Irrigation management zones for precision according to intra-field variability

J.A. Martínez-Casasnovas, D. Vallés Bigorda and M.C. Ramos University of Lleida, Department of Environment and Science, Av. Rovira Roure 191, 25198 Lleida, Spain; [email protected]

Abstract

The present research shows a case study in precision viticulture to improve irrigation. The research was carried out in a field located in Raimat (NE Spain). This is a semi-arid area with continental Mediterranean climate and a total annual precipitation between 300-400 mm. The field (4.5 ha) is planted with Syrah vines in a 3×2 m pattern. The vines are irrigated by means of drips under a partial root drying schedule. The irrigation sectors have a quadrangular distribution, with a size of about 1 ha each. presents a coefficient of variation of 32.2%, with an average yield of 6.9 t/ha. The re-design of the irrigation sectors was based on multi-variant statistics analysis of soil properties (pH, electric conductivity, organic matter content, calcium carbonate content, water retention availability for plants, texture and multi-temporal profile water volumetric content). The sampling density for these properties was of 8 samples/ha. Other data used in the analysis were the normalized difference vegetation index (NDVI) from Quickbird-2 satellite images of the years 2004 to 2007, and yield data acquired from a Canlink 3000 Farmscan monitor for the years 2004 to 2006. The results show that the best spatial prediction of yield is mainly explained by the NDVI of images acquired during and the sand content. It explains 85.6% of variance. NDVI variability is mainly explained by the volumetric water content of the soil profile, explaining 72.1% of its variance. These properties: average yield, average NDVI, average volumetric water content and sand content, were used in a cluster analysis performed by means of the ISODATA algorithm implemented in Image Analyst for ArcGIS 9.1 to distinguish two management zones within the field. Those zones were finally used to re-design the irrigation sectors to adapt them to the variability expressed in the two zones.

Keywords: vineyard, NDVI, yield, soil properties

Introduction

The application of (PA) techniques in viticulture is relatively recent. The first results began to be published from projects initiated in Australia in the wake of the appearance on the market of yield sensors and monitors (Bramley and Proffitt, 1999). Since then, a wide number of experiences and applications have been developed in this field, demonstrating that variable-rate application of inputs and selective harvesting at parcel level can be productive strategies which can provide significant benefits for winegrowers. Vineyard variability is a known phenomenon of which viticulturists are generally well aware, understanding that vine performance varies within their (Bramley and Hamilton, 2004). Recently, the development of the spatial information technologies tools in the last decades (geographical information systems, remote sensing, global position systems and electrical conductivity sensors, among other) and the advent of yield sensors and monitors has allowed obtaining information on vine performance as well as soil variability across the vineyard fields (Proffitt and Malcom, 2005; Proffitt et al., 2006). In this respect, variation in fruit and quality

EFITA conference ’09 523 has been the focus of much of the work currently undertaken (Bramley and Lamb, 2003). The results of some of those research works have demonstrated that variation in yield as well as in fruit quality exhibit marked spatial structure, but that the patterns of variation are not necessarily the same. Nevertheless, due to the absence of an on-the-go sensor to monitor fruit quality parameters, in the same line as the existing yield monitors, it is suggested that zonal management to differentiate grape quality could proceed on the basis of zones of characteristic yield productivity (Bramley, 2005). This system of differential management has been referred to as zonal vineyard management (Bramley, 2005). Several examples of the commercial implementation of this approach to improve the uniformity of fruit parcels delivered to the have been already demonstrated (Proffitt and Pearse, 2004; Bramley et al., 2005). Mainly, those applications are addressed to selective harvesting, since the actuation in the fields to diminish crop variability are difficult because it is mainly related to soil property differences, which are difficult to change (Bramley, 2001). Other experiences to improve labour at or to yield forecasting have also been reported (Martínez-Casasnovas and Bordes, 2005; Proffit and Malcom, 2005). Nevertheless, some experiences have been specifically addressed to apply cultural practices differentially, as for example irrigation water, with distinct amounts in different management zones along the growing season (Proffitt and Pearse, 2004; Proffit and Malcom, 2005). In this reported case study in the Margaret river region (Proffit and Malcom, 2005), irrigation was managed differentially in high vigorous areas with respect to less vigorous areas, so that water was restricted in the vigorous areas in order to reduce vegetative growth. The application of less water during the season appeared to reduce vegetative growth, with the greatest decrease in surface area being recorded in the most vigorous areas. In view of this background, and according to the interest of precision viticulture as a tool for improving the efficient use of inputs and for diminishing vigour crop variability across the vineyard fields to deliver a more uniform output to the winery (in terms of yield as well as of fruit quality), the objective of the present work is to present a case study in precision viticulture to re-define irrigation management zones according to intra-field variability in a commercial vineyard block located in Raimat (NE Spain).

