverse test. The CV is likely to achieved by changing these levels. increase travelling across a slope. Estimates of economic loss found that poor spreading from aircraft Internationally recognised testing Paddock shape was found to have was typically costing $10/ha mainly methods (transverse tests) for an impact on performance with due to changes in aircraft speed and fertiliser spreaders report that the irregular shaped paddocks having uneven swath matching, resulting in expected coefficient of variation (CV) a high application variation, up to overlaps or missed strips. Reducing for products containing nitrogen is 40.8%. The highest calculated fertiliser use by not spreading on at least 15% and 25% for products paddock variation was 62.3% on without nitrogen. a 0.8 hectare, irregular shaped unproductive areas provided a paddock. further cost saving of $9/ha per Using Geographic Information annum. Assuming perfect spreading Systems (GIS) and the positional Our research has shown, that where achieved the adoption of information from the spreading by improving driving accuracy variable rate application technology operation, the actual CV achieved considerable improvements in field has the potential to produce a was calculated for ground spreading CV could be achieved. For dairy financial benefit of $85/ha; raising operations on 102 paddocks on four farmers using small paddocks (1-2ha) the financial performance of these dairy farms. it is estimated that an achievable areas from around $320/ha to field CV for fertiliser distribution is Fifty nine of the paddocks were $405/ha. 25-30%, if GPS guidance is used. uniform in shape and 43 non- Within the study the typical cost of It is hoped that these results will uniform; paddock size ranged from lost production due to inaccurate help encourage producers to invest 0.48 hectares to 7.45 hectares. A spreading for the dairy farmers was in guidance and systems of fertiliser target rate of 80kg/ha of fertiliser between $50/ha and $70/ha over delivery that accurately deliver the was set for each paddock. The work the full year. desired fertiliser rate. assumes a mechanical drive is used for the spinners, so spread rate is Similar methods were used for For more information Dr Ian Yule linked to speed and that application testing the accuracy of fertiliser 00 11 64 (0)6 3504340, variation due to speed fluctuations applied by aircraft over a 20 hectare [email protected]

Precision Ag News 15 Research - cropping

Data layers for soilborne disease Dr John Heap and Dr Alan McKay, SARDI

oilborne diseases continue Our recent research assessed the Six zone models were generated, to cause substantial losses in usefulness of the many data layers using different combinations of Sbroadacre field crops. The available for mapping soilborne data layers (Table 3). Inoculum was spatial distribution of diseases such disease inoculum (Table 3). measured for points on a grid over as take-all, rhizoctonia and crown rot the whole paddock, and subsets is often variable within a paddock of this grid were used to calculate and is correlated with variation in soil root disease the level within each zone. All attributes, topography and patterns inoculum sampling zone models (except (f); Table 3) of plant growth (eg yield, biomass). zones can be derived combined multiple data layers using a clustering technique in Earlier research supported by SARDI/ from the cheapest JMP (“Jump”) computer software. GRDC showed that disease inoculum Zone models a) and b) were also levels frequently differed between available data compared over an additional Precision Agriculture (PA) zones. eight paddocks. Knowledge of which zones are at Spatial data layers (approximately 30 risk from diseases is valuable for layers per paddock) were collected A “Partition Index” (PI) was devised designing management strategies, so over five paddocks to determine to compare the relative ability testing soil from different zones was which combinations of layers are of each zone model to partition encouraged. However, further work most suitable for defining zones for paddocks into zones with different disease inoculum measurement. We was required to identify which data inoculum levels. This index takes compared zone maps based on a sources provided the most valuable into account differences between single data layer through to maps information on inoculum levels. zone inoculum means and the whole based on combinations of many paddock mean, as well as the size of Many data layers (eg yield, elevation, layers. The results suggest new each zone. biomass etc) can be mapped, using guidelines for the best method to a range of proximally- and remotely- identify differences in inoculum level Averaged over the five paddocks sensed data collection systems. between PA zones. proximal (a), satellite (b) and custom

16 Precision Ag News Table 3 Six zone models were generated for each paddock, and Research - cropping inoculum levels were measured from points within each zone.

Model Data layers used for zone map a) Proximally-sensed data (yield, ECa, elevation) b) Satellite NDVI biomass data. c) Custom disease zones (using correlation matrices and Forward Stepwise Multiple Linear Regression to choose layers - “Custom MLR” - selected from all available data layers).

d) “Biological” layers (yield, NDVI biomass, aerial photography, N-Sensor). e) “Geological layers” (ECa, elevation, slope, gamma-radiometric and magnetic susceptibility). f) ECa (eg EM38) alone.

