indices for precision agriculture
TRANSCRIPT
INDICES FOR PRECISION AGRICULTURE
INTERDRONE 2015 PRECISION AGRICULTUREPRESENTED BY ALEX CHARLES
INTRODUCTION
The spearheads of precision agriculture are various indices that give the grower a comprehensive overview of the crop’s overall health. The two most promising precision agriculture indices are Normalized Difference Vegetation Index (NDVI), and Crop Water Stress Index (CWSI). The session will address:• The history of NDVI and CWSI• NDVI and CWSI algorithms• The types of data and corresponding sensors required to acquire NDVI and
CWSI• Ways to interpret NDVI and CWSI• Integrating NDVI and CWSI with existing farm management software
NDVI
• Rationale behind NDVI• History of NDVI• NDVI Algorithm• Collecting NDVI data
Fig 1: Side by side comparison of RGB and NDVI maps
NDVI
• Normalized Difference Vegetative Index is a Vegetative Index model that describes the relative “greenness” of the inspected area.
• Photosynthetic cells absorb red and blue light and reflect near-infrared(NIR) and green light
• Comparing the amount of red light absorbed with the amount of NIR reflected is a reliable indicator of photosynthetic activity. [1] Fig 2: Absorption spectrum of Chlorophyll [2]
HISTORY OF NDVI
“Landsat 1 was launched on July 23, 1972; at that time the satellite was known as the Earth Resources Technology Satellite (ERTS). It was the first Earth-observing satellite to be launched with the express intent to study and monitor our planet’s landmasses.” [3]2 Camera Systems:• Return Beam Vidcon (RBV)• Multispectral Scanner (MSS)
Fig 3: Sketch of Landsat1 Satellite [3]
HISTORY OF NDVI• MSS recorded data in four spectral bands: Red, Green, and two
infrared bands.• Great Plains Corridor project: “The Landsat-1 Great Plains Corridor project… was designed to evaluate Landsat capability for quantitatively monitoring the progression of phenological development from South Texas northward through North Dakota.” [1] • Deering and Hass developed a correlation between the spectral
response seen on channels 5-7 on landsat 1 (Red, and the two NIR bands) and volume of Green biomass
NDVI ALGORITHM
• THE NDVI EQUATION AS FORMULATED BY DEERING AND HASS IS [1] :
(1)• Near IR: reflectance of Near IR light (>700nm)• Red: reflectance of Red Light (650-700nm)• Results in a number value between 0 and 1• 0.1: bare soils• 0.9: dense vegetation
Fig 2: Absorption spectrum of Chlorophyll [2]
COLLECTING NDVI DATA
2 pieces of information needed:• Reflectance of Red light• Reflectance of Near-Infrared Light
2 approaches to collecting reflectance data:• True NDVI: Record incident light (Irradiance)
and contrast with reflected light (radiance)• Relative NDVI: Statistically determine the
amount of incident light
Fig 4: Analyzing an NDVI orthomosaic in Pix4D
TRUE NDVI VS. RELATIVE NDVI
TRUE NDVI• Usage of either incident light
sensor or reflective panels to record irradiance
• High confidence level for data• Expensive to install and maintain
reflective panels in the field• Compromise: capture
“calibration” image before flying field
RELATIVE NDVI• Using statistical probability
models to estimate irradiance• Medium confidence level for
data (good enough)• Cheaper to implement, no
maintenance cost• Eliminates extra step in data
acquisition
CONSTRUCTION OF NDVI CAPTURING DEVICE• Imaging sensors in modern digital cameras have
the natural ability to record in the near-infrared region of light
• The most common single-sensor implementation:• remove the infrared filter installed by camera
manufacturer• Install a notch filter that replaces the green or
blue channel with an NIR channel• Post Processing: Read radiance values of Red and
NIR channels and compute NDVI value
Fig 5: Public Labs NIR Camera conversion kit
CWSI
• Rationale behind CWSI• History of CWSI• CWSI Algorithm• Collecting CWSI data
Fig 6: Thermal image of a cotton field in Arizona [5]
CWSI
• Crop water Stress Index is a vegetative index that describes the water stress level of a crop by proxy of the difference between air temperature and canopy temperature
• Rationale: The more water a plant has access to, the more the plant will transpire water, resulting in a cooling effect right the canopy.
• The difference between air temperature and canopy temperature has been proven to be a reliable indicator of water stress in crops
Fig 7: AgriImage drone flying over an irrigation pivot
HISTORY OF CWSI [6]
• 1843: French biologist Rameaux was the first scientist to attempt to record the temperature of the leaves of a plant : beginning of research in correlation between plant temperature and water stress
• 1923: Miller and Saunders used thermocouples to measure the temperature of the canopy of a crop in Kansas and recorded an average temperature difference of 1⁰C
• 1981: Idso and Jackson developed a formal relation between canopy temperature and crop water stress, leading to CWSI
CWSI ALGORITHM
• The CWSI equation as formulated by Idso and Jackson is [5] : (2)
• dT is the difference between the air and canopy temperature• dTl is the lower limit of the difference between air and canopy
temperature• and dTu is the upper limit of the difference between air and canopy
temperature
CWSI ALGORITHM
• Idso et al. developed a method to estimate the upper and lower limits of the temperature variance by taking the vapor deficit into account [5]:
(3) (4)
• b and m are the empirically determined intercepts and slopes, respectively, for the crop of interest
• VPD(T) is the recorded vapor pressure deficit as a function of temperature
• VPsat(T) is the saturation vapor deficit as a function of temperature
CWSI ALGORITHM
• VPD and Vpsat are describes as[7]: (5) (6)
• temperatures are in degrees Celsius, and Td is the dew point• Upper and lower difference limits can be determined using a weather
station or access to localized weather data.
