It is often a challenge to simulate soil moisture dynamics accurately when limited data are available about the soil. A forward simulation based on pedotransfer function (PTF) estimates of hydraulic parameters indeed produced severely biased moisture fluxes. This study therefore evaluated a new model inversion strategy which is able to use limited data by optimizing factors that scale PTF estimates. An important feature of this strategy is that the model inversion is performed for all profiles simultaneously, thus ensuring that the hydraulic parameters among the different profiles are adjusted at the same time by changing the scale factors while also ensuring that sufficient data are available to make inversion mathematically possible. We evaluated three main inversion models of different complexity. The simplest model (GW5), in which there were five scale factors with no horizon distinction was the most parsimonious in terms of the corrected Akaike’s Information Criterion (AICc). GW10 (with two sets of five scale factors for topsoil and subsoil), and GW30 (six sets) performed less well in terms of AICc, but were able to simulate the mean observed moisture profile much better than GW5. Additional inversions demonstrated that textural variation was important to achieve the best performance, thus indicating that the PTF estimates are a necessary part of the inversion. The GW5 model parameters appeared to be insensitive to assumptions made about total root water uptake and its distribution within the profile. The inversion method is simple and can be interfaced without much difficulty with existing vadose zone models.
ASJC Scopus subject areas
- Soil Science