Development of Soil Test Calibrations Using Precision Agriculture Technologies.

For decades, soil-test correlation and calibration have been based on conventional plot research. Precision agriculture technologies allow for much easier variable-rate nutrient application and consideration of within-field soil-test and crop yield variation. Especially important has been increased adoption of accurate Real Time Kinematic (RTK) GPS systems and improvements of yield monitors and variable-rate technology. This presentation discusses already proved use of these technologies for soil-test correlation and possibilities for calibration. Usually there are different soils, topographic positions, soil-test values, and yield potential within fields that could be considered as different "sites". However, there is uncertainty about ways for collecting and using data for soil-test correlation and calibration. In Iowa we developed and implemented a method to adapt the new technologies to classic, weigh-wagon, replicated, strip trials for soil-test correlation. As a minimum, soil-test correlation requires a non-fertilized control and a rate that maximizes yield but is not excessive. Therefore, even large custom fertilizer applicators can be effectively used to apply a fixed fertilizer rate to field-length strips. Other than GPS and yield monitors, the method involves a dense soil sampling approach adapted to the experimental layout, a control and a high but not excessive fertilizer rate, and GIS to calculate absolute and relative yield responses for small areas (0.1 to 0.2 ha) and for digitized soil maps. Soil-test calibration requires multiple nutrient application rates, however. Although strip trials with several fixed adjacent fertilizer rates or ramped techniques that deliver a range of rates along strips length can be used for soil-test calibration, this objective requires very careful consideration of the size of the area fertilized with each rate, soil-test small-scale variability, border effects, and the accuracy of fertilizer applicators and yield monitors. For both objectives, no single data management procedure can or should be recommended for all conditions. The results of implementing these techniques do show, however, that replication, several fields and years, and careful interpretation of response models are required to develop reliable soil-test interpretations.