Crop and Irrigation Water Management Using High-resolution Airborne Remote Sensing Christopher M. U. Neale and Hari Jayanthi Dept. of Biological and Irrigation Eng. Utah State University & James L.Wright USDA-ARS, Kimberly, Idaho
Remote Sensing of Crop Evapotranspiration Three basic approaches are being developed 1) Energy balance approach: Ground meteorological data and remotely sensed inputs in the shortwave and longwave bands are used to obtain parameters for estimating Rn, G and H => Energy balance equation obtain instantaneous latent heat fluxes (LE) that are extrapolated for the entire day and over time between satellite overpasses or airborne image acquisition flights
Remote Sensing of Crop Evapotranspiration 2) Reflectance-based crop coefficient approach: Remotely sensed inputs in the shortwave along with reference ET from ground meteorological stations are used to obtain daily actual crop ET in each field Interpolation over time between satellite or airborne images 3) Process models (SVAT models) that use RS for some input variables
Crop and Water Management Outline of Presentation Show how ground-based and airborne high resolution remotely sensed data can be used to develop and verify ET and yield models (precision agriculture perspective) Short description of the USU airborne multispectral digital remote sensing system Show recently developed Kcr for new crops Examples of applications at field and irrigation command area scales
Crop Coefficients Wright (1982) introduced the concept of basalcrop coefficients (Kcb): Kc = Kcb * Ka + Ks Where Ka and Ks are adjustments for limiting water in the root zone and wet soil surface respectively Allen et al (1998) FAO 56 presented crop coefficients for several crops
Reflectance-based Crop Coefficients Jackson et al. (1980), later Heilman et al. (1982) showed similarities with a ratio of the perpendicular vegetation index (PVI) with crop coefficient of wheat and alfalfa. Bausch and Neale (1987) and Neale et al. (1989) developed reflectance-based crop coefficients for corn using the Normalized Difference Vegetation Index (NDVI) from the TM bands Bausch (1993 and 1995) proposed the Soil Adjusted Vegetation Index (SAVI) instead of NDVI for the reflectance-based crop coefficient for corn Neale et al (1996) developed reflectance-based crop coefficients for cotton
Development of Reflectance-based Crop Coefficients: Rationale for Site Selection Kimberly, Idaho Was an area planted with the target crops of potato, sugar beets and beans Same agro-climatic region where Wright (1982) developed the basal crop curves with measured actual crop ET using lysimeters Wright (1982) conducted extensive crop biophysical parameter measurements which we used to relate the ET measurements to our remotely sensed data (Leaf area index, plant height, plant cover)
Approach: Development and Verification at Different Scales Small field scales: Ground-based radiometry and soil-moisture measurements Larger fields: Airborne imagery, flux measurements with Bowen Ratio and Eddy Covariance systems and soil moisture measurements Irrigation command areas and irrigation system scales: Airborne or satellite imagery, flux measurements and water balance (requiring inflow and outflow irrigation water measurements)
Source of Remotely Sensed Data EXOTECH 4-band radiometer with Thematic Mapper Bands TM1 4 => Canopy reflectance obtained with barium sulphate standard reflectance panel with know bidirectional properties High-resolution multispectral imagery from the USU airborne system with pixel resolution varying from 0.20 to 2.5 meters
Description of the USU Airborne Multispectral Digital System 3 Kodak Megaplus 4.2i digital frame cameras with Nikon lenses contain narrow band interference filters in the green (0.55 µm), red (0.67 µm) and near- infrared (0.80 µm) Pentium III computer, 133 Mhz bus, with EPIX PCI controller boards and special software, 30 Gigabyte hard drive and 18 Gigabyte removables GPS for navigation and geo-positioning of images GPS encoded video-tape in color for record of flight
USU Piper Seneca II Remote Sensing Aircraft
Kodak 4.2i digital cameras, Nikon lenses with interference filters
Green Red NIR 3 band
Image Rectification using Digital Orthophotos or ground control points obtained with GPS Digital Orthophoto Map Base 3 Band Multispectral Imagery
Portion of 3-band Image Mosaic of Rio Grande Valley where 1400 Km2 of Irrigated Agriculture were Mapped
Processing of Multispectral Digital Imagery Backup of raw images from the aircraft and import to ERDAS Imagine Correction for geometric and radial distortions Correction for lens vignetting and radiance nonuniformities Registration into 3-band images Geo-registration and/or rectification to a map base Absolute radiometric calibration Formation of large image mosaics Spectral classification Editing and recoding Generation of final GIS layers and products
Processing: Lens Vignetting correction A typical lens will concentrate more energy in the center of the image than along the edges. This must be corrected before image registration and mosaicking and the analysis of the imagery. Images are acquired over a uniformly illuminated halon panel, and a correction coefficient matrix is obtained to bring the brightness to a common plane. A correction coefficient matrix is developed for each lens/camera combination and for several f-stop apertures.
