The environmental models are mostly very complicated and virtually unbounded. This complexity results in uncertain modelling results. The two main sources of uncertainty are model and parameter uncertainty. The available mathematical and numerical groundwater models may not represent the exact natural process in the groundwater aquifers. The material properties are poorly known due to limited number of measurements and the natural high heterogeneity in soil properties.
The groundwater model parameter values are generally estimated using calibration processes using limited number of observations. The real-world systems are highly complex and even large calibration databases can provide multiple estimates of all system parameters which are equally probable. This is called non-uniqueness issue.
There are various methods presented in the literature to quantify the groundwater model uncertainty and/or estimate the reliability/possibility of the outcomes of the groundwater model. This paper tests the Iterative Ensemble Smoother (IES) tool in a regional scale real-work groundwater model with hundreds of parameters. The IES is an open-source and model independent tool which was developed to overcome the computational burden associated with matching large database of groundwater history in real-world scale environmental models. IES is a tool to quantify the uncertainty in highly dimensional parameter spaces. This paper explains the application of IES in real-world groundwater models and its effectively and efficiency to estimate posterior forecast uncertainty when the future projection involves large number of parameters.
The model independent IES is an emerging technology which enables environmental modellers to account for model input uncertainty at realistic spatial and temporal scale. The application of this technology will lead to better forecast uncertainty estimation and improve the usability of environmental models in decision making. The results of this case study can help environmental managers to select informed decisions by having reliable information about likelihood of environmental phenomena.