Back to Contents4.3 Soil Sampling
Several conditions are essential for successful site-specific management of crop production. The most basic condition is that the variation in soil properties be known with some degree of accuracy (Pierce and Nowak, 1999). Poor map quality and the use of an inadequate mapping scale may explain why the results of some site-specific management agronomic and economic studies have had mixed or negative results. Map accuracy and quality depends on sampling method (grid or zone sampling), scale (sampling intensity), analytical laboratory errors, and prediction errors (interpolation). Because various forms of these techniques are being applied on a commercial basis in Kentucky, it is critical to determine the most appropriate sampling strategies to assess soil variability in this state.
Because some studies have shown that grid sampling may be cost prohibitive researchers have turned to using co-varying secondary information to reduce the required number of samples necessary to achieve a greater level of accuracy (Mueller, 1998). This secondary information includes terrain attributes (e.g. slope, aspect, and curvature; all derived from an elevation map), electromagnetic conductivity, and ground penetrating radar. Because it is one of the soil-forming factors terrain contains a great deal of information about soil properties. Advances in GPS technologies have allowed for very accurate measurements of elevation within agricultural fields. Electromagnetic conductivity sensors (Veris Technologies, Geonics Inc. EM38) have been found to be related to soil cation exchange capacity (CEC), depth to soil claypan, water holding capacity, organic matter, and salinity. Ground penetrating radar has been found to be a reliable way to measure soil and water table depth. Geostatistical techniques such as cokriging or kriging with an external drift allow this secondary information to be used to improve spatial estimates of soil properties.
Yield monitors can be used to identify regions within a field that may have an economic response to changes in soil and crop management. Therefore it is important to understand factors that affect the spatial and temporal variability of grain yield (e.g. soil type, soil physical, chemical, and biological properties). Since these factors are difficult to measure economically while in the field, it is important to determine how sensed information (remote sensing, digital terrain models, electromagnetic conductivity, and ground penetrating radar) relate to these factors, and ultimately, grain yield.
Another cost-effective alternative to grid sampling is composite sampling within zones. These zones should be based on established differences within a field based on soil map delineations, management history, yield potential maps, or aerial imagery. Unfortunately, NRCS soil surveys were not created for the purpose, hence are not at the scale necessary for site-specific management of agricultural fields. Smaller scale soil surveys may provide better map unit delineations for zone sampling. Furthermore, digital terrain information, electromagnetic conductivity sensors, and ground penetrating radar may reduce time and costs associated with creating a first order soil survey while improving overall accuracy.