Material and methods

Study area The research was carried out in a commercial vineyard field located in Raimat (Lleida, NE Spain) (X= 292420, Y= 4614740, UTM 31n). It is included in the Costers del Segre Designation of Origin. This is a semi-arid area with continental Mediterranean climate and a total annual precipitation between 300-400 mm. The field, with an extension of 4.5 ha, is planted with Syrah vines in a 3×2 m pattern. The vines are irrigated by means of drips under a partial root drying schedule. At present, the irrigation sectors are have a quadrangular distribution, irrigating a homogeneous area of about 1 ha each. The sectors were designed previous to the plantation in 2002, without having into account the possible spatial variability of soil characteristics. At present, yield in this field presents a coefficient of variation of 32.2%, with an average yield of 6.9 t/ha.

Spatial variability of yield, vigour and soil properties The re-design of the irrigation sectors was based on multi-variant statistics analysis of yield, plant vigour, through the normalized difference vegetation index – NDVI (Rouse et al., 1973) and soil properties. Yield data was acquired from a Canlink 3000 Farmscan monitor for the years 2004, 2005 and 2006. For each year, yield maps were produced following the protocol of Bramley and Williams (2001).

524 EFITA conference ’09

The re-design of the irrigation sectors was based on multi-variant statistics analysis of yield, plant vigour, through the normalized difference vegetation index – NDVI (Rouse et al., 1973) and soil properties.

Yield data was acquired from a Canlink 3000 Farmscan monitor for the years 2004, 2005 and 2006. For each year, yield maps were produced following the protocol of Bramley and Williams (2001). Data refinement involved normalising the data (μ = 0, s = 1) after removal of data recordsData with refinement zero yield involved or GPS erronormalisingrs, and thenthe data removing (μ = 0, recordss = 1) after for removal which the of data records with zero normalised yieldyield was or greater GPS errors,or less andthan then ±3 st removingandard deviations records fromfor which the mean. the normalised The resulting yield was greater or less yield data werethan used ±3 to standardinterpolate deviations 3 m grid byfrom local the block mean. kriging The resulting (10 m x10 yield m blocks) data were using used to interpolate 3 m VESPER (Minasnygrid byet al. local, 2005). block kriging (10×10 m blocks) using VESPER (Minasny et al., 2005). In addition, three Quickbird-2 satellite images where acquired and processed for plant vigour In addition, three Quickbird-2 satellite images where acquired and processed for plant vigour monitoring. The dates of images acquisition were: 29-07-2004, 13-07-2005 and 13-07-2006. They monitoring. The dates of images acquisition were: 29-07-2004, 13-07-2005 and 13-07-2006. They are withinare the within range theof ±2range weeks of ±2 the weeks mome thent of moment veraison, of whichveraison, has which been referred has been to referred be to be the optimal the optimal timetime for for image image ac acquisitionquisition in in precisi precisionon viticulture viticulture applications applications (Lamb (Lamb et et al. al.,, 2004). The spatial 2004). The spatialresolution resolution of the of multi-spectral the multi-spectral images images was 2.8was m. 2.8 The m. images The images were werecorrected for atmospheric corrected for atmosphericscattering by scattering applying bythe aCOSTpplying model the COST proposed mode byl proposedChavez (1996). by Chavez Then, digital values were (1996). Then, converted digital values to reflectance were converted according to to re flectancethe radiance according conversion to theof Quickbird-2 radiance data technical note conversion of Quickbird-(Krause, 2003).2 data Aftertechnical this note process, (Krause, the images2003). After were this projected process, to the the images European Datum 1950 and were projected to the European Datum 1950 and the UTM 31n coordinate system. To improve the positionalthe UTM accuracy, 31n coordinate the project system.ed images To improve were ortho-rect the positionalified based accuracy, on: a) thea set projected images were of ground controlortho-rectified points collected based from on: a (a)0.5-m a set resolution of ground ort controlho-photo points produced collected and b)from a 5 ma 0.5-m resolution ortho- resolution digitalphoto elevation produced mode andl, both(b) a produced5 m resolution by the digital Cartographic elevation Instit model,ute ofboth Catalonia. produced by the Cartographic The normalizedInstitute difference of Catalonia. vegetation The index normalized (NDVI) wasdifference computed vegetation according index to (NDVI)Equation was 1 computed according (Rouse et al., 1973).to Equation 1 (Rouse et al., 1973).

