DETERMINATION OF VARIANCES FOR MAXIMUM AND MINIMUM TRANSMISSIVITIES OF ANISOTROPIC AQUIFERS FROM MULTIPLE WELL PUMPING TEST DATA

An extension of the least squares method given by Neuman et al. (1984) for estimating aquifer parameters is presented, and statistical tests for anisotropy are derived. Methods are also given to construct varia nce formulas for the principal axes of transmissivity and confidence i ntervals for the angles of orientation. The techniques developed here treat each monitoring well as the experimental unit, in contrast to in verse modeling methods which treat each time step for each well as the experimental unit. We argue that the latter is a form of pseudo-repli cation in that the number of time steps may be arbitrarily selected. T he example aquifer test data used by Neuman et al. (1984) are analyzed . The maximum transmissiviy was found to have a magnitude of 8.60 +/- 5.30 m(2)/day oriented N29.6 degrees W +/- 11.3 degrees at the 95% con fidence level with a ratio of anisotropy of 1.91 +/- 0.39. Both statis tical tests indicated anisotropy. Results from a multiple well pumping test from the Powder River Basin, Wyoming, are also analyzed, The max imum transmissivity was found to have a magnitude of 230 m(2)/day +/- 84.4 m(2)/day oriented N51.7 degrees W +/- 5.3 degrees at the 95% conf idence level with a ratio of anisotropy of 2.12 +/- 0.48. Both statist ical tests indicated anisotropy. The methods given in this paper can b e performed with standard statistical software packages and spreadshee ts. A program that uses Matlab(R) software is available from the autho rs to perform the necessary calculations at .com/pub/users/FTP/west-in c/groundwater/adap1.zip.