|Milliken, G - KANSAS STATE UNIVERSITY|
|O'Hara, C - MISSISSIPPI STATE UNIV|
Submitted to: Applied Statistics In Agriculture Conference Proceedings
Publication Type: Proceedings
Publication Acceptance Date: February 1, 2005
Publication Date: April 1, 2005
Citation: Willers, J.L., Milliken, G.A., O'Hara, C.G., Jenkins, J.N. 2005. Information technologies and the design and analysis of site-specific experiments within commercial cotton fields. In: Milliken, G.A., editor. Proceedings 16th Kansas State University on Applied Statistics in Agriculture, April 25-27, 2004, Manhattan, Kansas. p. 41-73. Interpretive Summary: The need for the use of spatial information in the analysis of cotton production is described. Natural spatial variability in soil types, elevation, and drainage patterns can be measured by lidar, multispectral images, and other sensor systems. Current technology provided by variable rate controllers allows producers to apply variable rates of fertilizer, insecticides, herbicides, and many other inputs on a spatial basis. Crop harvesters equipped with differential, global positioning systems (DGPS) yield monitors obtain information about yield at spatial locations. We describe how these various spatial information layers can be analyzed in order to provide feedback to the producer about the value of the spatial decisions made during a production season. An example involving 17 cotton varieties and one spatial application of a plant growth regulator is provided to demonstrate the concepts of the integrated analysis procedure.
Technical Abstract: Information products derived from multi-spectral remote sensing images, LIDAR elevations, or data products from other sensor systems (soil electrical conductivity measurements, yield monitors, etc.) characterize potential crop productivity by mapping bio-physical aspects of cropland variability. These sensor systems provide spectral, spatial, and temporal measurements at resolutions and accuracies describing the variability of in-field, physical characteristic phenomena, including management practices from cropland preparation, selection of crop cultivars, and variable-rate applications of inputs. In addition, DGPS-equipped (differential, global positioning system) harvesters monitor yield response at closely spaced, geo-referenced points. Geographic information system and image processing techniques fuse diverse information sources to spatially characterize cropland, describe management practices, and quantify the variable yield response. Following fusion of information sources, effectiveness of spatially applied management practices may be evaluated by designed experiments assessing impacts on yield caused by geo-referenced relationships between (1) uncontrollable spatial components (the environment) and (2) controllable management practices (cultivar selection, fertility management, herbicide, insecticide, and plant growth regulator applications, etc.). These kinds of experiments can be designed because farming equipment can be computer controlled through DGPS giving farmers the ability to continuously change applied treatments for many farming operations. A mixed linear model involving both uncontrollable and controllable management attributes attached as spatial descriptors to yield monitor points evaluates effects of management practices on yield. An example based upon cotton production demonstrates the methodology. Additional strategies for designing studies in commercial cotton fields involving spatial information are discussed.