ϕ −ϕ NDVI = NIR RED (1) ϕ +ϕ Equation 1 NIR RED

where φNIR and φRED are the spectral reflectance measurements acquired in the near-infrared (760- ϕ ϕ where NIR and900 RED nm) are and the red spectral (630-690 reflectance nm), respectively. measurements These acquired spectral in the reflectances near-infrared are themselves ratios of (760-900 nm) andthe incomingred (630-690 radiation nm), respectively. that is reflected These in speachectral spectral reflectances band individually, are themselves hence they take on values ratios of the incomingbetween radiation 0.0 and that1.0. is By reflected design, in the each NDVI spect itselfral band thus individua varies betweenlly, hence -1.0 they and +1.0. take on values Variationbetween 0.0 in soiland properties1.0. By design has been, the recognizedNDVI itself to thus be ava keyries driver between of vineyard -1.0 and variability (Brambley +1.0. and Lamb, 2003). This has raised the question as how vineyard should be surveyed. In the Variation in soilpresent properties research, has beensoils were recognized surveyed to be on a a key regular driver grid of each vineyard 10 row variability × 20 stocks (30×40 m). This (Brambley andrepresents Lamb, 2003). a density This ofhas about raised 8 samples/ha,the question which as ho wis almostvineyard five soils times should more be than the practical use in surveyed. In theother present viticulture research, regions soils (Brambley were surveyed and onLamb, a regular 2003). gr Inid this each grid, 10 rowsoil samples x 20 of the top horizon -1 stocks (30 x 40(0-20 m). Thiscm) wererepresents taken a and density analysed of about in the 8 sampleslaboratory. ha In, whichtotal, 35is almostsoil samples five were analysed. The times more thancharacterised the practical propertiesuse in other were viticultur pH, electrice regions conductivity, (Brambley anorganicd Lamb, matter 2003). content, In calcium carbonate this grid, soil samples of the top horizon (0-20 cm) were taken and analysed in the laboratory. content, water retention availability for plants and texture. Also the volumetric soil moisture at In total, 35 soil samples were analysed. The characterised properties were pH, electric conductivity, organicdifferent matter depths conten (eacht, 20calcium cm, up carbonate to 80 cm content, or up to water a contrasting retention layer) availability was measured using a TDR for plants and texture.sensor (TimeAlso the Domain volumetric Reflectometer) soil moisture atwith diffe a tuberent depthsprobe. (eachTube 20probes cm, upwere to permanently installed 80 cm or up in to the a contrasting sampling points layer) to was facilitate measured multi-temporal using a TDR measures, sensor (Timewhich Domainwere taken monthly between Reflectometer) November with a tube 2006 prob e.and Tube December probes were2007. permanen Once a tlypreliminary installed indata the analysissampling (central tendency and points to facilitatedispersal multi-temporal statistics) wasmeasures done, andwhich data were normality taken monthlyverified, betweenmaps for November all sampled variables (including grape yield) were obtained with VESPER (Minasny et al., 2005) geostatistical analysis software using the parameters as proposed by Arnó et al. (2005). From the measured values at the grid points 3 the experimental variograms were computed and modelled and global block kriging (10 m squared blocks) undertaken with a grid spacing of 3×3 m.

