Site Specific Weed Management:
Remote Sensing Detection and Site Specific Control

Jointed goatgrass is a nationwide problem estimated to cost producers over $145 million annually. In Colorado alone over 200,000 acres are infested with Jointed Goatgrass annually causing severe grain yield losses to farmers. Efficient management techniques are needed so that jointed goatgrass infestations can be controlled without causing any yield damage. One way of doing this is by mapping fields (using a handheld GPS) that are infested with jointed goatgrass and then doing site specific control of the infestations.

However, this process is time consuming, labor intensive, and costly. Alternatively, we can use remote sensing (i.e. aerial or satellite imagery) to identify field areas that are infested with jointed goatgrass and then treat specifically those areas with herbicide for control. The big question is: Can we detect, identify, and map jointed goatgrass using remote sensing & then control it effectively??
goatgrass possible.

The overall idea of this study is to "remotely" detect and map areas of wheat fields that are infested with jointed goatgrass and control them the very same year or to use prior information from previously mapped fields to treat jointed goatgrass in the following years. The specific objectives of this ongoing study, for the last three years were: (1) To determine if jointed goatgrass could be detected remotely in winter wheat fields at wheat maturity using digital color infrared aerial imagery. (2) To determine the minimum jointed goatgrass density that is accurately identified by using these techniques. (3) The third and final objective of this research was to control jointed goatgrass via site-specific application of Imazamox (Beyond herbicide) to test the herbicide efficacy in controlling weed infestation and impact on grain yield. Figure 1.0 above shows the patchy nature of jointed goatgrass infestations at study field 8 during 2001 - 2002 growing season.Figure 1.0. Jointed goatgrass patches mixed with wheat at study field 8, during 2001 - 2002 growing season.

This ongoing study was conducted over three consecutive years (1999 - 2002 wheat growing seasons) at eight winter wheat fields located in Northeastern Colorado with historical infestations of jointed goatgrass (Figure 2.0). These fields were planted with both standard and Clearfield wheat. Data collected during the second and third years growing season include: digital color infrared imagery, weed locations using a global positioning (GPS) system, and several other field-crop parameters. This includes jointed goatgrass weed seed banks, wheat grain yield, as well as samples of height, density, and biomass for both wheat and jointed goatgrass.

Figure 1.0. Jointed goatgrass patches mixed with wheat at study field 8, during 2001 - 2002 growing season.

The first year of research (1999 - 2000 growing season) gathered preliminary data regarding the spatial aggregation of jointed goatgrass using new technologies including global positioning systems (GPS). Results from the first year data indicated potential for mapping jointed goatgrass with remote sensing technology. Results from the second and third years of research indicate jointed goatgrass can be detected and mapped at accuracies ranging from 8% to 56% depending on the distribution of jointed goatgrass across the study field. These results indicate the optimum techniques for accurately (>80% accuracy) mapping jointed goatgrass within winter wheat are still to be investigated. However, results indicate jointed goatgrass detection accuracy was positively influenced by the density and the size of weed infestations.

Accuracy results for assessing different densities of jointed goatgrass infestations is currently underway. From a farm management perspective, determining all types of weed infestations within a farm field is of practical value. Using the above remove sensing techniques, kochia, buckwheat, and several other "green" colored invasive weeds within our study fields were identified with accuracy range of 75% to 97%. Such detection accuracy makes remote sensing a very viable tool for site-specific weed management on farm fields.

Spatial statistical procedures were also utilized in an innovative approach to predict field variables with a high degree of accuracy. Table 1.0 below shows the highest accuracy achieved for predicting individual field variables. These field variable predictions are extremely useful in assessing and mapping wheat and jointed goatgrass patterns within a field.

Our site-specific control and management of jointed goatgrass study objective indicated that the efficacy of the herbicide when applied site-specifically versus uniform conventional application was same. Such a finding indicates that site-specific control of jointed goatgrass would save money to farmers on expensive herbicide and also be beneficial to the environment.

Figure 3. Visible distinction between Clearfield wheat (right) and standard wheat (left) with intermixed kochia infestations at Study Field 3 during the 2001 - 2002 growing season.
	These results indicate the first objective of detecting jointed goatgrass remotely in winter wheat fields at wheat maturity using digital color infrared aerial imagery needs further research before detection can be done with consistency and high accuracy levels. However, results indicate these techniques are very useful for identifying other "green"weeds within a winter wheat field such as kochia, buckwheat, and many others in an efficient manner. The second objective of determining the minimum jointed goatgrass density that is accurately identified by using these techniques is currently underway. Results of this objective are expected to be consistent with those of the first objective. Lastly, site-specific control of jointed goatgrass with applications of Imazamox (Beyond herbicide) were found to save money to farmers on expensive herbicide and also be beneficial to the environment.
Figure 4, shows the color infrared imagery for the same field with jointed goatgrass and kochia patches indicated.	1Chris Woodward, 2Raj Khosla, and 3Phil Westra