On présente une analyse critique des modèles décrivant les transferts d'eau et de substances chimiques dans les sols partiellement saturés en eau. Pour la clarté de l'exposé, ces derniers sont classés en modèles déterministes (mécanistes et fonctionnels) et stochastiques. Les avantages et limites de ces différentes approches sont discutés. Finalement quelques recommandations et voies de recherche relatives à ce domaine sont suggérées.
- modèle déterministe,
- modèle stochastique
Modeling of solute transport in the vadose zone
The importance ofthe unsaturated (vadose) zone as an inextricable part of the hydrologic cycle has long been recognized. Theoretical and experimental studies on both water flow and solute transport in this zone have been further motvated by attempts to manage the root zone of agricultual soils optimally as well as by concerns about soil and groundwater pollution. These studies have greatly increased our conceptual understanding of the many complex and interactive physical, chemical and microbiological processes operating in the unsaturated zone. They have also led to a large number of models which vary widely in their conceptual approach and degree of sophistication, and are strongly influenced by the environment training and preoccupations of their developers.
A number of conceptual models for solute transport in partially saturated soils are reviewed and categorized. A key distinction is made between deterministic models which assume that a system behaves in such a way that the occurrence of a given set of events leads to a uniquely definable outcome, and stochastic models, which presuppose the outcome to be uncertain and are structured to account for this uncertainty. A second main distinction is between mechanistic and functional models. Mechanistic implies that the model takes into account the most fundamental mechanisms of the processes, as presently known and understood (e.g. Darcy's law for water flow, combination of mass-flow and difrusion-dispersion mechanisms for solute transport). The term functional refers to models that incorporate simplified treatments of solute and water and make no claim to fundamentality. However, their use requires less input data and computer expertise as compared to mechanistic models.
Additionally, it may be useful to distinguish between models that are primarily research tools (most of the mechanistic models) developed either to aid the testing of assumptions or to contribute to better understanding of the physical processes, and those (functional models) that are mainly useful as guides to the management of soil and water resources.
Although no attempt is made here at an exhaustive litterature review, the main features, the strengths and weaknesses of these approaches are presented and discussed. This analysis and other reviews published in recent years have revealed progress in many areas. Investigators have uncovered a number of inadequacies with existing models of soil transport processes and have made significant steps toward a better understanding of these phenomena. Some new research areas such as biodegradation modeling, immiscible phase transport, water and solute transfers in structured or swelling soils, and multi-interative ion transport are emerging.
A number of challenges still remain for both the theoretician and the practitioners. They include: i) how best to deal with preferential water flow and transport, ii) how best to model the effects of local and regional spatial and temporal variabilities of soil hydraulic properties on solute transport, iii) how to couple multi-component geochemical submodels efliciently with available unsaturated-saturated flow modes, iv) how to improve field methods for estimating vadose zone transport parameters and v) how best to predict the long-term consequences of short-term management decisions.
It is apparent that the complexity and variety ofthe physical processes have led to increasing specialization within the area of transport modeling. Soil physicists, soil chemists, soil microbiologists and agronomists have the propensity to limit their consideration and vision to their respective disciplines. A natural consequence of this specialization has been the evolution of scientific jargon specific to each subdiscipline. This may be overcome by reinforcing interdisciplinary cooperation among investigatory by training of students both at the graduate and postgraduate levels and by encouraging topical workshops and publications in interdisciplinary journals. Another general observation gleaned from this review is that very few solute transport models have been exhaustively tested under field conditions. Indeed, the quantitative criteria for validating models do not seem to be clearly identified or universally recognized. It appears very important that such criteria should be established and used to make an objective comparison of the abilities of the various types of model to simulate the results of field experiments. Without such tests and without comparisons between models there is a risk that disagreements between the predictions of different models and the resulting confusion could greatly diminish the usefulness of modeling techniques. While computer codes excalate in number as pressures mount for improved managements strategies it is time for asking the question: should the scientific community continue to develop more and more sophisticated general or even spectific models or should it put an emphasis on field experiments ? Obtaining an answer will probably be ofgreat importance in the near future. As a matter of fact, because of decreasing computer costs and relative increase in the cost of carefully designed field experiments, there is a worldwide trend to « observe the water and solute movements through computer screens » ! The modelers should be aware that without reliable estimates of the input parameters as well as in-situ validation their models will appear more as intellectual games of academic interest rather than as tools to help the practitioners in their management decisions.
- deterministic model,
- stochasiic models