MLR (c) zone models were found Figure 1. Comparison of usefulness (Partition Index) for zones based to be equally robust, consistently on proximally-sensed (Prox) and Satellite NDVI (Sat) data for a range producing zones with different levels of soilborne diseases over 13 sites. (n= observations) of inoculum. “Biological” (d) and “Geological” (e) models were equal Comparison of mean Partition Index for zones based on but less useful than a), b) and c), proximally-sensed (Prox) and Satellite NDVI (Sat) data for a while ECa alone was least useful. range of soilborne diseases over 13 sites. (n=observations Use of custom MLR zones (c) is probably unwarranted, because it is 1.4 n=4 less practical for growers. n=5 n=4 n=2 1.2 n=3 When proximal (a) and satellite (b) n=6 1 models were compared over 13 n=33 sites (Figure 1) the results suggested 0.8 n=9 that they were equally useful in the identification of zones with different 0.6 inoculum levels for the major diseases take-all, rhizoctonia 0.4

and CCN. Partition Index (mean) 0.2 Rs TA CCN CRp Pn Pt CCR Total Satellite NDVI data appeared to 0-0.2 be better at predicting crown rot distribution, while the proximal Prox Sat model was better for common root rot, Pratylenchus neglectus, and Abbreviations for diseases: Pratylenchus thornei. CCN = cereal cyst nematode; Given the complexity of the CRp = crownrot: Fusarium interactions involved it appears psuedograminearum; difficult to predict in advance the CRR = common root rot; usefulness of specific models for Pn = Pratylenchus neglectus; specific diseases. However, broad Pt = Pratylenchus thornei; suggestions can be made. Rs = rhizoctonia; TA = take-all. Based on our work to date it is considered that proximal and satellite For more information models are similarly useful when Dr John Heap (08) 8303 9444 testing for the full range of diseases. [email protected] Therefore, it is suggested that root disease inoculum sampling zones be Proximal data sources such as yield and ECa derived from the cheapest available or remotely sensed satellitebiomass data where all found to be useful data sources proximal or satellite data. for identifying cereal root disease.

Precision Ag News 17 Research - viticulture

Practical applications of precision

viticulture Photo Richard Merry, CSIRO Rob Bramley, CSIRO Sustainable Ecosystems

Selective harvesting the use of remotely sensed imagery, In this example from the Margaret with or without yield mapping, as a River region of Western Australia, in Margaret River basis for ‘selective harvesting’. an evaluation of selective harvesting n general, the main goal of was conducted in a 3.3 hectare Precision Agriculture (PA) is to the gross retail section of a much larger Cabernet Igain control over a variable Sauvignon vineyard. The vineyard production system so that there is an value of production manager and winemaker were increased likelihood that the outputs was estimated to interested to see whether there from the production system are the was a cost benefit from selective harvesting. desired ones. Whilst intuitively very be increased by similar, Precision Viticulture (PV) approximately Figure 2a is a remotely sensed differs markedly from PA in terms of $139,480 image of the study area acquired the objectives of their early adopters. in 2002 at veraison, the time when Broadacre cereal producers have Selective harvesting is the split grapes begin to soften, colour and overwhelmingly used yield mapping picking of fruit at harvest according ripen. The image was acquired and other tools, such as high to different yield/quality criteria, in using airborne digital multispectral resolution soil survey (eg EM38) and order to exploit observed variation video imagery, the most common elevation modelling, to promote the – generally in fruit quality. Rather commercially available form of variable rate application (VRA) of than focussing on differential vineyard remote sensing in Australia. inputs to the production system. In management of production inputs, Veraison has been shown to be the contrast, the early adopters of PV selective harvesting involves the most informative time for acquisition have placed much greater focus on differential collection of outputs. of such imagery enabling variation

