DanielOdushe/Saline.Freshwater-Zonation_Resistivity-Boreholes-Machine-Learning-Approach

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 …