Influence of the spatial and temporal monitoring design on the identification of an instantaneous pollutant release in a river

https://doi.org/10.1016/j.advwatres.2020.103788Get rights and content
Under a Creative Commons license
open access

Highlights

  • Identification of pollution source parameters after an instantaneous pollutant release.

  • Simulation of pollutant concentrations by an analytical solution of the 2D advection-dispersion-reaction equation.

  • Analysis of structural and practical parameter identifiability by the profile likelihood approach.

  • The spatial and temporal monitoring design influences parameter identifiability and uncertainty.

Abstract

In recent years, several approaches in the field of pollution source identification have been developed to identify the unknown pollution source parameters (PSP) of a river pollution incident. These approaches rely on monitoring data which have to be collected in the case of the pollution incident. Up until now, the influence of the often arbitrarily selected monitoring data on the identification results is rarely evaluated. Nevertheless, when considering an inverse problem, monitoring data can be of major importance regarding parameter identifiability and the reliable estimation of unknown parameters. Accordingly, in this work, the influence of the spatial and temporal monitoring design on the identification of an instantaneous pollutant release is analyzed. Assuming steady flow conditions, an analytical solution of the two-dimensional advection-dispersion-reaction-equation is used to model the pollutant transport in the river. The mirror-image technique is applied to capture no-flux conditions across the lateral boundaries. Parameter identifiability is analyzed using the profile likelihood approach. The approach additionally enables the user to determine causes in the case of poor or non-identifiability and to derive likelihood-based confidence intervals. The results imply that the spatial and temporal monitoring design has a significant influence on the identifiability and reliable estimation of PSP. This work provides an overview of first recommendations to guide the setup of a monitoring campaign in the case of a pollution incident.

Keywords

Surface water
Pollution source identification
Monitoring network design
Identifiability analysis
Profile likelihood
Parameter uncertainty

Cited by (0)