Modern vineyard managers need to consider a wide range of factors when making decisions on practices such as pruning techniques, irrigation, canopy management, and harvest date, to name but a few. The wine producing areas of South Africa pose a unique challenge with high variability in climatic conditions, topography, soil types and management practices. Several tools are available to the producer to aid in decision-making, including satellite and aerial photography derived products, as well as near real time technologies implemented in the vineyard. Managers and producers can benefit greatly from plant (growth), soil (water) and nutrient data products derived from remote sensing data collected throughout the season.
Two complementary projects are currently focussing on deriving spatial data products. A Winetech funded project aims at integrating and validating geographical information systems (designed to capture, analyse, manage and present different layers of geographical and other data), thermal remote sensing and climate data, with data collected in Cabernet Sauvignon and Shiraz vineyards (including transpiration, soil temperature, canopy growth and grape ripening). In addition, the “FruitLook” project, funded by the Western Cape Provincial Department of Agriculture and conducted together with eLeaf, provides spatial maps on growth, water and nutrients through an operational platform for vineyards and deciduous fruit produced in the Western Cape through the portal www.fruitLook.co.za.
In this article we evaluate the validity and interpretation of spatial data sources in relation to field data as generated through these projects.
Study areas and selected fields
FruitLook has been providing spatial data products related to growth, water and nutrients for most of the grape and deciduous fruit producing areas of the Western Cape since 2011 (Figure 1). To evaluate the accuracy of these and other spatial data products available to the wine and fruit industries, three Cabernet Sauvignon vineyard blocks were selected and studied in the 2012/13 growing season. These vineyards are situated in the Stellenbosch, Somerset West and Elgin regions and represent different climatic conditions.
The Somerset West and Stellenbosch study sites are classified as temperate to warm and the Elgin site as temperate. The vineyards have normal VSP training systems and normal viticulture practices were followed at each of these blocks, with spur pruning and basic canopy management throughout the growing season. A permanent cover crop is present at the Elgin site, while the other two vineyards only had a winter cover crop (oats/barley) that was killed in the beginning of the summer. Drip irrigation is present in the Stellenbosch and Elgin vineyards and a drip system was installed towards the end of the 2012/13 season in the Somerset West vineyard.
Within each of the three vineyards, areas with different growth vigour (low to high) were selected. The NDVI (Normalised Difference Vegetation Index) data derived from multispectral aerial photographs in January 2013 (at 0.5 m resolution) confirmed the variations in growth vigour in each vineyard (Figure 2).
At each of the vigour areas selected, 15 vines were demarcated from high resolution aerial photography (NDVI images) for detailed growth measurements, which commenced in October 2012. In addition to growth measurements, stem water potential, soil moisture, stomatal conductance, canopy light interception and leaf chlorophyll content were monitored at the three vineyards throughout the 2012/13 season, with measurements performed around key phenological stages. In this article we focus on data collected at the Somerset West vineyard during the 2012/13 season, with reference only to the other two sites and previous seasons.
Climatic data obtained from the Agriculture Research Council and the Hortec weather station networks. Tinytag sensors monitored in-vineyard ambient temperature and relative humidity at different growth vigour areas. In each of the treatment blocks, canopy temperature and relative humidity were measured in the bunch zones.
The 2012/13 growing season was slightly cooler and wetter than the 2011/12 season (Table 1 & 2). Growing degree days were 1 886 and the annual and seasonal rainfall was 837 mm and 328 mm, respectively. About 40% of the annual rainfall occurred within the growing season. This was the wettest year compared to the previous five years. Rainfall recorded at the Stellenbosch and Elgin vineyards in the 2012/13 season was above average with 1 452 and 1 199 mm recorded, respectively.
The average temperature recorded around bud break at the beginning of November (DAB = 2) was 11’C and seasonal daily maximum temperatures exceeding 35’C were recorded in December 2012 (DAB = 80) and again in March 2013 (DAB = 156) (data not shown). Mean February temperature (MFT) for the season was 21.8’C (long term MFT = 21.05’C), compared to the mean December temperature of 22.9’C.
Daily average wind speeds varied from 0.5 – 16 m/s in the month of November 2013. High wind speeds on 30 November 2012 recorded throughout the Western Cape caused widespread damage to the vineyards, including the Somerset West vineyard monitored. Here a maximum wind speed of 19.7 m/s was recorded.
Shoot growth and biomass production
Growth measurements were focused around important phenological phases. Weekly measurements were however performed until vegetative growth ceased. Shoot growth was monitored in the selected 15 vines in each subplot. Once the shoot length exceeded 5 cm, only selected shoots were monitored further, by way of monitoring shoot length, the number of nodes and the shoot plastochron index. Leaf area was measured destructively and grape ripening was monitored at about weekly intervals by way of triplicate sampling in each of the three vineyards and their treatments. Measurements included total soluble solids, pH, titratable acidity and berry mass and volume. Grape yield per vine and bunch number were determined at harvest. Detailed cane measurements were also performed during winter pruning on selected canes, including total cane mass and number of canes.
