Simultaneous identification of groundwater contamination source and aquifer parameters with a new weighted-average...

Area of Science:

• Environmental Science
• Hydrogeology
• Computational Science

Background:

• Groundwater contamination source and aquifer parameter identification is crucial for environmental management.
• Existing methods face challenges with noisy data and high computational costs.
• Wavelet threshold denoising methods have limitations in improving identification accuracy.

Purpose of the Study:

• To develop a parallel heuristic search strategy for simultaneous identification of groundwater contamination sources and aquifer parameters.
• To propose a new weighted-average wavelet variable-threshold denoising (WWVD) method for improving data quality.
• To create an optimal ensemble surrogate model using a hybrid algorithm to reduce computational cost.

Main Methods:

• A parallel heuristic search strategy was developed for simultaneous identification.
• A weighted-average wavelet variable-threshold denoising (WWVD) method was proposed.
• A differential evolution-tabu search (DE-TS) hybrid algorithm was developed to construct an optimal ensemble surrogate model (OES) using Gaussian process, kernel extreme learning machine, and support vector regression.

Main Results:

• The WWVD method significantly improved the denoising effect of concentration data and enhanced subsequent identification accuracy.
• The DE-TS algorithm improved the approximation accuracy of the OES model to the simulation model.
• The parallel heuristic search strategy proved effective for simultaneous identification.

Conclusions:

• The proposed WWVD method offers superior denoising performance for environmental data.
• The DE-TS algorithm effectively creates accurate ensemble surrogate models, reducing computational burden.
• The integrated approach enhances the accuracy and efficiency of groundwater contamination source and aquifer parameter identification.