DanielOdushe/Saline.Freshwater-Zonation_Resistivity-Boreholes-Machine-Learning-Approach
Domain:
environment and energyRecord type:
project
Creator:
Dan
Host:
Integrated resistivity surveys and borehole logs fed into supervised ML to classify and map saline versus freshwater zones in Igbo Efon, Southwestern Nigeria.
# Integrated Geophysical Investigation with Machine Learning for Seawater Intrusion Delineation in Igbo Efon, Southwestern Nigeria
## Project Overview
This comprehensive hydrogeophysical study integrates **Electrical Resistivity Tomography (ERT)**, **Vertical Electrical Sounding (VES)**, and **Machine Learning (ML)** techniques to delineate seawater intrusion in coastal aquifers of Igbo Efon, Southwestern Nigeria. The project successfully maps subsurface resistivity variations, identifies saline-freshwater interfaces, and develops predictive models with **92.4% accuracy**.
## Research Objectives
1. **Conduct geophysical surveys** (VES and ERT) to delineate subsurface resistivity variations and saltwater interface
2. **Assess vertical and lateral aquifer continuity** using 2D resistivity imaging
3. **Construct integrated earth models** revealing subsurface lithology
4. **Validate subsurface models** with borehole data for accurate lithological representation
5. **Identify saline water invasion zones** and their occurrence depths
6. **Implement machine learning** for validation and predictive modeling of saline-freshwater boundaries
## Project Structure
### Data Collection Components
- **Vertical Electrical Sounding (VES)**: 20 stations across 5 traverses
- **Constant Separation Traversing (CST)**: 5 ERT profiles (TR1-TR5) covering 160-200m each
- **Borehole Data**: 2 wells (BH_1 and BH_2) with comprehensive geophysical logging
- **Field Instrumentation**: OHMEGA Resistivity Meter with full accessory suite
### Data Processing Workflow
- **VES Processing**: 1D inversion and modeling using PyGIMLi
- **ERT Processing**: 2D inversion and modeling using PyGIMLi
- **Data Integration**: Spatial correlation of VES, ERT, and borehole data
- **Machine Learning**: Random Forest classification and regression models
### Key Deliverables
1. **2D Resistivity Models**: ERT inversion results for all 5 traverses
2. **Integrated Lithological Sections**: Geo-sections combining …
Similar
DanielOdushe/NGA-Weather-AnalysisEthnicity Classification: A Machine Learning ApproachPractical Hydrocarbon Allocation – A Machine Learning ApproachGene expression prediction: A machine learning approachmaryamajibade/Predictive-Machine-Learning-Approach-in-AgricultureA Machine learning approach for Shape From Shading