The key phenological stages achieved at the Somerset West vineyard are shown in Table 1. Typical GDD were estimated for the key stages and it is shown that most of the phenological stages were reached slightly later than average. Long term data sets for the Somerset West vineyard are not available yet, as detailed monitoring of the site started in 2012/13. With a longer record of phenology and climatic conditions, comparison between years and sites will be possible with normalised models such as the “precocity index” (index of relative earliness) and this will allow us to investigate the impact of climate and site conditions on phenology.
Primary shoot growth was monitored at regular intervals for the different vigour sites within the Somerset West vineyard, with longer shoots and faster growth in both the medium and high vigour areas compared to the low vigour plot (Figure 3). Shoot growth was plotted with biomass production (dry biomass) estimated from the FruitLook data for the same plots (Figure 3). Several interesting observations can be made. Firstly, there seems to be a good correspondence in the general slopes of the low and high vigour shoot growth and FruitLook biomass production curves. Shoot growth generally progressed strongly until about 50 DAB when it weakened in reaction to a few days with low mean temperature. The reduction in shoot length (means) between 57 and 65 DAB is due to many shoots being damaged by the gale force winds, forcing reselection of generally slightly shorter shoots. After about 80 DAB primary shoot growth ceased, but not biomass production. This is possible, considering that secondary shoot growth is still taking place, especially in the higher vigour plots.
In order to obtain a better idea of the correspondence between the FruitLook and field biomass data, the yield measured per grapevine and total pruning mass were combined into a “derived total biomass” parameter in order to compare it to the satellite data for accumulated biomass. It has to be considered that the field data is a combination of wet (fruit yield) and dry (cane mass) parameters, and that it does not include sub-canopy (weeds and cover crop) biomass. FruitLook biomass data present total above plus below ground biomass production within a pixel. Nevertheless, a good correspondence can be seen between the ratio of the low to high vigour biomass, indicating that the satellite data is able to convey the differences seen within vineyard blocks quite effectively. From Table 3 it is clear that vigour has a clear effect on pruning mass, as expected for the Somerset West and Stellenbosch sites, but not for the Elgin site. The latter was caused by plots not being laid out using late-season imaging, where the site differences become clear. This was addressed in the recent season by adding a low vigour site. The Somerset West sites showed the largest extremes for both vigour and pruning mass, but not the highest biomass according to satellite data. The highest accumulative biomass values were indicated at the Elgin site, which is probably the result of the permanent cover crop present in every alternate row.
Continuous evapotranspiration (ET) measurements from the Somerset West vineyard started on 19 December 2012. An OPEC eddy covariance (EC) system was installed in the middle of the vineyard (Figure 4) and data collected is representative of the entire vineyard. This system estimates ET not from the soil water balance, but directly from fluctuations in vertical wind speed and water concentration and indirectly by solving the surface energy balance, all at 30 minute intervals (Klaasse & Jarmain, 2012).
FruitLook data (growth, water and nutrient related) is estimated at a 30 m spatial resolution and weekly temporal resolution. Data from a number of satellites (UK-DMC 2 and DEIMOS-1 satellites of the DMC, VIIRS, MSG ) is combined with field weather data (NOAA, GSOD, ARC ) in a number of different algorithms (SEBAL, METEOLOOK, NLook and others) to provide the data. Both the eddy covariance and FruitLook ET data represent water losses from a surface, including transpiration (vines, weeds or cover crop), and soil evaporation of intercepted water.
Weekly ET from the Somerset West vineyard extracted from FruitLook, showed an increase in ET from 15 mm/wk (DAB = 7) to reach a maximum rate of 38 mm/wk in December 2012 (DAB 73), the same as the reference evapotranspiration (ETo) at the time (Figure 6). This was followed with a period of lower ET during January and February 2013 (DAB 87 – 122) with weekly ET (26 – 28 mm) about 35% lower than the reference ET recorded (40 mm). The ET again increased towards the end of February to 32 mm/wk (DAB 150), whereafter it showed a continued decrease towards the end of the season (9 mm/wk around DAB 123). The ET data followed changes in solar radiation and vapour pressure deficit closely throughout the season (data not shown).
The FruitLook and measured ET generally agreed well, with measured ET exceeding the FruitLook ET estimates by 1 – 10 mm/wk in summer (DAB 87 – 129), whereas in autumn the ET measured was occasionally lower than the FruitLook ET estimates (DAB >178).
