Takow Kel - KEL Home
Transcript of Takow Kel - KEL Home
Elvis A.Takow1, Edward W. Hellman2, Andrew G. Birt1, Maria D. Tchakerian1,
Robert N. Coulson1
Modeling Viticultural Landscapes: An Environmental Viticulture Information System
1Knowledge Engineering Laboratory, Department of Entomology, Texas A&M University, College Station, TX 77843 USA 2Texas A&M University, AgriLife Research and Extension Center, 1102 East FM 1294, Lubbock, TX 79403 USA; Department of Plant and Soil Science, Texas Tech University
Rationale
• Growing US and Texas wine industry.
• Increased demand for quality grapes and wine.
• Limited knowledge base of varietal suitability in US and Texas in particular.
• Match appropriate grape varieties to existing environmental conditions.
Assumptions
• Grape varieties in established European winegrowing regions are ‘optimal’ for the prevailing climatic and edaphic conditions.
• Relationships between environmental conditions and varieties in the “Old World” can be used as reliable predictor of grape variety selection in new regions.
Goal • Understand the environmental factors that
drive grape variety selection and use this knowledge in the establishment of vineyards in the “New World”.
Objectives
1. Develop a spatial database of environmental information.
2. Develop statistical models that relate environmental conditions to selection of appropriate grape varieties.
3. Deliver a web-based technology for further analysis and interpretation of models towards site selection – Decision Support
Data Sources
• National Climatic Data Center (NCDC)
• World Meteorological Organization (WMO).
• Soil Survey Geographic (SSURGO) Database
• Harmonized World Soil Database (HWSD)
• Topography-US Geological Survey
Raw
• Mean temperature (.1 Fahrenheit)
• Mean dew point (.1 Fahrenheit)
• Mean wind speed (.1 knots)
• Maximum temperature (.1 Fahrenheit)
• Minimum temperature (.1 Fahrenheit)
• Precipitation amount (.01 inches)
Climate Data
• Organic Carbon • pH • Available Water Capacity • Soil Depth • Cation Exchange Capacity • Salinity • Soil Texture Class • Elevation • Slope
Soil/Topography Data
Data
Grape Varieties
• Cabernet Sauvignon
• Chardonnay
• Pinot Noir
• Riesling
• Sangiovese
• Tempranillo
Methodology
• Identify relevant indices of climate, soil and topography suitable to grapevine growth.
• Use quantitative (data driven) statistical methods to analyze location based environmental factors important for grapevine growth.
Statistical Methods Multiple Logistic Regression
• Predict or estimate the probability that variety (Y) can be grown at a particular location.
• Understand the functional relationships between variety and environmental conditions.
• Determine which conditions might be causing the variation in the variety choice.
GSAT Tvar Srad ET FF LF Variety
55.43 8.42 20.98 23.47 16.00 111.30 NoCab
57.73 14.32 21.45 26.00 44.64 69.15 NoCab
57.94 9.83 20.72 25.41 26.58 107.86 NoCab
53.69 10.81 20.72 23.85 45.73 65.96 NoCab
57.73 14.32 21.45 26.00 44.64 69.15 NoCab
65.26 15.31 21.74 30.00 21.59 111.65 NoCab
61.36 10.52 21.31 27.69 17.00 113.00 NoCab
57.73 14.32 21.45 26.00 44.64 69.15 NoCab
53.69 10.81 20.72 23.85 45.73 65.96 NoCab
57.89 8.20 20.99 26.29 38.78 70.22 NoCab
49.63 7.38 18.16 17.55 57.95 56.95 Cab
45.00 5.51 19.03 16.05 56.95 63.25 Cab
50.72 4.05 18.28 18.26 62.40 76.60 Cab
46.94 5.54 18.32 0.00 61.20 59.56 Cab
49.81 6.59 19.03 17.94 43.97 73.71 Cab
60.83 11.80 21.00 23.24 22.25 110.68 Cab
55.29 9.24 19.79 20.85 41.28 79.76 Cab
53.36 3.55 19.83 19.95 41.38 103.88 Cab
36.05 5.18 19.43 13.28 106.72 30.84 Cab
57.19 5.08 20.42 21.99 7.29 117.36 Cab
Sample Data
Var Prob (Y)= α0 + α1Prcp + α2MaxT + α3MinT+ α4Tvar + α5Srad+ α6ET + α7FF+ α8LF
Statistical Methods Discriminant Analysis
• Predict variety membership based on a linear combination of environmental variables.
• Observations of conditions at selected locations are used to best separate varieties based on average of the conditions under which that variety is grown.
• A Likelihood function expresses the probability of the observed data as a function of the unknown parameters.
Modeling Approach
Climate Soils
Topography
Environmental Database
Geoprocessing
Climate
Soils
Topography
Site Selection
Web Page (Name of APP)
Online Mapping Define Area of Interest
View Tabular Data View Graphical Data
Download Data
Server Technologies Data analysis, Web Services
Raw Data Weather, Soils, Topography
Explore Retrieve raw and interpreted data
Compare Compare data for two or
more areas of interest
Analyze Model output relates data
to production &varietal selection
Environmental Viticulture Information System
Incre
ased
Inte
rpretatio
n an
d U
tility
System Architecture
Describe a Location
Bordeaux, France Napa Valley, California
McLaren Vale, Australia Santiago, Chile
Compare Locations
Analyze a Location
Case Study Soil Series Horizon Depth(cm) Organic
Matter (%)
Available Water
Capacity(cm)
pH(1:1 H20)
Texture
Dev-River wash complex H1 24 4 0.09 7.9 Loam Dev-River wash complex H2 70 4 0.09 7.9 Sandy
Loam River wash complex H1 203 0.5 0.03 8.2 —
Region Analogous Regions Existing Vars Recommeded Vars
Leakey,TX Malaga, Esp Shiraz Tempranillo
Jerez, Esp Sangiovese
Muscat