An investigation into time-variant subsidence potentials using inclusive multiple modelling strategies

Area of Science:

• Environmental science and hydrology
• Geosciences and remote sensing
• Water resource management

Background:

• Groundwater over-abstraction is a primary driver of land subsidence in aquifer regions.
• Effective management plans are often absent, exacerbating subsidence risks.
• The Salmas plain in Iran faces significant environmental challenges due to groundwater depletion.

Purpose of the Study:

• To investigate land subsidence in the Salmas plain using the novel ALPRIFT framework.
• To develop Time-variant Subsidence Vulnerability Index (TSVI) mapping capabilities.
• To identify areas vulnerable to subsidence hotspots caused by groundwater over-abstraction.

Main Methods:

• Utilized the ALPRIFT framework for mapping Subsidence Vulnerability Indices (SVI).
• Implemented a three-module strategy for Time-variant SVI (TSVI) mapping.
• Employed Inclusive Multiple Modelling (IMM) with Fuzzy Logic (FL) and Support Vector Machine (SVM) for predictive modeling.

Main Results:

• Developed and validated SVI and Groundwater Level (GWL) predictive models using IMM.
• Achieved defensible model accuracy through IMM strategies focusing on accuracy enhancement and residual analysis.
• Identified northern Salmas plain as a subsidence risk hotspot due to unplanned groundwater over-abstraction.

Conclusions:

• The developed TSVI mapping capability effectively identifies areas vulnerable to land subsidence.
• Inclusive Multiple Modelling (IMM) enhances the accuracy and reliability of subsidence prediction models.
• Urgent management interventions are needed for the Salmas plain to mitigate subsidence risks.