Sensitivity, uncertainty, and reliability in groundwater modelling

Groundwater numerical models are powerful and efficient tools for groundwater management, protection, and remediation. However, groundwater modelling, which requires a huge amount of data, is not an easy. To build a predictive model, and to get reliable results, input data should be accurate and representative of the real situation in the field. Because of the randomness inherent in nature and heterogeneity of aquifers, it is very difficult to accurately determine the hydrological properties of the aquifers. Classical groundwater models usually handle input parameters in a deterministic way, without considering any variability, uncertainty, or randomness in these parameters. Thus, the results of deterministic modelling are questionable. To account for uncertainty in physical, chemical, and geological data, stochastic modelling is usually used. Many approaches have been developed and used including sampling approaches, reliability methods, and Monte Carlo simulation. In this chapter, different approaches of stochastic and probabilistic modelling are introduced and discussed.