Vignetting Correction Uncorrected image with bright Spot in the center Corrected image with uniform brightness
Calibration for Removal of Geometric Distortions A typical optical lens will create radial distortions which can be severe along the edges of the imagery. These must be removed before band registration and image rectification Images are acquired over a grid with known coordinates. Pixels are re-mapped to proper locations using the nearest neighbor transformation A transformation matrix is developed for each spectral band and camera and then applied to acquired imagery from aircraft
Radiometric Calibration of a Multispectral System
Absolute Image Calibration Curves Red Band Camera 0.2284*DN-2.662 R2=0.9991 Radiance (w/m2) 60 50 40 30 20 10 0 0 50 100 150 200 250 Image digital number Digital Camera Calibration Red Band 250 Radiance (w/m2) 200 150 100 50 7 ms 15 ms 0 0 50 100 150 200 250 Digital Number
Calibration in terms of Reflectance Concurrent measurements of incoming radiation in same spectral bands to obtain, along with system calibration the reflectance of each pixel in the image Use of Radiation Transfer Models such as MODTRAN to obtain reflectance and/or adjust for atmospheric effects: Need profile of temperature and water vapor in atmosphere
Advantages of Image Calibration: Allows for the spectral classification of imagery taken at different times of the day over a region Allows for the development of reflectance-based vegetation indices and their relationships with biophysical canopy parameters such as biomass, LAI, % cover etc. In vegetation and land use change monitoring it allows for the comparison of imagery taken in different years, under different conditions Results from developed models (yield, biomass etc.) can be reproduced and are consistent from year to year.
Basis of the Kcb to Kcr Transformation: Basal crop coefficients Kcb reach peak or maximum crop ET (effective full cover) when LAI is around 3 and percent cover is around 80% At this crop canopy development stage the NDVI and/or SAVI are becoming asymptotic (saturated) SAVI or NDVI values at effective full cover depend on canopy geometry, leaf distribution and whether the crop is planted in rows, drilled or broadcasted
Reflectance-based Crop Coefficient for Beans
SAVI versus LAI Relationship for Beans
Simulation of soil moisture in the bean crop root zone using both the basal and the reflectance-based Kc Kcr = 1.326 * SAVI + 0.02863
Comparison of Accumulated Seasonal ET for Beans Kcr estimate of crop ET (mm/m) 270 260 250 240 230 220 210 200 190 180 170 170 180 190 200 210 220 230 240 250 260 270 Measured crop ET (mm/m) Kcr estimate Kcb estimate
Basal crop coefficient for beans adjusted for date of emergence
Development of the Reflectance-based Crop Coefficient for Potatoes
SAVI versus LAI Relationship for Potatoes
Simulation of soil moisture in the root zone of potato crop using both the basal and the reflectance-based Kc Kcr = 1.085 * SAVI + 0.0507
Measured vs Estimated Soil Moisture
Cummins center pivot #3 with Russet Burbank Potatoes
In-field variability of crop growth has implications for both crop ET estimates and also yield
Verification of Crop Coefficient Methodology at Irrigation Command Area Scales Smithfield canal company (gravity fed sprinkler system with 347 ha) Irrigation water deliveries measured at inlet Crops: Alfalfa, barley, corn, pasture Airborne imagery acquired over 1993 season using USU airborne system GIS database with soil layer, crop layer Reference ET from nearby weather station Calibrated imagery used to obtain reflectance based crop coefficient for each field 7 flights throughout growing season
Cropping Pattern 1993 of Smithfield Command Area
Soil Type Layer of Smithfield Command Area
Crop Development Progression Alfalfa Barley Corn
Seasonal SAVI Trends
Estimated Demand and Measured Irrigation Water Deliveries for Smithfield Command Area in 1993
Final Comments Future research should consider the effects of irrigation water management on yield Energy Balance or Kcr models should be verified in irrigation systems where water balance can be performed Simple Kcr models can perform well Future potato and sugar beet yield models under development will include distributed actual ET estimates and localized stress within the fields High-resolution imagery is useful for precision agricultural applications
Thank you