Airborne Remote Sensing for Precision Viticulture in Niagara Ralph Brown School of Engineering University of Guelph
Why the interest in precision viticulture? Highly variable regions in Niagara due to unique geological history and location topography, soil type, micro-climate (terroir) Different areas in same block may differ in vigour, nutrient availability, water status, fruit quality, etc. Can apply spatially-variable management to try to even out production (e.g., fertilizing, irrigation, thinning) or Adapt to variability by managing zones differently and segregating fruit at harvest for unique character reserve wines
Napa, Australia and New Zealand the beginning of remote sensing for viticulture Started with the development of a tool to monitor phylloxera spread ~ 20 years ago Napa work started with NASA (Lee Johnson) - Developed tool to monitor phylloxera spread and found that RS data had other uses: Crop scouting Vineyard management Harvest planning to maximize reserve wine production Commercial RS services began 1999
Spectral differences in grape canopy Typical green vegetation reflectance chlorophyll absorbs at 420, 490 nm, green peak ~ 540-560 nm second chlorophyll trough at 660-680 nm red edge to NIR plateau 700-740 nm water overtone troughs at 1450 and 1940 nm Stressed leaves reflect more strongly than healthy leaves in green-yellow-orange (540-640 nm) and in the red (660-700 nm), lower in NIR Reflectance in spectral bands combined as indices to emphasize soil or vegetation, e.g. NIR+red or NIR+green for Normalized Difference Vegetation Index (NDVI)
Leaf Reflectance spectral differences between phylloxera-infested and healthy vines (CSU 2002) t-value Non-infested Infested 0.9 Mean Reflectance 0.7 0.5 0.3 differences 0.1 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400-0.1 Wavelength, nm
NDVI Green Ratio (R 740 R 550 )/ (R 740 + R 550 ) used to separate healthy from infested vines Non-infested Invested 0.75 0.7 0.65 0.6 Green-Red Ratio 0.55 0.5 0.45 0.4 0.35 0.3 0.25 0 5 10 15 20 25 30 35 40 45 50 Leaf Number
RS images contain other information too! Spatial patterns reveal underlying variability in soil type, moisture, fertility, disease, etc. Time series of images show temporal effects e.g., weather effects (drought), disease spread, insect infestation Spatial information in a geo-referenced image (i.e., image elements tagged with geographic position) useful for determining areas, GPS location in vineyard, etc.
RS images from 4-band CMOS cameras in small aircraft at 3500 ft AGL
Multi-band images co-registered e.g., RGB colour Red band Blue band Green band
NIR, red and green gives colour-infrared composite (CIR) NIR band Green band Red band
CIR highlights vegetative canopy (red) Little canopy Active canopy
Many vineyard blocks show canopy variation, reveal underlying variability of site 1 2 1 2 June 29, 2007 August 28, 2007
Band reflectance in multi-spectral image used to classify and interpret image Red is grape canopy (and trees in bush) Black is soil Green is other vegetation (floor)
NDVI highlights canopy vigour yellow (low) to bright green (high) for one date - August 28, 2007
Change in NDVI shows canopy development from July 20 to August 14, 2007 green = +ve change, blue = -ve change canopy increase
30-Bench Winemakers Project Large Riesling block Divide into zones based on vigour Harvest fruit and vinify separately Determine variability of fruit, wine Stability of zones across years
30-Bench vineyard management zones from 2005 RS images - canopy vigour from NDVI (red) LV1 HV1 LV2 HV2 LE Triangle
Geo-referenced image also contains spatial information e.g., Area of zones? GPS for vines? 7,393 sq m 7,296 sq m 8,660 sq m Length of 1 st row = 233.9 m 4,286 sq m + = GPS coordinates of sentinel vines
