Kriging of Lognormal Stochastic Field
| Autor: | Masaru Hoshiya, Shigeru Noda |
|---|---|
| Rok vydání: | 1998 |
| Předmět: | |
| Zdroj: | Journal of Engineering Mechanics. 124:1175-1184 |
| ISSN: | 1943-7889 0733-9399 |
| DOI: | 10.1061/(asce)0733-9399(1998)124:11(1175) |
| Popis: | The methods of simple kriging (SK) have been studied by many researchers for interpolation and extrapolation of a Gaussian stochastic field conditioned on observations at a number of points, and basic findings on theory are made clear including techniques of simulating a conditional sample field. This study discusses how to extend SK to a lognormal stochastic field, whose mean field and covariances are prescribed, and it points out some important differences in the results from those for a Gaussian stochastic field. Next, universal kriging (UK) is proposed for the same lognormal field, where its mean field is unknown. It is found that the SK estimator with known mean and covariances is superior to the UK estimator, since the latter does not have specification of the mean function, where the kriging variance cannot be evaluated and only a variation indicator can be obtained. Finally, a numerical example of a one-dimensional lognormal stochastic field is demonstrated to show the characteristic differences b... |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|