Mapping soil water retention curves via spatial Bayesian hierarchical models

作者：Yang Wen Hsi^*; Clifford David; Minasny Budiman

来源：Journal of Hydrology, 2015, 524: 768-779.

DOI：10.1016/j.jhydrol.2015.03.029

• Water retention curve
• Spatial van Genuchten model
• Bayesian hierarchical modeling
• Markov chain Monte Carlo
• Spatial confounding
• WinBUGS

Soil water retention curves are an important parameter in soil hydrological modeling. These curves are usually represented by the van Genuchten model. Two approaches have previously been taken to predict curves across afield - interpolation of field measurements followed by estimation of the van Genuchten model parameters, or estimation of the parameters according to field measurements followed by interpolation of the estimated parameters. Neither approach is ideal as, due to their two-stage nature, they fail to properly track uncertainty from one stage to the next. In this paper we address this shortcoming through a spatial Bayesian hierarchical model that fits the van Genuchten model and predicts the fields of hydraulic parameters of the van Genuchten model as well as fields of the corresponding soil water retention curves. This approach expands the van Genuchten model to a hierarchical modeling framework. In this framework, soil properties and physical or environmental factors can be treated as covariates to add into the van Genuchten model hierarchically. Consequently, the effects of covariates on the hydraulic parameters of the van Genuchten model can be identified. In addition, our approach takes advantage of Bayesian analysis to account for uncertainty and overcome the shortcomings of other existing methods. The code used to fit these models are available as an appendix to this paper. We apply this approach to data surveyed from part of the alluvial plain of the river Rhone near Yenne in Savoie, France. In this data analysis, we demonstrate how the inclusion of soil type or spatial effects can improve the van Genuchten model's predictions of soil water retention curves.

• 出版日期2015-5
• 单位CSIRO