18 Precision Ag News Research - viticulture

would have been allocated to the lower end-use product. Based on these wine prices, the tonnage of fruit harvested from each zone, and the assumption that one tonne of fruit produces 750 litres of wine, the gross retail value of production was estimated to be increased by approximately $139,480 over the 3.3 hectares using selective, as opposed to uniform harvesting. This is equivalent to $3,653/t fruit harvested or $42,267/ha. The additional costs involved in selective harvesting were confined to the costs of running the second chaser bin/tractor. Winemaking costs associated with the two products were the same. Figure 2 a&b. Performance of a 3.3ha area of Cabernet Sauvignon in Margaret River in terms In 2003, this block was again split of (a) PCD at veraison 2002, and (b) vintage the same year. In (a), blue indicates relatively higher vigour and red indicates low vigour. The horizontal line in both figures delineates the lower and into the two zones and harvested higher yielding zones, the fruit from which was selectively harvested into separate bins. differentially. However, instead of picking each zone on the same day, in photosynthetically active biomass vigour) and southern (low PCD/low the more vigorous northern zone (PAB) to be identified. Remote vigour) zones. It was subsequently was harvested nine days later in sensing provides a surrogate selectively harvested using a order to enable the fruit to become estimate of PAB – in this case mechanical harvester fitted with a more physiologically ripe. This through the so-called ‘plant cell grape yield monitor and differential strategy had the additional benefit density’ index (PCD), which is GPS. Two chaser bins ran alongside of removing the requirement for the calculated as the ratio of reflected for the separate collection of fruit second chaser bin during harvest, infrared to red light. A large PAB (ie from the two zones. The yield map reducing the cost of selective higher values of PCD) is a reflection produced (Figure 2b) showed that harvesting. While wine made from of a large, healthy (ie vigorous) whilst the average yield for the the less vigorous southern zone canopy, whereas a low PAB reflects whole block was 13t/ha, the higher was again allocated to the varietal either that the canopy is small and/or vigour northern zone had an average Cabernet Sauvignon brand, wine that it is under stress (low vigour). yield of 16t/ha whereas the southern made from fruit harvested from the As Figure 2a indicates, this particular zone yielded 8t/ha. The variation in northern zone was allocated to a vineyard has a more vigorous (high yield across the block was found to PCD) northern section and a less resemble closely variation in PCD. higher end product than in 2002 vigorous (low PCD) southern section. (a Cabernet Merlot blend with a The two parcels of fruit were Note that the rows run north-south, retail value of approximately $22.50/ processed separately in the winery. which means that a single row may bottle), because of the enhanced After vinification, differences in wine express the full range of vigour seen fruit ripeness. quality between the wines from in the whole block. the two zones were deemed large In 2004 and 2005, the block was again selectively harvested. As in Two weeks prior to the expected enough to justify allocation of the harvest date, vines in areas of low wines to different end products. 2002, each zone was picked on the and high PCD were assessed for same day but fruit was separated Wine made from fruit harvested during harvest using one tonne canopy vigour, and samples of fruit from the northern zone was were analysed for sugar (baumé), pH capacity bins mounted on a four bin allocated to the ‘Classic Dry Red’ trailer (4t total). Individual bins were and titratable acidity (TA). Sensory brand (retail price approximately used to separate the low and high assessment of the fruit by the $19/bottle), while wine made from PCD fruit by simply changing the winemaker was also carried out. The fruit harvested from the southern filling position on the picking trailer. results confirmed that differences in zone was allocated to a varietal This strategy, as in 2003, removed PCD translated into real differences Cabernet Sauvignon brand (retail the requirement/cost of running a on the ground. price approximately $30/bottle). second chaser bin/tractor, but still Accordingly, the block was divided If the block had been harvested allowed selective harvesting to be into northern (high PCD/high as a single unit, the resulting wine carried out in a single harvest event.

Precision Ag News 19 Research - viticulture determined each year. Nevertheless, study was conducted wanted to selective harvesting offers both know whether, and to what extent, Research conducted over several grapegrowers and winemakers such salinisation was impacting on years in other vineyards by Richard the opportunity to take advantage vineyard performance. Hamilton and I, along with of variability within their colleagues from CSIRO, CRCV and production systems. An EM38 soil survey was carried Foster’s has shown that patterns out at the site in December 2000 of spatial variation in vineyard and the following vintage the block productivity (yield, vigour) tend to Improved was yield monitored (Figure 3). An be fairly stable in time – as was the natural resource elevation model of the site was also case in this example. The research produced following survey with also strongly suggests that patterns management in the a real time kinematic GPS (RTK) of variation in fruit quality follow Clare Valley (normally done with the EM38 survey). While the differential GPS, patterns of variation in yield. So, he Clare Valley has limited until on-the-go fruit quality sensing as used for yield mapping and EM38 supplies of good quality water; survey, is accurate to about ± 50 cm technologies are available, the use much of the groundwater is of remotely sensed imagery and T in the horizontal planes, its accuracy salty and annual average rainfall is yield monitor data appears to be an in the vertical plane is several only 650mm, with most falling in metres. In contrast, RTK is accurate appropriate basis for delineation of winter when vines are dormant. In zones for pre-vintage fruit quality to 2-3cm in both the horizontal order to secure a supply of irrigation and vertical planes. assessment and subsequent water for the growing season, many selective harvesting. vineyards have installed surface Soil samples collected from positions Of course, whether inter-zone water dams. However, this strategy chosen to cover the full range of quality differences are large enough increases the risk of soil salinisation variation in the EM38 signal (Figure to move the resultant wines from as a consequence of the raised 3a) were analysed for a range of one price point to another, as in saline water table in the vicinity soil properties, including electrical this Margaret River example, is of the dam. The manager of the conductivity (EC), clay content something that will need to be 24 hectare vineyard in which this and exchangeable cations. Note

