Submitted to: Intnl Conference On Geospatial Information In Agriculture And Forestry
Publication Type: Proceedings
Publication Acceptance Date: 11/7/2001
Publication Date: 11/7/2001
Citation: Jung, W., Kitchen, N.R., Sudduth, K.A. 2001. Evaluating claypan soil productivity using sensors and soil sampling. In: Proceedings Third International Conference on Geospatial Information in Agriculture and Forestry. Ann Arbor, Michigan. CDROM. Interpretive Summary: Crop yield is affected by many factors, including soil, weather, and agronomic management practices. Farmers have expressed interest in knowing how soil resources might be better managed to improve crop production with greater use efficiency of management inputs, such as fertilizer and seed. This investigation was conducted to pinpoint what soil and landscape properties were the most dominant in reducing yield for poorly drained claypan soils in Missouri. We measured standard soil fertility properties at 282 points in a 34-acre field to assess their effect on crop production. In addition, sensor-based soil electrical conductivity, slope, and elevation measurements were obtained. We found from the soil fertility samples that soil pH and base cations could only explain up to 30% of yield variability. With sensor-based information included, however, up to 60% of the yield variability could be accounted for. This finding supports our belief that having sensor-based information (of which many more measure- ments can be obtained with given time and expense constraints) will be the most valuable in helping producers understand yield maps. These results show that sensor-based soil property measurements alone explain yield variations within a field much better than standard soil fertility measurement. These results will be useful to producers and crop consultants to improve site-specific nutrient management plans.
Technical Abstract: Understanding the relationship between crop yield and field properties is essential for developing site-specific field management. This study was conducted to understand relationships between corn and soybean yields and soil and landscape properties. We collected 282 soil samples on a 25-m grid spacing to a 15-cm depth from a 14-ha MO claypan soil field in 2000. Soil samples were analyzed for texture, total N, pH, neutralizable acidity, OM, P, Ca, Mg, K, and CEC. Soil profile electrical conductivity (EC), slope, elevation, and yield were also obtained on the same grid. From principal component (PC) analysis using soil chemical and physical properties, the first 4 principal components explained 84% of the data set variance. These PC variables were affected primarily by base cations, acidity, soil EC, soil particle size, and elevation. Stepwise multiple regression analysis was performed on yield using both the actual soil properties and then using gPC variables. While PC analysis removed colinearity between soil measure- ments, regression results of PC variables generally gave much lower R**2 values in soybean crop-years and similar R**2 values in corn crop-years when compared to multiple regression using the individual soil properties. In all, 10 to 40% of soybean yield variability and 20 to 60% of corn yield variability could be accounted for using these measuresments. With both crops, soil EC, elevation, and slope were most often the first variables included in the stepwise models indicating they predominated over the soil fertility variables in accounting for yield variation. These results support our belief that having sensor-based information (of which many more measurements can be obtained with given time and expense constraints) will be the most valuable to producers in understanding yield variability.