Optimal sampling strategies for mapping soil types. II. Risk functions and sampling intervals |
| |
| Authors: | T M BURGESS R WEBSTER |
| |
| Institution: | Schlumberger Cambridge Research, PO Box 153, Cambridge CB2 3BE;Rothamsted Experimental Station, Harpenden, Herts AL5 2JQ |
| |
| Abstract: | This paper develops the concept of the quality of a soil map in terms of the proportion of soil boundaries that have remained undetected as a result of point sampling. This proportion is equated to the risk of omitting such boundaries during survey. A risk function is defined, and the optimal strategy is such that the risk is maintained at some predetermined acceptable level. In general the risk depends on the distribution of boundary spacings, the distance travelled from the last boundary and the projected sampling interval. Where the distribution is exponential, the position of the last encountered boundary is immaterial, and for a constant risk the sampling interval is also constant. The optimal strategy is a regular grid. The same principles enable the total sampling effort for a survey of constant risk to be forecast once the distributions of boundary spacings have been estimated from reconnaissance. The sampling effort for a risk of 0.1 is shown to be within the range of traditional practice. |
| |
| Keywords: | |
|
|
