Saturation-Dependent Hydraulic Conductivity From Critical Path Analysis: Comparison with the Unsoda Data Base and Van Genuchten Model.

We have analyzed predictions of critical path analysis (CPA) for the hydraulic conductivity of unimodal pore-size distributions by investigating their predictive capabilities for the UNSODA data base. For comparison we also checked the accuracy of the van Genuchten model. We found that at values of the hydraulic conductivity exceeding (10^-3 cm/day) the CPA calculations performed best. We have traced the majority of the problems of the CPA calculations at lower moisture contents to the inaccurate extraction of the PSF model parameters from the soil water retention curve. The remainder of the discrepancies can be attributed to the relevance of multi-modal pore size distributions. Interestingly, the reduced sensitivity of the van Genuchten model to the exact pore-size distribution causes its performance to be less degraded by use of an inappropriate model, allowing it to perform better at low moisture contents and low hydraulic conudctivity values. Conversely the van Genuchten results cannot be improved as significantly by modeling the pore space more accurately.  We thus argue that reliance on the van Genuchten treatment tends to remove the motivation to start with the best model.