Statistical analysis and irrigation sector’s delineation Yield, NDVI and soil properties data were statistically analysed at the sample grid points using Statgraphics Plus 5.1. Previously, average yield and average NDVI values were captured and associated to the sample points using ArcGIS 9.1 tools. Different types of statistical analysis were performed: correlation matrix among the variables and stepwise multiple regression having,

EFITA conference ’09 525 respectively, average yield and average NDVI as dependent variables. The objective of these analyses was to test the influence of soil properties on yield and vigour variability and determine the variables that could be useful to establish zones for differential management to improve irrigation in the vineyard block. This spatial zoning was performed by means of a cluster analysis using the ISODATA algorithm implemented in Image Analyst extension for ArcGIS 9.1. From a set of input variables, the ISODATA is a K-means algorithm for clustering that uses minimum distance to assign a cluster to each candidate pixel in an iterative process (Jensen, 1996). It removes redundant clusters or clusters to which not enough samples are assigned. It needs that the initial number of clusters is defined. In the present case study the target clusters to compute were considered two or three, according to previous experiences of definition of management zones in other study areas (e.g. Bramley and Hamilton, 2004; Proffitt and Malcom, 2005).

Results and discussion

Spatial variability of yield, vigour and soil properties Figure 1 shows the spatial variability of the average yield and average NDVI computed from the respective data acquired in 2004, 2005 and 2006 for the case study vineyard block. The average yield is 11.0±2.7 Mg/ha, with an average coefficient of variation of 24.9% (range 24.0-35.4%); while (R2= 0.12, p-value < 0.05), calcium carbonate content (R2= 0.14, p-value < 0.05) and the average NDVI is 0.29±0.06,2 with a coefficient of variation of 20.7% (range 16.6-25.8%). These coefficientselectrical conductivity express the (Rexistence= 0.14, of p-value an important < 0.05). heterogeneity However significantof yield and co vigourrrelation (at wasveraison) not withinfound thebetween vineyard the blockaverage and NDVI also theand persistence soil texture or components. stability of theThe variation available along water the content years, was not significant correlated either with the yield or with the NDVI, as found by Bramley such as other researchers have found in other study areas (e.g. Bramley and Hamilton, 2004), and (2001), Bramley and Hamilton (2003) or Proffitt and Malcom (2005), among other. couldNevertheless, be indicating this thecould opportunity be explained for site-specific by the lack ofcrop soil management water content according data for to theits variability.horizons below the top soil, which were not sampled.

FigureFigure 1. 1. Average Average NDVI NDVI at veraisonat veraison and and average average yield yield (years (years 2004, 2004, 2005 2005 and 2006)and 2006) of the of Syrah the vineyardSyrah vineyard block located block located in Raimat in Raimat(NE Spain). (NE Spain).

