In this paper, we present a stochastic-analytical approach for uncertainty modeling in two-dimensional, statistically nonuniform groundwater flows. In particular, we develop simple closed-form expressions that can be used to predict the variance of Darcy velocities caused by random small-scale heterogeneity in hydraulic conductivity. The approach takes advantage of the scale disparity between the nonstationary mean and fluctuation processes and invokes an order-of-magnitude analysis, enabling major simplifications and closed-form solutions of the nonstationary perturbation equations. We demonstrate the accuracy and robustness of the derived closed-form solutions by comparing them with the corresponding numerical solutions for a number of nonstationary flow examples involving unconfined conditions, transient conditions, complex trends in mean conductivity, sources and sinks, and bounded domains.
