|Sudduth, Kenneth - Ken|
Submitted to: European Conference on Precision Agriculture Proceedings
Publication Type: Proceedings
Publication Acceptance Date: 6/20/2001
Publication Date: N/A
Interpretive Summary: Soil drainage affects plant growth, water flow, and solute transport in soils. Accurate and inexpensive prediction and mapping of soil drainage class (SDC) would be beneficial for both agricultural and environmental management. In this study, we examined the use of topographic data and soil electrical conductivity (EC) for predicting SDC. We chose these data sources because they are easily obtained with automated sensors. Topography and EC data from a central Illinois field were analyzed using a statistical technique called discriminant analysis. We also used two geostatistical methods, kriging and cokriging. These two methods take into account not only the data values but also the location of the data points within the field. Because they include location information, geostatis- tical methods are generally preferred for developing maps of precision agriculture-related data. We found that the most accurate mappings of SDC could be obtained by combining discriminant analysis with cokriging. Because a number of agronomic and environmental decisions are based in part on SDC, this new approach to estimating SDC will be of use to producers, crop advisors, extension personnel, and environmental management professionals, such as NRCS.
Technical Abstract: Multivariate methods that utilize easily obtained topographical and soil information can be of substantial value for delineation of zones. We applied discriminant analysis (DA) and geostatistics to create soil drainage maps with topographical and soil electrical conductivity (EC) data as sources of auxiliary information. Among the studied variables, soil EC, ,terrain slope, and distance to drainage-way affected soil drainage the most. When these variables were used as additional information in predicting soil drainage classes (SDC) using either discriminant analysis or cokriging procedures, they improved prediction accuracy, comparing with the indicator kriging that used only drainage data and with the soil survey map. For determining zones, we propose to use stepwise DA to select the secondary variables with the highest influence on soil drainage and, second, to create drainage maps using indicator cokriging with the selected dvariables.