Joint identification of groundwater contaminant sources: an improved optimization algorithm

Abstract

Rapid identification of contaminant source information is critical for solving sudden groundwater contamination events. This paper constructs a combined EnKF-SPSO algorithm based on the ensemble Kalman filter (EnKF) and survival particle swarm optimization (SPSO) algorithms to groundwater contamination source identification, which includes determining the location of the source, initial concentration, and emission time. The proposed hybrid architecture improves upon conventional single-algorithm approaches by decoupling the identification process into two stages. First, the EnKF searches for the contaminant source’s location, thereby reducing the search space. Next, the SPSO estimates the initial concentration and emission time within the reduced domain. This two-stage process effectively mitigates the curse of dimensionality often encountered in standalone optimization methods. We set up two solute transport scenarios with different numbers of contaminant sources to examine the effectiveness of the algorithm and compare it with the EnKF, particle swarm optimization (PSO), and SPSO algorithms. The results show that the EnKF-SPSO algorithm can identify the contaminant characteristics more accurately without falling into a local optimum, and the average relative error is less than 1%. In addition, the EnKF-SPSO algorithm, for cases with measurement errors, is highly reliable. The combined algorithm can provide technical support for groundwater contamination remediations, risk assessments, and liability determinations.