Mark W. Brennan and George R. Waddington, Jr.
Variations in crop yields reflect differences due to specific
crop varieties, soil conditions, disease, pests and nutrient imbalances.
Crop growers are concerned with controlling field specific variability
to maximize crop yields and reduce input costs, in order to realize
financial profits. This paper describes the utilization of spatial data technologies in making pre-season crop management decisions, processing remotely sensed data and field specific
information, and analyzing available yield maps. The application of ArcView and MapObjects in a new agricultural information
service is discussed and demonstrated using field studies from
the 1996 growing season near Jacksonville, IL.
Variability exists within agricultural fields. Effective management of this crop variability can lead to increased financial profits, by improving yields and farm productivity, and by reducing input costs. Often the variability within a specific field is related to undulating topography and may involve more than one parameter, such as crop variety, soil condition, disease, pests and nutrient imbalances. For example, Hollands (1996) found that relatively high areas in a field have a higher amount of residual nitrogen, whereas the surrounding low areas are usually lower in residual nitrogen.
The introduction of cartographic technologies, including Global Positioning Systems (GPS), Geographic Information Systems (GIS) and remote sensing, has resulted in more accurate and efficient mapping of field variability. Soil sampling, yield monitors, and digital imagery with various spectral and spatial resolutions are also being exploited for managing site specific variability (Kirk and Tom, 1996; Barnes et. al., 1996). Anderson and Yang (1996) demonstrated the utility of aerial imaging for identifying and mapping differences in organic matter that were apparent in photographs of bare soil before planting. Curran (1985) has reviewed the use of aerial photography for assessing the condition of various crops. Satellite imagery has been shown to provide useful crop management information for potatoes under irrigation in the Pacific Northwest (Waddington, 1994). Remotely sensed imagery acquired in the visible and near infrared wavelengths has considerable potential for mapping soil variability (Johanssen and Dacosta, 1980). This information, coupled with other measurements and field scouting information, allows for effective application of site specific management practices.
Given the technology available today, crop growers have the ability to gather geo-referenced measurements throughout the fields under their management. This permits the analysis of cause and effect relationships within a field that otherwise may go unnoticed. Growers also have a choice of many tools for organizing, analyzing and presenting the geo-referenced information that is collected about their fields. Successful growers involved with site specific management use the available geo-referenced information to monitor spatial relationships and evaluate the derivative information for in-season adjustments or future planning.
The remainder of this paper is dedicated to demonstrating the utility of spatial data technologies for site specific field analyses through a case study from Jacksonville, Illinois. The imagery and graphics shown are extracted from products delivered through TASC's new agricultural information service, emerge(tm). The emerge information service provides value-added products based on high-resolution, geo-referenced, multi-spectral mosaicked imagery. Imagery described in the paper was obtained during the 1996 growing season using a airborne three-channel sensor configuration. ArcView was the primary tool used to ingest and analyze the imagery and related field data. MapObjects was used to develop a spatial data viewer for the imagery, as referenced in the TASC paper entitled "Development of a MapObjects-Based Spatial Data Viewer and Analysis Tool for Remote Sensing Applications", by W. Thomas Ofenstein and Kim Rauenzahn.
Growers in the mid-western United States face numerous crop management problems. Common problems include disease, pest infestations, drought, flooding, soil chemistry imbalances and weed pressure. Some problems can be detected early, located precisely on the ground and treated, resulting in healthier plants with greater yield potential. Often, solutions require more than one season to implement and see results.
The primary crops that were monitored during this study are corn and soybeans. These two crops are subject to specific pest and disease problems, some of which were identified using the emerge imagery. Figure 1 shows a color composite image over a corn field near Jacksonville, Illinois. The emerge sensor system collects information in three channels: green, red and near infrared, and displays them as blue, green and red respectively. Therefore, healthy vegetation in these images appears as red, since healthy vegetation has a strong infrared reflectance. This imagery is useful for many applications, including pest or disease detection, drainage location and soil condition analysis. Additionally, image enhancement techniques (contrast stretch, histogram matching) can be used to provide easier image interpretation.
