|English, Patrick - Mississippi State University|
Submitted to: National Cotton Council Beltwide Cotton Conference
Publication Type: Proceedings
Publication Acceptance Date: 2/8/2010
Publication Date: 4/28/2010
Citation: English, P.J., Defauw, S.L., Thomson, S.J. 2010. Patterns of zone management uncertainty in cotton using tarnished plant bug distributions, NDVI, soil EC, yield and thermal imagery. National Cotton Council Beltwide Cotton Conference. p 843-848.
Technical Abstract: Management zones for various crops have been delineated using NDVI (Normalized Difference Vegetation Index), apparent bulk soil electrical conductivity (ECa - Veris), and yield data; however, estimations of uncertainty for these data layers are equally important considerations. The objective of this study was to evaluate the extent of spatially non-autocorrelated areas in an irrigated cotton field in the Mississippi Delta (5.8 acres - with substantial contrasts in soil texture and water-holding capacities) using NDVI, ECa, yield, and thermal imagery as well as Tarnished Plant Bug (Lygus lineolaris) distribution maps (the latter taken at peak bloom). Three year composites for plant NDVI and TPB distributions (analyzed using univariate Local Moran’s I Spatial Autocorrelation – LISA) exhibited the highest levels of uncertainty (with 2.7 and 2.6 acres or 46.0% and 45.5% of the field extent, respectively); whereas, multi-year compositing of yield monitor data more closely matched the shallow ECa map acquired in October 2004 (involving 2.0 and 1.5 acres or 34.7% and 26.5% of the field extent, respectively). Comparison of thermal datasets (acquired from July to September over two growing seasons) with the multi-year yield map highlighted the significance of the linkages of low yield zones with areas of the field subjected to the highest temperatures as well as the pairing of high yield zones with cooler canopy temperatures. These types of uncertainty assessments demonstrate that the fusion of multi-year datasets may allow predictive field-specific models to be created and then used by producers to more effectively manage risk.