Need for Process Based Empirical Models for Water Quality Management: Salinity Management in the Delaware River Basin

Abstract

Managing salinity in the Upper Delaware Estuary is an important operational goal within the Delaware River Basin (DRB). High salinity concentrations can create water quality and operational challenges which increase treatment costs for downstream water utilities and cause ecological damage. This study reviews the advantages and limitations of process based empirical models (PBEM) as an alternative to complex hydrodynamic models or statistical models (i.e., multivariate regression) for salinity management. PBEMs involve choosing a parsimonious form of equation(s) that logically reproduces important physical relationships. A PBEM was developed to model specific conductivity (SC) (proxy for salinity) at three locations within the DRB more than 50 years. The resulting models explain most of the variations in historic SC and give comparable performance to a much more complex hydrodynamic model. The PBEM was then combined with streamflow, tidal forecasts, and an error model to develop an operational tool for assessing salinity impacts of potential reservoir releases and for generating ensemble forecasts of chlorinity. The authors also document how such ensemble forecasts can be employed to generate probabilistic forecasts of future salinity levels under various water resource system operating assumptions.