distribution of contamination, such as geo and climatic information, contaminants’ level and form, etc. can be approximated using spatial interpolation methods, in which geostatistical modelling (GM) is most popular and proved effective one.
So far, number of GM methods have been proposed, developed and widely used in environmental science, technology and engineering. They include the ordinary kriging, universal kriging, empirical Bayesian kriging (EBK), and regression kriging (RK). Fundamentally these geostatistical approaches predict the spatial profile/distribution or spatial map of contamination solely in terms of the variation against distance in reference to the sampling points on the base of the average of the values at all sampling points. For the reason, the accuracy of GM highly depends on the number of the available sample data, and their representative nature. In addition, geostatistical modelling addresses each contaminant individually, giving little consideration for the interplay and potential inherited correlation between multi contaminants.
On the fast advance in artificial intelligence, machine learning (ML) has become more and more versatile and been productively used in data analysis. An important feature of ML is the ability to learn from and deal with a complex, multidimensional and less directly correlated dataset. In contrast to spatial interpolation adopted by GM, ML can handle a high number of cross-correlated parameters so may depict complex nonlinear interactions between contaminants and their properties in related physics inside and beyond (extrapolate) the sampling region.
Physics-based modelling (PBM) use fundamental physical laws to describe the working mechanisms of environmental system. Multi-disciplinary physics underpin the description, which includes, surface hydrology, subsurface hydrology, soil physics, multi-phase flow in porous media, thermodynamics, reactive mass transport in soils, etc. Classical theories, descriptive mathematical models, and computational analysis technologies, such as finite element method, have provided reliable methodology and tools to quantify the contamination process and results. So, PBM can be more explanative and accurate used for contaminating assessment and prediction. However, the effectiveness of PBM still depends on the reliability of the input parameters of the mathematical models defining the underlying physics.
Recently, the hybrid technology using the GM and ML in the environmental science, contamination, remediation/restoration and resource management has had raising interest and great advance. However, the attempts to integrate the GM, PBM and ML to together and develop relevant technology are little reported. Integrating the PBM into the hybrid GM and ML approach can create further improved robust, accurate, and interpretable models for complex systems. This interdisciplinary approach is to be explored in the proposed international partnership for the application in the hydrocarbon soil contamination caused by crude oil release, and the bioremediation design for both situation and result assessment and prediction. The collaboration is going to be carried on a well-established preliminary research works on the PBM for subsurface hydrology, reactive mass transport modelling for porous material contamination, GM for hydrocarbon contamination of the oil lakes in Kuwait, bio- and wetland-remediation technologies in both partner universities. These previous work are completed and ongoing PhD research projects, a completed EU horizon project, and number of completed industrial consultation project. The expected research outputs will bring in wider international collaboration in wider prospective environmental contamination and risk assessment domains, such as, historic mining tailings and the soil erosion under climate change.
Funding Notes
To inquire about University of Salford funding schemes – including the Widening Participation Scholarship – visit this website:
https://www.salford.ac.uk/doctoral-school/phd-studentships#:~:text=PhD%20Widening%20Participation%20Scholarships&text=For%20entry%20in%20September%202025,may%20change%20in%20future%20years
References
https://salford.worktribe.com/record.jx?recordid=1159986