Regarding the multiple regression analysis between the average yield and average NDVI as 526dependent variables, the results show that the best spatial predictionEFITA of conference yield is mainly ’09 explained by the average NDVI from the multi-spectral images acquired during veraison and the sand content: avgYield = 7.18 – 0.20 %Sand + 33.0 avgNDVI, (R2= 85.6%, p-value <0.01). Then, the volumetric soil moisture content (SMC), measured in the whole profile by means of TDR, is the soil property most correlated with the yield, explaining almost 70% of their spatial variability by itself: avgYield = -6.15 + 0.91 avgSMC (R2= 69.5%, p-value <0.01). This is also the case of the results of the stepwise multiple regression between the average NDVI and the soil properties, which produced the following equation: avgNDVI = - 0.26 + 0.03 avgSMC (R2= 60.9%, p-value <0.01). This relationship is improved if soil texture is considered: AvgNDVI= -0.36 + 0.005 %Sand + 0.026 avgSMC (R2= 72.1%, p-value < 0.01). The results confirm the importance of the soil texture and soil moisture content in the variability of vineyard development and yield and, therefore, the opportunity to actuate differently through vineyard zoning based on soil water status information (Proffitt and Malcom, 205; Acevedo-Opazo et al., 2008) to deliver to the winery a more homogeneous production.

5 Relationship between variables The simple regression analysis among yield, NDVI and soil properties variables reveals the high correlation between yield and NDVI (R= 0.89, P<0.01) and yield or NDVI with the average volumetric soil moisture content of the soil profile (R= 0.83, P<0.01; R= 0.82, P<0.01). These results are in agreement with research by other authors in precision viticulture, which have found significant correlations between yield and the multi-spectral response of the crop in the read and near infrared spectrum (Bramley and Lamb, 2003; Proffitt et al., 2006; Acevedo-Opazo et al., 2008). Regarding simple correlations with soil properties, significant relationships were found between average yield and sand content (R2= 0.12, P<0.05), calcium carbonate content (R2= 0.14, P<0.05) and electrical conductivity (R2= 0.14, P<0.05). However significant correlation was not found between the average NDVI and soil texture components. The available water content was not significant correlated either with the yield or with the NDVI, as found by Bramley (2001), Bramley and Hamilton (2003) or Proffitt and Malcom (2005), among other. Nevertheless, this could be explained by the lack of soil water content data for the horizons below the top soil, which were not sampled. Regarding the multiple regression analysis between the average yield and average NDVI as dependent variables, the results show that the best spatial prediction of yield is mainly explained by the average NDVI from the multi-spectral images acquired during veraison and the sand content: avgYield = 7.18 – 0.20%Sand + 33.0 avgNDVI, (R2= 85.6%, P<0.01). Then, the volumetric soil moisture content (SMC), measured in the whole profile by means of TDR, is the soil property most correlated with the yield, explaining almost 70% of their spatial variability by itself: avgYield = -6.15 + 0.91 avgSMC (R2= 69.5%, P<0.01). This is also the case of the results of the stepwise multiple regression between the average NDVI and the soil properties, which produced the following equation: avgNDVI = -0.26 + 0.03 avgSMC (R2= 60.9%, P <0.01). This relationship is improved if soil texture is considered: AvgNDVI= -0.36 + 0.005%Sand + 0.026 avgSMC (R2= 72.1%, P<0.01). The results confirm the importance of the soil texture and soil moisture content in the variability of vineyard development and yield and, therefore, the opportunity to actuate differently through vineyard zoning based on soil water status information (Proffitt and Malcom, 205; Acevedo-Opazo et al., 2008) to deliver to the winery a more homogeneous production.

Intra-field zones and proposal of irrigation sectors Since the assessment of the spatial variability soil properties to guide irrigation is costly and requires a high density of sampling and monitoring the soil moisture content and, that from the above results it is clear the existence of a significant relationship between yield and NDVI with the volumetric soil moisture content, we propose the clustering of yield maps and/or NDVI maps as an alternative approach to define management zones for irrigation. Figure 2 shows the results of the clustering of the average NDVI, average yield and average soil moisture content (SMC) for the study period. Note that the average SMC was only sampled and interpolated in a sub-block of the vineyard field. From Figure 2 it is clear the correspondence between the clusters calculated from the average NDVI and the average yield across the vineyard block. Also, as confirmed in the statistical analysis, there is a good direct correspondence between the clusters identified with these variables and the ones of the soil moisture content. Therefore, either the average NDVI or the average yield cluster map can be used as basis to re-design the irrigation sectors in the block. A proposal of the re-formulated irrigation sectors is presented in Figure 2. It has been implemented during the 2008 , adapting the irrigation tubes to the new configuration of sectors. According to the estimated evapotranspiration and measured rainfall in the study area, the average irrigation water necessities to satisfy the necessities of the crop were 265 mm/year. This amount of water was supplied to the

