Abstract

A scouting protocol for cotton insect pests was developed which combines high resolution, multispectral remotely sensed imagery with a belt transect that crosses rows of cotton. Imagery was used to determine sample site selection while estimating plant bug abundance in a more than 200 ac. cotton field in 1997. Tarnished plant bug (Lygus lineolaris) counts were acquired using a standard drop cloth for each of eight rows along a transect. The sample data indicated that plant bug population densities spatially vary as a function of different spectral (color) classes present on the imagery. We postulate that such classified images correlate to differences in crop phenology, and plant bug populations (especially from early to mid-season) aggregate themselves by these habitat differences. Therefore, the population dynamics of Lygus, and possibly other species, can be better understood by combining the transect-based sampling plan with remotely sensed imagery. To verify and validate this claim, a computer intensive approach was utilized to simulate the performance of different sampling plans. The comparison is accomplished with a combinatorial algorithm that exhaustively enumerates the original data into unique subsets. These subsets correspond to results that could be expected from the use of traditional or alternative sampling plans and compared to results from the candidate plan actually used. The results of the enumerative analysis show the benefit of multi-band, remotely sensed imagery combined with the use of large sized sample units to improve sampling efficiency (and without the need to have large sample sizes). It is of great benefit that the enumerative algorithm provided answers to questions of interest without having to complete additional fieldwork.

Keywords

Sampling, enumeration, integrated pest management, remote sensing, precision agriculture

Creative Commons License

Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.

Download

Share

COinS
 
Apr 30th, 11:00 AM

APPLICATION OF COMPUTER INTENSIVE METHODS TO EVALUATE THE PERFORMANCE OF A SAMPLING DESIGN FOR USE IN COTTON INSECT PEST MANAGEMENT

A scouting protocol for cotton insect pests was developed which combines high resolution, multispectral remotely sensed imagery with a belt transect that crosses rows of cotton. Imagery was used to determine sample site selection while estimating plant bug abundance in a more than 200 ac. cotton field in 1997. Tarnished plant bug (Lygus lineolaris) counts were acquired using a standard drop cloth for each of eight rows along a transect. The sample data indicated that plant bug population densities spatially vary as a function of different spectral (color) classes present on the imagery. We postulate that such classified images correlate to differences in crop phenology, and plant bug populations (especially from early to mid-season) aggregate themselves by these habitat differences. Therefore, the population dynamics of Lygus, and possibly other species, can be better understood by combining the transect-based sampling plan with remotely sensed imagery. To verify and validate this claim, a computer intensive approach was utilized to simulate the performance of different sampling plans. The comparison is accomplished with a combinatorial algorithm that exhaustively enumerates the original data into unique subsets. These subsets correspond to results that could be expected from the use of traditional or alternative sampling plans and compared to results from the candidate plan actually used. The results of the enumerative analysis show the benefit of multi-band, remotely sensed imagery combined with the use of large sized sample units to improve sampling efficiency (and without the need to have large sample sizes). It is of great benefit that the enumerative algorithm provided answers to questions of interest without having to complete additional fieldwork.