Scientific Investigations Report 2006–5316
U.S. GEOLOGICAL SURVEY
Scientific Investigations Report 2006–5316
DOE/ID-22201
The relative sediment content of the eastern Snake River Plain aquifer is a key geologic variable that affects the assignment of hydraulic conductivity values during calibration of ground-water flow models. Geostatistical analysis was used to estimate sediment content in the uppermost 300 ft of the U.S. Geological Survey (USGS) subregional ground-water flow model in order to improve future model calibrations.
The statistical characteristics of the sediment content of the aquifer, as measured in 333 boreholes located across the Idaho National Laboratory, were analyzed using conventional (nonspatial) and spatial statistical methods. Borehole data were segregated into various groups by stratigraphic age, depth, and geographic location in order to determine the degree of statistical stationarity that characterizes the model domain. The results indicate that the stratigraphic distribution of sediment within the aquifer, when evaluated over sufficiently large intervals of time (200 thousand years or greater), appears to be statistically stationary in a vertical sense as well as in a geographic sense. This finding makes it possible to model the spatial distribution of sediment in the uppermost three layers of the USGS subregional ground-water flow model using two-dimensional kriging within individual layers.
Multiple indicator variogram analysis and multiple indicator kriging (mIK) were used to model the spatial distribution of sediment. The use of mIK has three principal advantages over ordinary kriging: (1) local uncertainty of the kriging estimate is quantified more accurately than is possible in ordinary kriging; (2) differing degrees of spatial autocorrelation can be used to describe sediment-rich and sediment-poor areas; and (3) local frequency distributions that characterize a layer’s sediment contents can be used to delineate hydraulic-property zones that depend on sediment content.
The kriging approach honors all available borehole sediment data, provides better spatial resolution for delineating zones of relative sediment abundance, and provides a basis for making parameter assignments that reflect the uncertainty of the kriged sediment estimates. By delineating hydraulic conductivity zones with mIK and considering kriging uncertainty in the model calibration process, the ultimate benefit of such an approach will be to more tightly constrain parameter assignments in data-rich areas. This approach could be incorporated in future development of the USGS subregional ground-water flow model and development of methods for integrating the uncertainty associated with zonal boundaries into the parameter estimation process.
U.S.
Department of the Interior | U.S. Geological
Survey
Persistent URL: https://pubs.water.usgs.gov/sir20065316
Page Contact Information: Publications Team
Page Last Modified: Thursday, 01-Dec-2016 19:33:15 EST