FruitLook captured the in-field variation in ET within the Somerset West vineyard throughout the season (Figure 6), with higher weekly estimates of ET corresponding to the high growth vigour area and lower weekly ET estimates at the lower growth vigour area (Figure 2 & 5). Seasonal ET estimates (659 and 752 mm for the low and high growth vigour, respectively) varied by approximately 12% between the low and high vigour areas.
This trend persisted if the data of more than one year was considered. The FruitLook ET data over two seasons (2011/12 and 2012/13) shows that the higher ET estimates consistently corresponded with higher biomass production related to the higher vigour areas delineated. The seasonal ET (October – April) ranged between 394 and 567 mm in 2011/12 and 650 and 752 mm in 2012/13 across the vineyard (Figure 7), corresponding to more biomass and rainfall (Table 2) during the 2012/13 season.
Interesting to note is that the vineyard blocks studied in Stellenbosch and Elgin showed similar trends in the relationship between seasonal ET and biomass production. However, the seasonal ET from the vineyard block in Elgin was found to exceed that of the other two vineyards, despite the cooler climate and can be ascribed to the presence of a permanent, well-established cover crop, which contributed to the higher ET estimate.
The ET data from FruitLook and that measured using the eddy covariance system, were also compared to long term ET data collected by Myburgh (2013) from a Cabernet Sauvignon block. Myburgh (2013) estimated ET from 2009 – 2013 from a vineyard with an aggressively growing summer cover/interception crop (Babala) (except for during 2009/10), which was irrigated frequently and situated in a warmer climate. It was found that the daily ET increased from <2 mm in spring (before bud break) to maximum seasonal estimates in summer of 7 – 9.3 mm, whereafter the ET decreased to estimates of around 1 mm/d in winter. During autumn estimates of 3 – 5 mm/d were found (Figure 8).
The FruitLook ET data for 2012/13, converted to daily ET estimates, was generally slightly lower than the Myburgh (2013) estimates. The ET estimates in October ranged from 1.5 – 3 mm/d (DAB >2), during maximum summer of 4.4 – 7.2 mm/d [lower than what Myburgh (2013) found] and during autumn 0.4 – 1.5 mm/d. The very different vineyard conditions probably contributed to the differences in the ET estimates.
ET deficit and stem water potential
Detecting stress in grapevines is important, especially in the period between véraison and harvesting. Stem water potential is typically measured to monitor this. The FruitLook ET deficit data possibly represent a useful alternative, representing an ET shortfall from the potential ET from the specific vineyard.
Less negative stem water potentials (-750 to -090 kPa) was measured at the high growth vigour area in the Somerset West vineyard compared to the low vigour area (-815 and -1 327 kPa). The highest water potentials were measured during the measurement cycle about two weeks following véraison. For the corresponding period, the weekly ET deficits from FruitLook remained less than 1mm/wk for the 2012/13 season, suggesting no (water) stress was experienced by the vines (data not shown). ET deficit in the 2011/12 season was slightly higher, with a seasonal total of 16 – 20 mm, but still with a maximum weekly ET deficit of only 4.5 mm/wk, indicating no water stress and an ET shortfall of <1 mm/d during that period.
The Stellenbosch and Elgin vineyards showed low ET deficits throughout the season, with the exception of one week in the Stellenbosch vineyard where ET deficits of 13.5 and 14.5 mm/wk were estimated (data not shown). This equates roughly 2 mm/d ET shortfall and was estimated around véraison. The highest stem water potentials were measured towards the end of the season (-1 300 to -1 500kPa), shortly before the grapes were harvested.
The low ET deficits and relatively high (less negative) stem water potentials measured during the field campaigns suggest that the vines experienced little (water) stress in 2012/13 and that no supplementary irrigation was required during 2012/13 at any of the sites. ET deficit could potentially provide an indication of water stress development over time. The spatial dimension of the ET deficit data set can also provide insights on stress variations across a vineyard over time.
In this research, we compared field data related to growth and water with satellite-derived data across climatic regions and growth vigour variations in three Cabernet Sauvignon vineyards. We showed that spatial data provided through the FruitLook portal and derived from aerial photography corresponds well to field conditions.
Growth data from field measurements, as well as satellite derived data, were shown to correspond well with regards to shoot growth data. Further work will involve using the data of other sites to investigate further the effect of the cover crop and the cultivar (Shiraz vs. Cabernet Sauvignon), as well as to incorporate other parameters measured in the field, such as leaf area, chlorophyll content and ripening parameters.
The ET estimates from the FruitLook portal agreed to within 17 – 25% of ET measured. The ET deficit showed agreements with stem water potential measurements and highlighted the small ET shortfalls over the season in areas which are typically only supplementary irrigated. However, longer term data sets are needed to draw definitive conclusions.