Growth of weeds in agricultural fields is common; associated weed pressure induces crop stress which can significantly impact yield. The ability to quickly locate and define the extent of weed patches in a field makes site specific treatment of weeds possible. In turn, this saves the grower money on input costs, and is also more environmentally friendly.
The emerge imagery was used to precisely locate and navigate to weed patches and other problem areas. Unique data collection and processing steps result in geo-referenced imagery mosaics with ArcView and MapObjects-compatible world coordinate information. By displaying the imagery in the custom-built MapObjects viewer, it was possible to locate and determine precise latitude and longitude coordinates for a weed patch in the northeastern (top right) corner of the field shown in Figure 1. Figure 2 displays a close-up view of this area, with the cursor in the center of the weed patch, and coordinate readings at the bottom of the screen. The brighter red area in the image indicates the growth of highly reflective vegetation. The spectral signature of this area is different from the surrounding darker red areas and is indicative of weeds rather than corn.
These coordinates were provided to an agronomist to enable navigation to the location for ground verification. The agronomist first attempted to locate the spot by approximating distances from the imagery. When he felt he was in the right location, the coordinate reading from the imagery and the mobile GPS unit indicated the correct spot was another 50 feet into the field. By moving 50 additional feet, he "discovered" the 2 acre patch of foxtail (tall grass). The use of GPS and navigation data obtained from MapObjects confirmed the information shown in the imagery.
The Normalized Difference Vegetative Index (NDVI) is used to indicate the relative health of vegetation. Figure 3 shows a monochromatic version of the NDVI computed over the corn field. Bright white areas indicate healthy foliage, shades of gray signify varying degrees of crop stress with black indicating the worst conditions. This image simplifies the interpretation of imagery, compared to three-channel color composite imagery, and highlights the varying degrees of crop vigor found in a field.
The northwestern portion (top left) of the corn field (Figure 4) contains an area with vertical rows of corn that are thin and interlaced with gray striping (the gray areas have low infrared reflectance, corresponding to soil visible between rows). The linear pattern associated with this area is possibly indicative of a human-induced artifact, for instance, a problem with planting or application equipment. This obvious area of decreased crop vigor warrants additional investigation in the field for visual assessment.
Imagery and crop scouting information should be correlated with other types of field data, such as soil chemistry tests and yield monitor data. Grid soil sampling is the most common method used for site specific soil testing . Usually, samples are collected at pre-determined uniform intervals (e.g., 1 sample per 5 acres). Soil chemistry results can be ingested into a mapping application to create contoured soil maps of individual parameters.
Figure 5 displays soil pH contour lines derived using ArcView-Spatial Analyst for a large portion of the Jacksonville corn field. Note that pH levels are relatively lower in the northwestern (top left) portion of the field, which correlates well with the low IR anomaly found in the aerial imagery at the same location (Figure 4). This region should be sampled again at an appropriate time with more sample locations distributed within this area. A targeted sampling pattern will refine the estimate of soil chemistry imbalances in this region of the field and enable a more accurate treatment program.
Yield monitors provide a record of spatially-related yield variability at the end of the season. These measurements can be used to evaluate obvious in-field spatial associations. For instance, yield monitor data can be correlated with soil sample results, derivative information from aerial images, or other periodic measurements obtained during the season. Yield monitor data are especially useful for confirming events (e. g., pest infestation) that occurred during the previous growing season, and for planning the next season.
Yield data for the top half of the Jacksonville corn field were converted from a text file into an ArcView point shapefile theme for comparison and overlay on the emerge imagery (Figure 6) . The yield data shows a similar pattern to both the soil sample results and the imagery. The northwestern (top left) portion of the field has lower yield readings, reflecting the decreased plant health seen in the imagery. Management strategies for next season might include collecting a denser set of soil samples in this region and potentially treating this low pH area with lime.
