Investigations on Statistics of Terrain Height.
Final rept. Jan-Oct 84,
NORTHEASTERN UNIV BOSTON MA
Pagination or Media Count:
Three related but independent analytical investigations are reported in this document. The first is investigation of a binary hypothesis test to determine whether a given population of terrain height samples is more likely to have originated from a Gaussian or an expotential probability distribution. The input data for the test are assumed to be finitely sampled and quantized. Two tests are considered, one based on the assumption of continuously varying input data and the other based on finitely sampled and quantized data. Bivariate distributions with multiple observations are assumed. The second investigation concerns two classes of interpolation problems. The first is the inference of terrain height values at points for which no direct measurements are available using measured height values at other points, and using maximum likelihood estimator theory as a basis for the interpolation. The second is that of inferring the joint PDF of terrain height values of points at which height measurements have not been made. In both of these investigations, the effect of external constraints on heights and height differences are also considered. The third investigation concerns a continuation of a study initiated on a previous contract, sampling, and quantization errors in estimator of mean, variance and covariance of terrain heights. Author
- Statistics and Probability