Author
 |
PACHEPSKY, YAKOV - DUKE UNIVERSITY |
|
Acock, Basil |
|
Submitted to: Geoderma
Publication Type: Peer Reviewed Journal Publication Acceptance Date: 10/7/1997 Publication Date: N/A Citation: N/A Interpretive Summary: The techniques of precision agriculture allow farmers to manage separate areas of land with different properties within a field. To do this they sample the soil in a grid to determine its properties. However, they really need to know soil properties in locations between the points they can afford to sample. The techniques used to estimate these intermediate points all introduce uncertainty. To determine how much uncertainty, we generated a detailed soil map for an imaginary field. Then we sampled our field using a 50x50m and a 100x100m grid, corresponding to typical research and commercial soil sampling densities respectively. We used two new techniques called genetic algorithms and stochastic imaging to generate many equally probable maps of soil properties for the field based on these samples. A soybean crop model GLYCIM was used to simulate crop yields for three years of weather data. We compared the calculated yields for each map with the yields for our original detailed map, and calculated the errors as the difference between them. Results showed that the errors in yield estimates were affected by weather pattern, and that neither the commercial nor the research sampling densities were high enough to adequately characterize the soil in the field. Although this was a purely theoretical exercise, it shows the type and magnitude of errors that will be encountered in real fields. It could also be used with real field data to assess the efficiency of a proposed sampling density. However, our results show that farmers cannot afford to sample soil densely enough to characterize it, so we must find other ways of estimating soil properties. Technical Abstract: Site specific agriculture requires estimates of soil properties in locations other than the sampling points. Techniques are needed to assess the uncertainty of these estimates. Such uncertainty assessments can be based on stochastic imaging of soil parameters: a technique that consists of generating many equiprobable maps of the parameters for the same site. The objective of this study was to use stochastic imaging of the available soil water capacity (AWC) and a soybean crop model GLYCIM to simulate variability and uncertainty in crop yield estimates as related to soil sampling density and weather patterns. First, we generated an AWC data set on a 25x25m (fine) grid, simulated yields at the fine grid nodes, and considered the results as the "true" yield values. Then we sampled the fine grid using sparse grids of 50x50m and 100x100m corresponding to typical research and commercial soil sampling densities respectively. We carried out stochastic imaging of AWC using genetic algorithms, simulated yields for each image, and calculated the errors in yield estimates as the difference between the "true" yields and yields from the images. The probability distributions of the errors were used to quantify the uncertainty. The simulations were repeated for three different weather patterns. Results showed that the distributions of errors in yield estimates were affected by weather pattern, and the temporal variability in yield error estimates could not be overridden by improvements in spatial variability estimates at the sparse sampling densities that we considered. Stochastic imaging of soil properties enables us to assess the efficiency of a sampling density to be used over several years. |
