Page Banner

United States Department of Agriculture

Agricultural Research Service

Research Project: Pesticide Application Technologies for Spray-drift Management, Maximizing In-field Deposition, and Targeted Spraying

Location: Crop Production Systems Research Unit

Title: Methods for Georeferencing and Spectral Scaling of Remote Imagery using ArcView, ArcGIS, and ENVI

Authors
item Bright, Jerry
item Thomson, Steven
item Huang, Yanbo

Submitted to: ASABE Annual International Meeting
Publication Type: Proceedings
Publication Acceptance Date: August 24, 2009
Publication Date: September 1, 2009
Citation: Bright Jr, J.R., Thomson, S.J., Huang, Y. 2009. Methods for Georeferencing and Spectral Scaling of Remote Imagery using ArcView, ArcGIS, and ENVI. ASABE Annual International Meeting. ASABE Paper No.09-5549

Interpretive Summary: Use of aerial platforms for variable application of pesticides or harvesting aids is a relatively new practice compared with use of ground-based application platforms for the same purpose. Remote sensing images can determine which areas of crop fields require treatment and which areas do not require treatment. For a variable-rate application system to determine where to apply material, these images must be “georeferenced” by global positioning or other methods. Georeferencing assigns location (latitude and longitude) to every point in the field. When applying material by air, certain issues are prevalent that do not apply to ground vehicles. It is commonly known that aircraft can roll or tilt causing potential distortion in remote sensing images. These distortions can affect accuracy of application due to corresponding georeferencing error. Studies we have conducted with agricultural aircraft, however, have shown that these errors may not be significant when compared with potentially larger application errors inherent in a spray delivery system that must accommodate rapid rate changes at ground speeds up to 70 m/s. Many procedures can be used to georeference images and create field prescriptions, but documentation of these procedures is scarce. We have outlined a set of straightforward procedures to georeference imagery using a pre-set field prescription that specifies different application rates. Procedures using both ArcGIS 9.2 and Arcview 3.2a Geographic Information System (GIS) software are outlined. Another remote sensing application involves use of a thermal imaging camera to detect the onset of crop water stress on a temporal basis. Obtaining the proper image representation of canopy temperature has been quite a challenge since differences in incoming solar radiation and wind from day to day can alter that representation. Inconsistent results have thus been obtained from imagery over long-term drying cycles that should have shown a clearly increasing trend in canopy temperature. Image manipulation tools in ENVI 4.4 image analysis software that could be useful for scaling imagery to account for varying solar radiation are proposed. It is hoped that appropriate scaling techniques for thermal imagery can be properly utilized and lead to the development of methods for detecting the onset of crop water stress over large field areas.

Technical Abstract: Remote sensing images can be used to support variable-rate (VR) application of material from aircraft. Geographic coordinates must be assigned to an image (georeferenced) so that the variable-rate system can determine where in the field to apply these inputs and adjust the system when a zone has been traversed. Resulting imagery obtained aerially can exhibit position distortions due to airplane roll or tilt. Depending on the position accuracy required for the image and assuming that rotational variables (roll, tilt, yaw, pitch) can be measured in the airplane with an Inertial Measurement Unit (IMU), image pixels can subsequently be shifted (georectified) to remove these distortions. However, a preliminary study using images obtained from aircraft has shown minimal improvement to positioning accuracy by compensating for these variables. This is coupled with the fact that effective experimental positioning accuracy on the ground for on-off rate changes has been found to be no better than + 5 m (with a standard deviation of error magnitude up to 7 m). This result was typical of experiments conducted using GPS receiver position updating of 0.2 s. Distortion correction under these error conditions would provide only minimal relative improvement in positioning accuracy. Rapid procedures for georeferencing and rectifying imagery are outlined using ground control points obtained by RTK GPS. Both ArcView 3.2a and ArcGIS 9.3 were used, and interesting procedural differences between the two GIS are discussed that could dictate preference of one method over the other. Positioning error due to rotational attributes was not factored into the processing procedure. Another type of processing involves image scaling. We have used a thermal imaging camera in an effort to detect the onset of crop water stress on a temporal basis. Obtaining the proper image representation of canopy temperature has been quite a challenge since differences in incoming solar radiation and wind from day to day can alter that representation. Inconsistent results have thus been obtained from imagery over long-term drying cycles that should have shown a clear trend (increase) in canopy temperature. Image manipulation tools in ENVI 4.4 image analysis software that could be useful for scaling imagery to account for varying solar radiation are proposed. It is hoped that appropriate scaling techniques for thermal imagery can be properly utilized and lead to the development of methods for detecting the onset of crop water stress over large field areas.

Last Modified: 10/1/2014
Footer Content Back to Top of Page