Submitted to: Soil Science Society of America Annual Meeting
Publication Type: Abstract Only
Publication Acceptance Date: June 2, 2008
Publication Date: October 9, 2008
Citation: Woodbury, B.L., Lesch, S.M., Eigenberg, R.A., Miller, D.N., Spiehs, M.J. 2008. Using EMI, Geospatial Statistics and Multi-Linear Regression for Identifying Areas of Manure Accumulation on Feedlot Surfaces. [abstract] In: Proceedings of Joint Annual Meeting of the Geological Society of America, Soil Science Society of America, American Society of Agronomy, Crop Science Society of America, and Gulf Coast Association of Geological Societies, Oct. 5-9, 2008, Houston, TX. Abstract 9123840, CDROM 2008. Technical Abstract: Accumulated feedlot manure negatively affects the environment. The objective was to test the validity of using EMI mapping methods combined with predictive-based sampling and ordinary linear regression for measuring spatially variable manure accumulation. A Dualem-1S EMI meter also recording GPS coordinates was pulled across the feedlot surface on 2 m intervals to collect horizontal dipole ECa data. A stratified random sampling (SRS) design was determined by ranking pen ECa data from highest to lowest value, segmented into 4 equal ranges. Five values from each range (N=20) were randomly selected. Another 20 sites were selected using the response surface sampling design (RSSD) program contained in the ESAP program suite. The associated site GPS coordinates were used to navigate for sample collection. Samples were analyzed for volatile solids (VS), total nitrogen (TN), total phosphorus (TP), and chloride. This sampling approach enables evaluations of the prediction-based ESAP sampling design with the SRS design by employing three different model validation tests. Correlation coefficients between ECa and VS, TN, TP and Cl-, data were 0.95, 0.95, 0.93 and 0.95, respectively, for the RSSD based model and 0.94, 0.88, 0.88 and 0.92, respectively, for the SRS based model. Composite F-tests showed the two sampling methods to be equivalent for each soil constituent. Additionally, joint predictive F-tests indicated that the regression models for the soil constituents were able to make unbiased predictions of average values at the SRS sites using the RSSD models. The mean predictive t-tests indicated no bias in average value predictions across the SRS sites that were predicted using the RSSD sites. The combination of geo-referencing ECa, directed soil sampling data and regression modeling provides a tool for accessing manure accumulation on feedlot surfaces. This knowledge will aid in the development of BMP for mitigating environmental contamination.