Regional groundwater flow models play an important role in decision making regarding water resources; however, the uncertainty embedded in model parameters and model assumptions can significantly hinder the reliability of model predictions. One way to reduce this uncertainty is to collect new observation data from the field. However, determining where and when to obtain such data is not straightforward. There exist a number of data‐worth and experimental design strategies developed for this purpose. However, these studies often ignore issues related to real‐world groundwater models such as computational expense, existing observation data, high‐parameter dimension, etc. In this study, we propose a methodology, based on existing methods and software, to efficiently conduct such analyses for large‐scale, complex regional groundwater flow systems for which there is a wealth of available observation data. The method utilizes the well‐established d‐optimality criterion, and the minimax criterion for robust sampling strategies. The so‐called Null‐Space Monte Carlo method is used to reduce the computational burden associated with uncertainty quantification. And, a heuristic methodology, based on the concept of the greedy algorithm, is proposed for developing robust designs with subsets of the posterior parameter samples. The proposed methodology is tested on a synthetic regional groundwater model, and subsequently applied to an existing, complex, regional groundwater system in the Perth region of Western Australia. The results indicate that robust designs can be obtained efficiently, within reasonable computational resources, for making regional decisions regarding groundwater level sampling.

<p>Setting groundwater allocation limits requires an understanding of recharge fluxes to the aquifer system. Very often rainfall percolation through the subsurface represents the critical recharge flux. In this groundwater limit setting context, recharge estimates are often established as a component of the groundwater flow model history matching process. Typically, there are many recharge models available, and the basis for selecting any particular model is often confusing.</p><p>Some of these recharge models are numerical solutions of variably saturated pressure head and flow and represent the full complexity of the soil-vegetation-atmosphere transfer of water. Such models require many parameters that may not be measured or verified, and/or are computationally expensive. This can make the history matching process and predictive uncertainty analyses difficult.</p><p>Simpler model representations of the recharge processes are also available, either through upscaling (e.g., by lumping together different soil profiles, with different vegetation) and/or by simplification of the recharge estimation method. Such simplifications may involve empirical equations to derive gross recharge, single bucket-type root zone water balance calculations, or solving net recharge with the help of analytical solutions of flow or pressure heads and linear approximations of gross recharge or evapotranspiration from groundwater as function of the groundwater head. These simpler models often have a greater utility (i.e., they are quicker to run and are more numerically stable) but may be accompanied by additional ‘simplification’ induced uncertainty.</p><p>Regardless of the method used, the uncertainty and bias of these recharge predictions can be high.  The uncertainty of groundwater model predictions underpinning the setting of allocation limits can also be high. However, the performance of a recharge model in terms of how it impacts the reliability of the predicted impacts relevant to the groundwater allocation limit, is currently not considered. This study addresses this issue, exploring the costs and benefits of recharge models of varying complexity, in the context of setting groundwater abstraction limits. This is demonstrated using a synthetic, but realistic case study in Western Australia.</p><p>We adopt a paired complex-simple model analysis workflow, and implement it using the Flopy-PyEMU Python-based scripting framework. This workflow is then used to explore the performance of more complex and simpler models within the groundwater allocation management context by measuring each model’s bias and uncertainty. We compare a cell-by-cell Richards’ equation-based recharge model, with a series of simpler recharge contender models. This scripted workflow supports the efficient deployment of the paired complex-simple model stochastic analysis and interpretation of its outputs.</p>