The data sets discussed here need to be explored further for specific applications, like winter pruning (timing), canopy management (when to top), water management (when to irrigate) and nitrogen management (when to fertilise). Assessments are also needed across a wider range of cultivars and in drier grape producing regions.
Vineyard growth vigour: The spatial data products derived from satellite data provide a source of data that could in future be used to monitor the growth vigour throughout the season, which may further be related to the vegetative:reproductive ratio, disease/virus incidence/expression or nutrient/grapevine water status relations.
Vineyard irrigation: The 2012/13 season was relatively cool and wet. Irrigation seemingly did not play such an important role at the sites that were studied. For drier and warmer conditions, including heat waves, the ET and ET deficit data could be used to optimise irrigation applications. The spatial nature of the data may assist in the division/demarcating of vineyards into different irrigation blocks, in this way facilitating more efficient water management.
Future work (FruitLook 2013/14)
The findings reported here specifically illustrate the potential of growth, NDVI, evapotranspiration and ET deficit data products derived from remote sensing data. The measurements will be extended to drier climates where the ET deficit is expected to be larger. The spatial nitrogen data provided through FruitLook will also be further investigated and other cultivars, such as Shiraz, exhibiting different physiological behaviour compared to Cabernet Sauvignon, will be included.
Two students have been involved in the work reported. Tara Southey (née Mehmel) is registered for a PhD (Agric) in Viticulture and Christo Kotze is registered for an MSc (Agric) in Viticulture.
We would like to thank the Western Cape Provincial Department of Agriculture and Winetech/THRIP for funding the projects.
We would also like to thank Ruben Goudriaan and colleagues (eLeaf) for producing all FruitLook data and extracting information for each treatment from the vineyard blocks studied.
We would also like to express our gratitude to the management of the Vergelegen, Thelema and Sutherland farms for allowing us to conduct research on their farms.
Technical assistance from Talitha Venter and Leonard Adams, Department of Viticulture and Oenology, is appreciated, as well as that of other students involved in some of the measurements.
Both the KwaZulu-Natal University and Stellenbosch University are thanked for supporting this research.
Goudriaan, R., 2013. FruitLook: An operational service to improve crop water and nitrogen management in grapes and other deciduous fruit trees using satellite technology for the season of 2012-13. Final report to Director: Sustainable Resource Management, Western Cape Department of Agriculture, 63 pages.
Klaasse, A. & Jarmain, C., 2012. FruitLook: An operational service to improve crop water and nitrogen management in grapes and other deciduous fruit trees using satellite technology for the season of 2011-12. Final report to the Western Cape Provincial Department of Agriculture, 98 pages.
Myburgh, P., 2013. The impact of waste water irrigation by wineries on soils, crop growth and product quality. Progress report to the Water Research Commission on Project no. K5/1881, September 2013, 58 pages.
A collaborative project was launched to integrate satellite and field data of grapevine growth and water status with the purpose of validating currently available products. Agreement between shoot growth and the biomass indicators from satellite data could be seen, with good distinction between the major delineated vigour areas. Biomass production curves from FruitLook showed different phenological stages clearly, with the onset of growth agreeing with high biomass production and a reduction in growth following véraison. The FruitLook evapotranspiration (ET) compared well with that measured, again showing differences across delineated vigour areas. FruitLook evapotranspiration from a site with permanent cover crop was higher than other study sites over the season, as expected. The effects of different elements of vineyards management need to be further considered. There was some agreement between the ET deficit estimated using satellite data and the plant water status. For the studied vineyards ET deficits were very low for the years considered, corresponding to limited plant water deficits as a result of above average rainfall years. For the sites considered in this pilot study the FruitLook growth and water status data accuracy is acceptable, when data is assessed in a qualitative and quantitative way. Specific ways of integrating spatial FruitLook and other spatial data with field based data (e.g. soil moisture) need to be shown before adoption will take place within the industry.
For more information contact Albert Strever at firstname.lastname@example.org, Caren Jarmain at email@example.com, Tara Southey at firstname.lastname@example.org or Kobus Hunter at HunterK@arc.agric.za.
Albert Strever1, Caren Jarmain2, Tara Southey1 & Kobus Hunter1,3
1 Department of Viticulture, Stellenbosch University
2 School of Agricultural, Earth & Environmental Science, KwaZulu-Natal University
3 ARC Infruitec-Nietvoorbij, Stellenbosch
1 DMC = Disaster Monitoring Constellation, VIIRS = Visible Infrared Imaging Radiometer Suite on board the National Polar-orbiting Operational Environmental Satellite, MSG = Meteosat Second Generation.
2 NOAA = National Oceanic and Atmospheric Administration, GSOD = Global Summary Of the day, ARC = Agricultural Research Council.