Sentinel vines are used to make ground measurements, chart stability of zones Vines are flagged and geo-referenced (GPS) Same vines are revisited year after year Collect canopy, soil, fruit characteristic data Fruit from sentinel vines grouped by water stress for small wine batches Fruit from each management zone harvested separately Winery keeps zone batches separate through process
Ground data collected from sentinel vines Soil moisture from portable TDR Vine water status from pressure bomb Leaf reflectance spectrum Harvest data yield, berry weight, Brix, ph, etc. Sensory and chemical wine data
Measuring leaf reflectance in management zones
2006 cool and wet! Average soil moisture 4780350 48 44 4780300 40 36 4780250 32 28 24 4780200 20 16 4780150 622350 622400 622450 622500 622550 622600 12
Yield per vine (kg) in 2006 season 4780350 12 4780300 10 8 4780250 6 4 4780200 2 4780150 622350 622400 622450 622500 622550 622600 0
Sugar (Brix) in 2006 season 4780350 22 4780300 20 4780250 18 16 4780200 14 4780150 622350 622400 622450 622500 622550 622600 12
Total monoterpenes in fruit (2006 season) 4780350 6 5.5 4780300 5 4.5 4780250 4 3.5 3 4780200 2.5 2 4780150 622350 622400 622450 622500 622550 622600 1.5
2006 cool and wet pattern of variability 4780350 4780350 48 12 44 4780300 40 4780300 10 36 8 4780250 32 28 4780250 6 24 4 4780200 20 4780200 2 16 12 0 4780150 4780150 622350 622400 622450 622500 622550 622600 Soil Moisture 622350 622400 622450 622500 622550 622600 Yield per vine 4780350 4780350 6 22 5.5 4780300 20 4780300 5 4.5 4780250 18 4780250 4 3.5 16 3 4780200 14 4780200 2.5 2 12 1.5 4780150 4780150 622350 622400 622450 622500 622550 622600 Brix 622350 622400 622450 622500 622550 622600 Total monoterpenes
2007 hot and dry! Average soil moisture 4780350 18 4780300 16 14 4780250 12 10 4780200 8 6 4780150 622350 622400 622450 622500 622550 622600 4
Yield per vine (kg) in 2007 season 4780350 8 4780300 7 6 5 4780250 4 3 4780200 2 1 4780150 622350 622400 622450 622500 622550 622600 0
2007 hot and dry pattern of variability 4780350 4780350 18 8 4780300 16 4780300 7 14 6 5 4780250 12 4780250 4 10 3 4780200 8 4780200 2 6 1 4 0 4780150 4780150 622350 622400 622450 622500 622550 622600 622350 622400 622450 622500 622550 622600 Soil Moisture Yield per vine
Are the zones stable? Well, yes they seem to be in this vineyard diagonal zones appear in all 3 years Zones are evident in aerial images Due to soil type and topography variation Effects of zones change from wet year to dry year are these predictable in advance? How can we manage this? What next?
Re-draw Zones based on 2 year dataset 16 15 14 13 12 11 10 9 8 7 6 5
Thermal (long-wave) infrared imaging Lakeshore Rd NOTL Evenly spaced flights dawn to dusk Changes in surface temperature due to canopy, soil, moisture Heating and cooling
Thermal image shows surface temperature differences canopy is cool (blue) soil surface is warmer (orange) Colour-near infrared image Thermal image
Thermal sequence heating pattern shows problem areas (warmer), active canopy (cooler) useful for irrigation scheduling? Morning Noon Afternoon Evening
Remote Sensing and Weather Data Integrated On-line System for Vineyard Management Partnership formed to develop and commercialize remote sensing services for viticulture in Ontario. System is needed to acquire, process and deliver imagery to users in a useful form. Geo-spatial information is extracted, combined with weather data as input to models for prediction of vine stress. Integrated as an on-line system, outputs are useful for decision making spraying, irrigation, harvesting.
Thanks to all those involved Ontario Centres of Excellence Etech NSERC Andrew Peller Ltd. 30-Bench Winemakers Brock University CCOVI Weather Innovations Inc. Aviation International Lakeview Harvesters Dr. Andy Reynolds Darryl Brooker Matthieu Marciniak Linda Tremblay David Ledderhof Lucas Baissas Aiman Soliman Jim Willwerth Javad Hakimi many others