|BIRRELL, STUART - UNIV OF MO
|Sudduth, Kenneth - Ken
|KITCHEN, NEWELL - UNIV OF MO
Submitted to: International Conference on Precision Agriculture Abstracts & Proceedings
Publication Type: Proceedings
Publication Acceptance Date: 12/1/1996
Publication Date: N/A
Interpretive Summary: Site-specific management, or precision farming, is a strategy in which cropping inputs such as fertilizers are applied at varying rates across a field in response to variations in crop needs. Variations in crop nutrient needs across a field are generally determined by laboratory analysis of soil samples collected on a regular grid pattern. Accurate nutrient maps require the collection and analysis of many samples from each field. The costs of sample collection and analysis, however, limit the number of samples a producer is willing to obtain. In this work we intensively sampled a 70 acre field to investigate the impact of soil variability and sample spacing on the accuracy of the nutrient maps obtained. Maps developed from soil samples taken on a 330 ft grid (a standard in the industry) were compared to maps from samples taken on a 82 ft grid. There were significant differences between the maps, and using the 330 ft maps for fertilizer application would result in large areas of the field receiving either too much or too little fertilizer. This work is significant because it indicates that, at least on the field studied, standard soil sampling practices are not adequate for measuring variations in crop nutrient needs. This finding may impact future agribusiness practices for soil sampling for precision farming.
Technical Abstract: A 28 ha field was soil sampled on a 25 m grid to investigate the mapping implications of short-range soil variability. Additional samples were taken at the 100 m grid intersections and at other locations near those intersections. Geostatistical analysis and Kriging were used to develop soil nutrient maps from the complete 25 m dataset and from the smaller datasets obtained on a 100 m grid. The range of spatial influence in this dataset was approximately 100 m for potassium, phosphorus, and soil pH. However, using data from only the 100 m grid resulted in unreliable semivariograms, as these datasets did not include enough data points. Soil potassium maps generated from 100 m grid data were relatively similar to those generated with the complete 25 m dataset. However, phosphorus and soil pH maps for the two datasets were quite dissimilar. At least for the study field, soil nutrient variability was not adequately characterized when samples obtained on a 100 m grid were used to generate nutrient maps.