Process-based numerical modeling of subsurface flow is an important tool in hydrogeological research and groundwater-resources management. The computational power (in terms of hardware and software) has improved drastically over the last decades. At the same time, models have increased in complexity, as new numerical methods have become feasible, more processes can been considered and evermore detail can be represented in the models (e.g., Venkataraman & Haftka, 2004;Y. Zhou & Li, 2011;Jakob, 2014). A side effect of this development is that modern models tend to have many adjustable parameters. The process of estimating values of model parameters so that the model output reasonably agrees with measured data is known as model calibration, inverse modeling, or parameter inference (e.g.,