A procedure is developed to estimate the spatial variability of soil properties in 3-D from irregularly spaced data. Two basic components of the soil property data are modeled: (1) the global trend which is non-stationary and is estimated using multiple linear regression analysis for the raw soil property data; and (2) the local trend which contains a stationary portion that is characterized by its autocorrelation structure. Autocorrelation between residual values is determined and the correlation structure is modeled using multiple linear regression analysis. Local trend is estimated using a kriging process that utilizes the autocorrelation structure. The estimation model was applied to a set of Dutch-cone penetration test data. Analysis on the residual values showed that autocorrelation was weak, which resulted in a large estimated variance for the estimated mean property value. It was concluded that better soil property estimates could be obtained using smaller sampling intervals. The model can also be used in the design of soil exploration programs.
|Original language||English (US)|
|Title of host publication||Unknown Host Publication Title|
|Publisher||Inst for Risk Research|
|State||Published - Dec 1 1987|
ASJC Scopus subject areas