EFITA conference ’09 527 Intra-field zones and proposal of irrigation sectors

Since the assessment of the spatial variability soil properties to guide irrigation is costly and requires a high density of sampling and monitoring the soil moisture content and, that from the above results it is clear the existence of a significant relationship between yield and NDVI with the volumetric soil moisture content, we propose the clustering of yield maps and/or NDVI maps as an alternative approach to define management zones for irrigation. Figure 2 shows the results of the clustering of the average NDVI, average yield and average soil moisture content (SMC) for the study period. Note that the average SMC was only sampled and interpolated in a sub-block of the vineyard field.

FigureFigure 2. Clusters2. Clusters formed formed from from the the average average NDVI, NDVI, average average yield, yield, average average soil soil moisture moisture content (SMC)content and (SMC) proposed and irrigationproposed sectors.irrigation sectors.

From Figure 2 it is clear the correspondence between the clusters calculated from the average NDVI and the average yield across the vineyard block. Also, as confirmed in the statistical cropanalysis, in the irrigation there is sectors a good of direct type 2 correspondence while the irrigation between sectors the of clus typeters 1 were identified supplied with with these a supplementaryvariables and amount the ones of of20% the with soil moisturethe objective conten to t.equilibrate Therefore, the either plant th developmente average NDVI and yieldor the withaverage respect yield to the cluster most mapvigorous can bezones. used as basis to re-design the irrigation sectors in the block. A proposal of the re-formulated irrigation sectors is presented in Figure 2. It has been Conclusionsimplemented during the 2008 vintage, adapting the irrigation tubes to the new configuration of sectors. According to the estimated evapotranspiration and measured rainfall in the study Thearea, present the workaverage supposes irriga tiona contribution water necessities to the application to satisfy of th zonale necessities vineyard ofmanagement the crop were to make 265 -1 moremm targeted year . managementThis amount decisions.of water wasThe resultssupplied confirm to the thecrop positive in the irrigationand significant sectors relationship of type 2 betweenwhile the irrigationvolumetric sectors soil moisture of type 1content were suppliedand the yieldwith aand supplementary NDVI, then theam ountcharacterization of 20% with the objective to equilibrate the plant development and yield with respect to the most vigorous of the spatial variability of those parameters (yield or NDVI) are useful to define the vineyard zones. management zones to improve irrigation. The benefits of this targeting management, however, need to be analysed in order to know if it reduces the spatial variability of yield to deliver a more homogeneous fruit to the winery in relation to conventional uniform management. 6 References

Acevedo-Opazo, C., Tisseyre, B., Guillaume, S. and Ojeda, H., 2008. The potential of high spatial resolution information to define within-vineyard zones related to vine water status. PrecisionAgric 9: 285-302. Arnó, J., Bordes, X., Ribes-Dasi, M., Blanco, R., Rosell, J.R. and Esteve, J., 2005. Obtaining grape yield maps and analysis of within-field variability in Raimat (Spain). In: Stafford, J.V. (ed.), Proceedings of 5th European Conf. Prec. Agric., Wageningen Academic Publishers, The Netherlands, pp. 899-906. Bramley, R.G.V. and Proffitt, A.P.B., 1999. Managing variability in viticultural production. Grapegrower and Winemaker 427: 11-16.

