Text S1. Model calibration

Gerke et al. (2018) calibrated this method. In addition, we explored the effect of resolution with Test 4 as an example with voxel size of ~2.3 and ~4.6 µm. The difference between two simulations is <3% for intrinsic permeability, <2% for effective permeability, and <2% for normalized permeability. This indicates that the trends demonstrated in this study (Figure 2) is credible.