CWSI ALGORITHM
Crop Intercept (b)
Slope (m)
Alfalfa 0.51 -1.92Barley (pre-
heading) 2.01 -2.25Barley (post-
heading) 1.72 -1.23Bean 2.91 -2.35Beet 5.16 -2.3Corn (no tassels) 3.11 -1.97Cotton 1.49 -2.09Cowpea 1.32 -1.84Cucumber 4.88 -2.52Lettuce, leaf 4.18 -2.96Potato 1.17 -1.83Soybean 1.44 -1.34Tomato 2.86 -1.96
Wheat (pre-heading) 3.38 -3.25
Wheat (post-heading) 2.88 -2.11
An example of the slope and intercept values for Soybeans is given below, followed by values for various other crops:
Fig 8: Example Vapor Pressure deficit Baseline for Soybeans [5]
Intercept and Slope values for various crops [5]
CWSI ALGORITHM
𝐶𝑊𝑆𝐼= (𝑇 𝑎−𝑇𝑐 )−(𝑏+0.6108𝑚 ∙(𝑒17.27 ∙𝑇 𝑎
𝑇 𝑎+265.5−𝑒17.27 ∙𝑇𝑑
𝑇 𝑑+ 237.3))0.6108𝑚∙ (𝑒
17.27∙𝑇 𝑎
𝑇𝑎+237.3 −𝑒17.27 ∙ (𝑇 𝑎+𝑏 )𝑇𝑎+𝑏+265.5 )
COLLECTING CWSI DATA• 4 parameters:
• Ta: Air Temperature in oC
• Td: Dew point in oC
• Tc: Canopy Temperature in oC
• Slope and intercept for crop in question
• Ta and Td are recorded using a weather station
• Tc is recorded using a thermal imaging camera capable of producing thermographic images
• Slope and Intercept values are determined during post processing
Fig 9: Flir Thermal Imaging Camera [8]
INTERPRETING NDVI AND CWSI DATA
NDVI• Focused areas for further soil and tissue
testing• Variable rate nutrient and pesticide
application• Stand count/Yield estimates• Damage assessment• Determining local water drainage
problems
CWSI• Variable rate water application• Quality control in vineyards• Quality control for fruits and
vegetables• Main focus for USDA research
Fig 9: NDVI image showing drainage issues [8]
INTEGRATING NDVI AND CWSI WITH EXISTING FARM MANAGEMENT SOFTWARE.
• THE KEY TO HAVING SUCCESSFUL ADOPTION OF A SYSTEM IN ANY GROW OPERATION IS SEAMLESS INTEGRATION
• KEY ATTRIBUTES THAT GROWERS LOOK FOR IN ANY PRECISION AG SYSTEM ARE:• INTUITIVE FLIGHT PLANNING• AUTOMATED FLIGHT AND DATA COLLECTION• EASILY GENERATED ORTHOMOSAIC MAPS IN THE SHAPE FILE FORMAT
REFERENCES[1] D. Deering and R. H. Hass, "Using Landsat Digital Data for Estimating Green Biomass," NASA, Greenbelt, MD, 1980.
[2] J. Whitmarsch and Govindjee, "The Photosynthetic Process," 1995. [Online]. Available: http://www.life.illinois.edu/govindjee/paper/gov.html#52. [Accessed 22 August 2015].
[3] NASA, "Landsat Science," NASA, 19 August 2015. [Online]. Available: http://landsat.gsfc.nasa.gov/?p=3172. [Accessed 22 August 2015].
[4] Public Labs, "Near Infrared Camera," Public Labs, June 2015. [Online]. Available: http://publiclab.org/wiki/near-infrared-camera. [Accessed 25 August 2015].
[5] USDA agricultural Research Service, "Crop water Stress Detection," USDA, 15 December 2005. [Online]. Available: http://www.ars.usda.gov/Main/docs.htm?docid=9715&pf=1. [Accessed 24 August 2015].
[6] R. D. W. P. K. a. B. J. C. Jackson, "A reexamination of the crop water stress index.," Irrigation Science, vol. 9, no. 4, pp. 309-317, 1988.
[7] K. P. U. R.L. Snyder, "Measuring Vapor Pressure Deficit in the Field," Regents of the University of California, 6 January 2006. [Online]. Available: http://biomet.ucdavis.edu/biomet/VPD/vpd.htm. [Accessed 24 August 2015].
[8] Flir, "FLIR Tau 2 LWIR Thermal Imaging Camera Core Series," Flir, 2015. [Online]. Available: http://www.flircameras.com/flir-tau-2/flir-tau-2-thermal-imaging-camera-core-series.htm. [Accessed 25 August 2015].