Submitted to: Unsaturated Zone Modeling: Progress, Applications and Challenges
Publication Type: Book / Chapter
Publication Acceptance Date: June 15, 2004
Publication Date: October 4, 2004
Citation: Pachepsky, Y.A., Smettem, K.J., Vereecken, H. 2004. Reality of models and fiction of data in soil hydrology. Unsaturated Zone Modeling: Progress, Applications and Challenges. pp.233-260.
Interpretive Summary: While predicting water movement in soils and landscapes, one would like to use the best model. Developments in instrumentation and data accumulation show the notion of 'the best' model may be wrong. Complexity of flow pathways may be easily perceived but difficult to represent in mathematical terms without making strong simplifying assumptions. This implies that many vastly different models can be consistent with observations. To illustrate typical choices made in modeling practice, we use a case study of using integrated data to build a model of ground water pollution for a watershed. Remote sensing, geophysical, and topographic data present opportunities to represent in models the spatial variation in soil, vegetation, and land use. As those data are not directly related to the ability of soils to transmit and retain water, they have to be converted into soil hydraulic properties. The conversion equations, known as pedotransfer functions, and prospects of their improvement, are discussed. Pedotransfer functions are derived from small-volume soil data, whereas hydrologic models have to use relatively large volumes of soil to calculate water transport in practically important cases. We show that the ability of soils to transmit and retain water in such large volumes cannot be deduced by simply combining many small soil volumes, and discuss current options to upscale the small-volume data. We note the paucity of literature on use and comparison of several models, and the absence of methods for comparative model evaluation. Given the current status of hydrologic models and data, it is best to use a variety of alternative models.
The objective of this chapter is to contribute to the ongoing discussion on strengths, weaknesses, opportunities and trends of existing modeling approaches in soil hydrology. The prevailing attitude in hydrology has been that if the modeling concept is sufficiently accurate, it should be possible to represent the site uniqueness with a specific set of model parameters. Complexity of models and detail of data have grown at increasing pace. Recent advances in instrumentation have revealed a complexity of flow pathways that may be easily perceived but difficult to represent in mathematical terms without making strong simplifying assumptions. This implies that many different model structures and calibrated parameter sets could be consistent with available observations. The multiplicity of models and the parameter deficit are the emerging issues that present both obstacles and opportunities for hydrologic modeling. We present a comprehensive case study on using integrated data to model groundwater pollution for a watershed, and use this case study to illustrate current opportunities in quantifying soil variability with remote sensing, geophysical methods, and topographic information. The value of pedotransfer functions and publicly available databases is discussed. Mismatch between measurement and modeling scales creates the need in incorporating scale effects in the hydrologic models. Techniques for the comparative evaluation of models need to be developed and tested. In absence of unique model selection criteria, it can be therefore best to consider a variety of alternative models based on reasonable alternative hypotheses.