������Ƶ

Topic

Analysis and Comparison of Spatial Interpolation Methods in Modelling the Movement and Remediation of Hydrocarbons in Soil

Department of Computer Science

Date & location

• Thursday, October 15, 2026

• 1:00 P.M.

• Virtual Defence

Reviewers

Supervisory Committee

• Dr. Kevin Stanley, Department of Computer Science, ������Ƶ (Supervisor)

• Dr. Teseo Schneider, Department of Computer Science, ������Ƶ (Member) 

External Examiner

• Dr. Laura Minet, Department of Civil Engineering, ������Ƶ 

Chair of Oral Examination

• Dr. Sara Ellison, Department of Physics and Astronomy, ������Ƶ

   

Abstract

Canada faces a growing number of abandoned oil wells and petrochemical contamination sites. The presence of petroleum hydrocarbons (PHC) in soil can pose serious risks to the environment, agricultural productivity, and human health in nearby communities if left unchecked. Remediating these sites is costly and time-intensive. In situ bioremediation technologies artificially accelerate biodegradation within the soil without requiring excavation of the affected material. Effective in situ biostimulation remediation is aided by accurately mapping PHC plume distributions from sparse sensor measurements, as interpolating these measurements into a continuous plume estimate can significantly improve remediation efficiency and outcomes. This thesis evaluates two traditional distance-based spatial interpolation methods, Inverse Distance Weighting (IDW) and Kriging, alongside three parameter estimation techniques, 4D-Variational Data Assimilation (4D-Var), the Gauss-Levenberg Marquardt algorithm (GLM), and the Iterative Ensemble Smoother (IES). The spatial interpolation methods are benchmarked on a physics-based synthetic dataset and validated on real-world sensor data to identify the most effective spatial interpolation method for interpreting petrochemical remediation sites in Canada. 

Across nearly all conditions tested, including measurement noise, sensor dropout, and reduced sensor counts, the physics-based parameter estimation methods (4D-Var, GLM, IES) substantially outperform the distance-based methods. Among these, IES achieves the best overall accuracy and provides an ensemble-based uncertainty estimate, though at an increased computational cost. For real-time or edge computing scenarios IDW offers an advantage over Kriging. Taken together, these results indicate that where computational cost is acceptable, IES is the recommended method for PHC plume reconstruction at active remediation sites.