Using a model’s purpose to guide model updates: The case of the Mississippi Alluvial Plain aquifer

Uncertainty and data-worth analyses evaluate the reliability of computer model outputs and estimate which model inputs are contributing most to model output uncertainty. These analyses direct decision makers to well-informed decisions and allow practitioners to estimate data importance for reduced model output uncertainty.

These techniques have been applied to the existing Mississippi Embayment Regional Aquifer System (MERAS) model to guide data collection and model dataset construction for the proposed Mississippi Alluvial Plain (MAP) aquifer model. The focus of the MAP model is to forecast aquifer water levels and surface-water/groundwater (SW-GW) exchange under different climatic and water-use conditions. Forecasts of interest (FOIs) simulated by the MERAS model include water levels and SW-GW exchange along major surface-water features under conditions of less recharge and increased water use. Preliminary results from the uncertainty analysis indicate streambed conductance values may better constrain uncertainty in water-levels and streamflow, which can be used to guide data collection in the field. These results will be presented and discussed in the context of data collection and model data preparation.