The Spring of 1996 was abnormally wet in parts of Indiana and Illinois. The impact of excessive rain and flooding was evident in many fields. Figure 7 a - c shows a second corn field at three distinct time intervals: early July with water damage, late July after replant, and late September. In this case the grower decided to replant after assessing the flooding in an attempt to recover some yield. By using ArcView's measurement tools with the crop scout, it was possible to measure the acreage represented by the replant area from the imagery.
As the crop matures, samples are often pulled to estimate the
yield of the crop. The integration of high-resolution imagery,
GIS software, and GPS navigation can greatly reduce the time involved
in yield estimation. In the example shown in Figure 7, strategically placed sampling
locations in both the replant and original crop areas gave a much
more accurate assessment than a typical grid-based sampling estimate
(which is very difficult when corn is shoulder high). This was
a clear example of how spatial data technologies allow more
fields to be covered with a higher degree of accuracy.
Spatially related variability exists in all agricultural fields. An integrated approach to site specific management is essential for developing yield preservation plans on a field-by-field basis. Combining information obtained using GPS, remote sensing (e.g., soil sampling, yield monitors, digital imagery), and GIS technologies provides the most comprehensive solution for managing spatially related variability. In this case study, the benefits of an integrated approach to field management were demonstrated by the use of ArcView and MapObjects in conjunction with the emerge image products.
The authors would like to extend thanks to additional emerge team members for their contributions to this project: Rob Comer, Steve Karels, Gerry Kinn, Tom Ofenstein, Kim Rauenzahn, Tom Parr, and Bill Pevear. Also, thanks are extended to Darren Bohannan and Paul Tewes, employees of UAP/Richter. emerge is a trademark of TASC, Inc.
Anderson, G. L., and C. Yang, 1996. Multispectral Videography and Geographic Information Systems for Site-Specific Farm Management. Proceedings of the 3rd International Conference, Minneapolis, MN, June 23-26, 1996. pp 681-692.
Barnes, E. M., T. R. Clarke, M. S. Moran, and P. J. Pinter, Tr., 1996. Multispectral Remote Sensing and Site Specific Agriculture: Examples of Current Technology and Future Possibilities. Proceedings of the 3rd International Conference, Minneapolis, MN, June 23-26, 1996. pp. 845-853.
Curran, P. J., 1985. Aerial Photography for the Assessment of Crop Condition: A Review. Applied Geography, Vol. 5. pp. 347-360.
Hollands, K. R., 1996. Relationship of Nitrogen and Topography. Precision Agriculture: Proceedings of the 3rd International Conference, Minneapolis, MN, June 23-26, 1996. pp 3-12.
Johanssen, C. J., and L. H. Dacosta, 1980. Using Soil Color Reflectance in Predicting Soil Properties. In Machine Processing of Remotely Sensed Data Symposium, Laboratory for the Applications of Remote Sensing, West Lafayette, IN. pp. 233-237.
Kirk, I. W., and H. H. Tom, 1996. Precision GPS Flow Control for Aerial Spray Applications. Proceedings of the 3rd International Conference, Minneapolis, MN, June 23-26, 1996. pp. 815-818.
Waddington, G. R., Jr., 1994. Managing Crops. Remotely Sensed Data Technology, Management and Markets. U.S. Congress, Office of Technology Assessment, U.S. GPO, Washington, DC. pp. 153-155.
Mark W. Brennan
TASC
55 Walkers Brook Dr.
Reading, MA 01867
508-262-0672 (O)
508-262-0700 (F)
mwbrennan@tasc.com
George R. Waddington, Jr.
TASC
55 Walkers Brook Dr.
Reading, MA 01867
508-262-0672 (O)
508-262-0700 (F)
grwaddington@tasc.com