Figure 3. Variation in (a) bulk electrical soil conductivity and (b) yield (vintage 2001) in a 24 ha vineyard in the Clare Valley, South Australia. The triangles in (a) indicate the location of soil sampling sites used for ground-truthing the EM signal. Different varieties are grown in each of the sub-blocks making up this vineyard. To account for between-variety differences in yield potential, the data were normalised (mean = 0, standard deviation = 1) on a per variety basis prior to mapping. There is a large surface water dam on the north-western side of this vineyard.

20 Precision Ag News Research - viticulture

benefits in terms of the productivity and long term sustainability of this vineyard. In addition to providing quantitative evidence that soil salinity has a detrimental impact on vineyard productivity, this case study demonstrates that the application of PV does not have to be confined to optimising production outcomes. These results strongly support the view that a PV approach to assessment of vineyard constraints offers a means of providing growers with knowledge of the precise location and extent of the constraints, and the basis for targeted ameliorative management. Figure 4. Results of clustering salinity (EM38) and yield (vintage 2001) data in a 24ha Clare Valley vineyard. For each cluster, the first value in the legend represents the mean bulk electrical Whole-of-vineyard soil conductivity (dS/m) in the cluster, whilst the second value is the mean cluster yield. For this analysis, the yield data were normalised to account for differences in yield potential between the experimentation in six varieties grown in block. The range in elevation from the lowest to highest point in the block is approximately 13m. Langhorne Creek that conductivity arises in soil as a • Cluster 2 separates out hen conducting consequence of the presence of salt, above-average yielding areas experiments, researchers the type and amount of clay, and characterised by moderate Whave traditionally used the amount of water and organic soil salinity, random location of treatments in matter. Calibration of the EM38 small plot trials to overcome the • Cluster 3 is made up of lower signal (bulk electrical conductivity effects of spatial variation. However, yielding areas in which salinity, in the soil, against soil analytical as the two previous examples as measured by EM38, is not data) demonstrated that the EM38 illustrate, spatial variation in expected to be yield limiting. signal was dominated by soil salinity. paddocks tends not to be random. So is there a better way of designing Therefore, for all practical purposes, The latter suggests that Cluster agricultural trials? Might there be an Figure 3a can be regarded a soil 3 areas may be subject to a yield advantage in doing the experiment salinity map. This conclusion is limiting factor not reflected by over the whole management unit? strongly supported by the fact that the EM38 signal. Cluster 1 occurs At CSIRO/CRCV we have been when the EM38 map is draped over mainly in the lower lying areas, addressing this question as a part the digital elevation model derived especially in the western part of the of our Precision Viticulture research. from the RTK survey, the areas of block close to the dam, where the The following example is based highest apparent salinity occur in groundwater table may be expected on work carried out at Langhorne the lowest lying areas (Figure 4) to be closest to the soil surface. – the expected result in a Creek, SA by Dean Lanyon. It salt-affected landscape. demonstrates the management salinity in this related benefits gained by As Figure 3 shows, both yield using whole-of-block rather and electrical conductivity varied vineyard is reducing than plot designs. markedly within the 24 hectare study yield by 5-27% The manager of this Langhorne area, with most of the apparently Creek vineyard was concerned that saline areas corresponding to areas Further analysis of the data production was being limited by of lower yields. This is especially underlying Figures 3 and 4 suggests poor soil conditions. On ground apparent when the yield and EM38 that salinity in this vineyard sampling had identified that the soil data are clustered into zones using a is reducing yield by 5-27%. volume explored by vine roots was statistical procedure known as Accordingly, it was suggested to the constrained by the combined effects k-means clustering (Figure 4). vineyard manager that a re-design of wheel track compaction in the • Cluster 1 is made up of areas of the irrigation system and inter-row, and in an inhospitable which are characterised by re-location of the surface-water sodic subsoil containing toxic both low yields and high soil dams to higher parts of the concentrations of boron. Since conductivities (ie high salinity), landscape may deliver substantial the soils at this site vary in clay

