Sensitivities. Table 4.1 lists the sensitivity analyses which are proposed.
Sensitivities. Most of the parameters estimated during model calibration were related to hydraulic conductivity (horizontal hydraulic conductivity, horizontal-flow barriers, drain conductances, vertical anisotropy, and depth decay) (fig. 25 and appendix 1). In addition, a number of pilot points were added to better quantify storage properties for the transient simulation (xxxxx- xxx 1). Of the more than 500 parameters, all were estimated in the combined steady-state and transient simulation. At times during the calibration process, some of the defined parameters were not estimated because they were relatively insensitive at that point of the calibration. Sensitivity was measured by using composite-scaled sensitivities (CSS) that allow compari- sons of parameters of different units by scaling them to obtain quantities with the same units. CSS present the amount of information that observations provide to estimate the individ- ual parameter (Hill and Xxxxxxxx, 2007). Like DVRFS v. 1.0, recharge and depth decay of horizontal hydraulic conductivity are some of the most sensitive parameters in DVRFS v. 2.0, although some volcanic-rock and basin-fill hydraulic-con- ductivity, vertical anisotropy, and storage parameters are also among the most sensitive parameters. Parameter Correlation Parameter correlation coefficients (PCC) were used to eval- uate whether model parameters could be estimated uniquely by regression (Xxxx and Xxxxxxxx, 2007, p. 53). A PCC with an absolute value close to 1.00 indicates that the two param- eters involved likely cannot be estimated uniquely. Generally, absolute values greater than 0.95 are cause for concern, but values as small as 0.85 can affect the uncertainty of parameter estimates. Of the more than 500 parameters only 4 had PCCs greater than 0.95, all of them pilot points. All of these sets (parameter pairs) are negatively correlated, meaning that as one parameter is increased the other will decrease, and all of these param- eters have relatively low sensitivities, less than 1 x 10-3. The correlation of these four sets of parameters may not be able to be uniquely estimated. Two of the best most highly correlated parameter sets were for hydraulic conductivity in the basin- fill deposits in northwest Pahrump Valley, and two of the sets were for pilot point specific-yield parameters occurring in the basin-fill deposits near Indian Springs, Nevada: kvf_18 and kvf_22: -0.98 kvf_24 and kvsu_u_24: -0.99 pp_ss88 and pp_ss92: -0.99 pp_ss91 and pp_...
