Nepal Earthquake Damage Assessment Using GeoAI

This repository contains a comprehensive workflow for post-disaster damage assessment using satellite imagery and GeoAI techniques, specifically developed for the November 3, 2023 earthquake in Nepal.

Event Information

• Date: November 3, 2023
• Magnitude: 6.4 ML (5.7 Mw)
• Epicenter: Ramidanda, Jajarkot District (28.84°N, 82.19°E)
• Affected Districts: Jajarkot, Rukum West, Salyan

Project Overview

This project implements state-of-the-art satellite imagery analysis and GeoAI techniques for earthquake damage assessment, including: - Multi-temporal satellite imagery analysis (Sentinel-2, Landsat 8/9, Sentinel-1) - Spectral change detection using various indices (NDVI, NBR, NDBI) - Machine learning classification for damage assessment - Building-level damage analysis - Landslide detection - Comprehensive reporting and visualization

Requirements

Prerequisites

1. Python 3.9+
2. Google Earth Engine account and authentication
3. API keys for satellite data providers (optional)
4. Sufficient storage for satellite imagery

Installation

```bash

Clone the repository

git clone https://github.com/S-bachir/NepalDammageAssessment_POC.git cd nepal-earthquake-assessment ```

Create conda environment

bash conda create -n earthquake-assessment python=3.9 conda activate earthquake-assessment

Install dependencies

bash pip install -r requirements.txt

Workflow

The complete workflow is documented in the main notebook: notebooks/damage_assessment_workflow.ipynb

Pipeline Steps

1. Data Acquisition (scripts/data_acquisition.py)

• Fetches satellite imagery from multiple sources:
  • Sentinel-2 Level-2A (ESA Copernicus)
  • Sentinel-1 SAR GRD (ESA Copernicus)
  • Landsat 8/9 Collection 2 (USGS)
• Downloads ancillary data:
  • OpenStreetMap building footprints
  • SRTM Digital Elevation Model (NASA)
  • Population density (CIESIN GPW v4.11)

2. Preprocessing (scripts/preprocessing.py)

• Image co-registration
• Radiometric correction
• Cloud masking
• Creation of analysis-ready data

3. Damage Analysis (scripts/damage_analysis.py)

• Spectral change detection
• Machine learning classification
• Building-level damage assessment
• Landslide detection

4. Visualization (scripts/visualization.py)

• Before/after satellite image comparisons
• Damage classification maps
• Interactive visualizations (e.g., via Folium or Plotly)
• 3D terrain analysis

5. Reporting (scripts/reporting.py)

• PDF reports (e.g., via ReportLab or LaTeX)
• Excel summaries (e.g., via Pandas)
• GIS outputs (e.g., Shapefiles, GeoJSON)
• Web dashboards (e.g., via Dash or Streamlit)

Data Sources

• Sentinel-2 Level-2A (ESA Copernicus)
• Sentinel-1 SAR GRD (ESA Copernicus)
• Landsat 8/9 Collection 2 (USGS)
• OpenStreetMap building footprints
• SRTM Digital Elevation Model (NASA)
• Population density (CIESIN GPW v4.11)

Results

The workflow produces:

• Damage classification maps
• Building-level damage statistics
• Landslide susceptibility maps
• Comprehensive reports for disaster response

License

This project is licensed under the Mozilla Public License 2.0. See the LICENSE file for details.

Acknowledgments

• ESA Copernicus Programme for Sentinel data
• USGS for Landsat imagery
• OpenStreetMap contributors
• NASA for SRTM elevation data

Contributing

Contributions are welcome! Please submit pull requests or open issues for improvements and bug fixes.