Precision Ag News 21 Research - viticulture enabled construction of treatment- benefit greater than that likely specific yield maps for the whole from ripping. content and depth to the B horizon, block (Figures 6a-c), even though Figures 6d-e provides the vineyard the response to soil amelioration in this case, the treatments were manager with a powerful basis treatments was expected to effectively only applied to a third for decision making. In particular, be variable. of block. they allow construction of Figure A highly replicated trial design On first inspection, comparison 6f which shows those areas where was applied across the whole of the three treatment-specific there was both a treatment benefit 6.8 hectare Shiraz vineyard block maps (Figures 6a-c) suggests little with respect to the control, and with treatments applied in strips difference between them. However, where each treatment delivered a four mid-rows wide (ie 12m). The subtraction of the control map greater benefit than the other; here experimental treatments used were from those for ripping and mulch ‘benefit’ is defined in terms of a yield aimed at increasing access to stored reveals that the response to the difference of 0.5t/ha or greater. soil moisture. They comprised of imposed treatments was highly The areas in Figure 6f where ripping ripping the wheel tracks to a depth variable, delivering both benefits is recommended are those where of 40mm, and use of a composted and penalties with respect to yield there was both at least a 0.5t/ha grape marc applied as a surface in excess of 1.5t/ha (Figures 6d-e). yield benefit from ripping compared mulch at a rate of 150m3/ha. The The final map 6f indicates where the to the control (Figure 6d) and at ripping was intended to improve greatest benefit will be gained from least a 0.5t/ha greater yield root penetration into the mid-row; applying either or both treatments. following ripping compared to the mulch was intended to suppress mulch (Figure 6e). Given the large difference between evaporation of soil water. (Figure 5). the cost of ripping ($185/ha) and Of course, Figure 6f may be Figure 6 shows the results obtained applying mulch ($2,500/ha), the constructed with respect to any from this experiment in 2004. The manager had a strong preference yield benefit of the manager’s highly replicated design coupled with for ripping and a desire to only use choosing. This is important because the use of a yield monitor at vintage mulch where it was likely to deliver a the statistics of classical experiments

Figure 5. Design for a whole-of-block experiment used to explore options for improved soil management in a 6.8ha Langhorne Creek vineyard. Use of such a highly replicated design, coupled with GIS-based analysis, allowed yield maps to be produced for each treatment over the whole of the block and promotes a spatially based analysis of results.

22 Precision Ag News determine whether or not a than is possible in the case of more Research - viticulture treatment delivers a benefit on the traditional experimental approaches. basis of the statistical significance However, arguably the most Acknowledgments of the difference between the important thing to emerge from this This work was funded by CSIRO, treatment and the control. Whilst research is the fact that the vineyard Vasse Felix Wines, Southcorp this is mathematically robust, I have manager had no problem either Wines (now Foster’s Wines) yet to meet a vineyard manager with the idea of giving the whole and Australia’s grapegrowers (or any other kind of farmer) who block over to experimentation or and winemakers, through their makes their decisions on the basis with the pragmatics of implementing investment body, the Grape and of statistical significance. More the treatments. The managers Wine Research and Development typically, the decision as to whether Corporation. The invaluable input of several other blocks in which a new practice should be adopted is of colleagues at CSIRO, Vasse whole-of-block designs have been made on the basis of considerations Felix and Foster’s is gratefully used for experimentation have such as the magnitude of the acknowledged. been similarly comfortable with the response (eg t/ha), the benefit:cost idea. In other words, even in the Note: This article is based on trade-off, or whether the benefit is absence of variable rate application, a paper presented at the 11th large enough to justify the additional these managers have not seen the Romeo Bragato Conference, held effort required in doing something complex nature of the experimental in Gisborne, New Zealand, 25-27 new, among a whole raft of other design as an impediment to its August, 2005. possible considerations. implementation. Further work For more information Another benefit of note in this being done by Kerstin Panten seeks Dr Rob Bramley (08) 8303 8594, approach is that through the use to explore improved designs for [email protected] of an indicator variable, in this case such experiments. Kerstin will be or visit clay content, the results can be more reporting on this work in a future www.csiro.au/csiro/content/ robustly translated to similar areas issue of Precision Agriculture news standard/ps1xy,,.html

Figure 6. Variable yield response of 6.8ha of Shiraz to two soil amelioration treatments.

Precision Ag News 23 We’re as precise as this!

10cm Accuracy

1800 062 221 www.omnistar.com.au

OMNI239126-FP-Precise.indd 3 3/11/06 4:10:46 PM