While sensitivity analysis and calibration are common practice in integrated hydrologic modeling, little work has been done to understand how the design of the sensitivity analysis and calibration affects the simulation outcome in these often highly nonlinear models. This is especially true for irrigated agricultural basins with a strong connection between land use, groundwater, and surface water. Using a range rather than a single set of initial parameter values, multiple sensitivity analyses, calibrations, and linearity tests were performed using UCODE_2014 on the Scott Valley Integrated Hydrologic Model. Calibration results show that parameters related to crop demand and applied irrigation water are most sensitive. Influence statistics show that low streamflow observations provide the most information during model calibration, indicating preference should be given to these observations during model development, sensitivity analysis, and calibration. Importantly, due to the nonlinearity of the integrated model, significant differences are found in results when initial parameter values are sampled from within their respective expected ranges. Estimates for some parameters varied up to an order of magnitude between calibrations, while all produced similar final objective function values, groundwater elevations, and stream flow. Confidence intervals for individual sensitivity analyses and calibration runs only spanned a fraction of the ensemble estimated parameter range across multiple runs. Our work suggests that a calibration design with multiple sensitivity analyses and calibrations of integrated hydrologic models, each using one of several widely varying sets of initial values, provides a frugal approach to identify parameters across the global parameter space.