528 EFITA conference ’09 Bramley, R.G.V., 2001. Progress in the development of precision viticulture - Variation in yield, quality and soil properties in contrasting Australian vineyards. In: Currie, L.D. and Loganathan, P. (eds.) Precision tools for improving land management. Occasional report No. 14. Fertilizer and Lime Research Centre, Massey University, Palmerston North, pp.25-43. Bramley, R.G.V. and Williams, S.K., 2001. A protocol for the construction of yield maps from data collected using commercially available grape yield monitors. Cooperative Research Centre for Viticulture. Available at: http://www.clw.csiro.au/staff/BramleyR/documents/CRCVYield_Mapping_Protocol.pdf. Accessed 21 November 2008. Bramley, R.G.V. and Lamb, D., 2003. Making sense of vineyard variability in Australia. In: Ortega, R. and Esser, A. (eds.) Precision viticulture. Proceedings of an international symposium held as part of the IX Congreso Latinoamericano de Viticultura y Enología. Centro de Agricultura de Precisión, Pontificia Universidad Católica de Chile, Santiago, Chile, pp. 35-54. Bramley, R.G.V. and Hamilton, R.P., 2004. Understanding variability in winegrape production systems. 1. Within vineyard variation in yield over several . Australian J. Grape & Wine Res. 10: 32-45. Bramley, R.G.V., 2005. Understanding variability in winegrape production systems. 2. Within vineyard variation in quality over several vintages. Australian J. Grape & Wine Res. 11: 33-42. Bramley, R.G.V, Proffitt, A.P.B, Hinze, C.J. Pearse, B. and Hamilton, R.P., 2005. Generating benefits from Precision Viticulture through selective harvesting. In: Stafford, J.V. (ed.), Proceedings of 5th European Conf. Prec. Agric., Wageningen Academic Publishers, The Netherlands, pp. 891-898. Chavez, P.S., 1996. Image-based atmospheric corrections. Revisited and improved. Photogrammetric Engineering and Remote Sensing 55: 339-348. Jensen, J.R., 1996. Introductory Digital Image Processing: A Remote Sensing Perspective. 2d ed. Prentice- Hall, Englewood Cliffs, NJ, USA. Krause, K., 2003. Radiance conversion of Quickbird data – Technical note. DigitalGlobe Inc., Longmont, Colorado, USA. Lamb, D.W., Weedon, M.M. and Bramley, R.G.V., 2004. Using remote sensing to predict grape phenolics and colour at in a Cabernet Sauvignon vineyard: Timing observations against vine phenology and optimising image resolution. Australian J. Grape & Wine Res. 10: 46-54. Martínez-Casasnovas, J.A. and Bordes, X., 2005. Viticultura de precisión: Predicción de cosecha a partir de variables del cultivo e índices de vegetación Revista de Teledetección 24: 67-71. Minasny, B., McBratney, A.B. and Whelan, B.M., 2005. VESPER version 1.62. Australian Centre for Precision Agriculture, The University of Sydney, Sydney, Australia. Proffitt, A.P.B. and Pearse, B., 2004. Adding value to the wine business precisely: using precision viticulture technology in Margaret River. Aust. NZ Grapegrower Winemaker. 491: 40-44. Proffitt, A.P.B. and Malcom, A., 2005. Zonal vineyard management through airbone remote sensing. Aust. NZ Grapegrower & Winemaker 502: 22-27. Proffitt, A.P.B., Bramley, R.G.V., Lamb, D. and Winter, E., 2006. Precision viticulture – A new era in vineyard management and wine production. Winetitles Pty Ltd., Ashford, Australia. Rouse, J. W. Jr., Haas, R. H., Schell, J. A. and Deering, D.W., 1973. Monitoring vegetation systems in the great plains with ERTS. In: Proceedings of the Third ERTS Symposium, NASA SP-351 1, U.S. Government Printing Office, Washington DC, pp. 309-317.

EFITA conference ’09 529