Design criteria and accuracy specifications for precise large -scale topographic surveys from sparse data sets

James Arthur Elithorp, Purdue University

Abstract

The purpose of this research is to develop design and accuracy criteria for large-scale topographic surveys. The large-scale topographic survey is a ‘sparse surface model’ consisting of sampling prescription, surface modeling algorithm, and partitioned terrain complexity. Predictable survey outcomes can be achieved by partitioning the site into unitary land forms such that the amplitude of resident terrain microform is less than one-half the accuracy specification. A paraboloid of revolution can be used to model unitary land form. The limiting radius of surface curvature and the accuracy specification can be used to compute a sampling interval. Where the resident microform, without regard to its distribution, is less than one-half the accuracy specification, sampling prescriptions based on curvature provide predictable sampling outcomes. The sampling in such cases is accomplished on the trend surface without regard to the distribution of the microform amplitude. Where the amplitude of resident microform is greater than one-half of the desired accuracy specification, composite sampling is the result. Predictable sampling outcomes cannot be achieved in such cases, however the better results are achieved within the one-quarter mean wavelength where the microform can be shown to be periodic. Such sampling takes place on the surface of the microform without regard to curvature on the underlying trend surface. The Draft National Standard for Spatial Data Accuracy (NSSDA) vertical accuracy statistic has a high probability of failure on unitary land forms because of the lack of a measure of central tendency. The error distribution on a unitary land form needs a measure of central tendency to locate the center of mass of the error residuals. Unsymmetrical error distributions that occur with sparse data sets produce inflated confidence intervals because variance is a second-degree function. This fact can be mitigated by the use of the median and average deviation to locate and describe the mass of the error residuals. These smaller intervals will encompass from 87 to 91% of error residuals. The residuals in the tails can be reported with a count and magnitude range. This provides much more information about the large-scale topographic survey to the engineer than a single statistic.

Degree

Ph.D.

Advisors

Johnson, Purdue University.

Subject Area

Civil engineering|Earth

COinS