Ralph D. Bagwell, B. Rogers Leonard, Jay W. Hardwick, Edwards Barham, Dale Magoun, Randy Price, Robert G. Downer and Kenneth W. Paxton

Geospatial tools offer great promise of increasing profitability of cotton production. These tools, however, must be adapted to the specific agronomic and plant protection needs of cotton production and made available in a user-friendly format that can be easily transferred to producers, commercial pesticide applicators and agricultural consultants.

In 2002, a multi-disciplinary team including LSU AgCenter scientists initiated a project to develop spatially variable pesticide applications based on remote sensing to improve the efficiency and profitability of cotton production. The objectives of the study are: 1) to correlate arthropod densities with field growth patterns, 2) to apply spatially variable pesticide applications to targeted zones in a field, 3) to perform economic comparisons of precision farming techniques to conventional plant protection strategies and 4) to develop and disseminate educational programs on the efficiency and value of precision agriculture techniques.

The study is being conducted at Hardwick Planting Co. near Somerset, La. Images are gathered with fixed wing aircraft flown at 12,000 feet. These multi-spectral images are geo-referenced and used to create a Normalized Difference Vegetation Index (NDVI) of the study site, which provides an estimate of plant vigor.

To correlate arthropod densities with multi-spectral images, test fields (300 to 450 acres total) were sectioned into one-acre grids. Five randomly selected sites are then sampled for the predominate arthropods in each one-acre grid. Arthropod density data are then correlated with multi-spectral image data.

To determine economic value of this technology, insecticide applications are compared with traditional broadcast applications when arthropod densities reach the treatment threshold. Areas selected for variable rates are determined by multi-spectral imaging. Generally, insecticides are applied to areas indicating high plant vigor (60 percent to 80 percent of the field) and not applied to areas with the lowest plant vigor. Four paired cotton fields (75 to 120 acres) are treated with either the variable or traditional sprays. Lint yields and production costs are then compared for the two treatments.

The third evaluation looks at yield history to determine areas where pesticide applications are rarely justified. Geo-referenced historical yields are identified and pesticide exclusion zones are created in low-yielding areas. Three to four replicates of these treatments are compared with traditional whole-field sprays. Lint yields and production costs are then compared.

The final evaluation is a replicated comparison of spatially variable defoliation treatments with traditional blanket defoliation treatments. NDVI images are used to develop multiple-rate defoliation treatments. Defoliant rates are varied by changing the output volume of the spray equipment. Treatments are replicated three to four times and are sufficient in size that one, preferably two, modules can be made from each plot. Lint yields, production costs and lint quality factors are then compared.

These research results will provide the basis for integrating geospatial technologies into the current cotton production system. A demonstration program will be developed for producers, pesticide applicators and agricultural consultants on the efficiency and value of precision agriculture in cotton IPM. The results of this project will contribute the necessary information to integrate geospatial technologies into current IPM systems. The days of the “pesticide treadmill” with its disastrous results of pest resurgence, secondary pest outbreaks and environment contamination in cotton production systems may be eliminated if a base of information can be developed to support the appropriate development of these technologies.

Ralph D. Bagwell, Associate Professor, and B. Rogers Leonard, Professor, Macon Ridge Research Station, Winnsboro, La.; Jay W. Hardwick, Operator, Hardwick Planting Co., Somerset, La.; Edwards Barham, Aerial Applicator, Barham Brothers Aviation, Oak Ridge, La.; Dale Magoun, Head, Department of Mathematics, Computer Science and Physics, University of Louisiana at Monroe, Monroe, La.; Randy Price, Assistant Professor, Department of Biological and Agricultural Engineering, LSU AgCenter, Baton Rouge, La.; Robert G. Downer, Associate Professor, Department of Experimental Statistics, LSU AgCenter, Baton Rouge, La.; and Kenneth W. Paxton, Professor, Department of Agricultural Economics, LSU AgCenter, Baton Rouge, La.

(This article was published in the 2003 spring issues of the Louisiana Agriculture magazine.)