Nepal Earthquake Trend Analysis and Seismic Risk Prediction Based on Severity Classification

Overview

This project analyzes historical earthquake data of Nepal to understand long-term seismic trends and applies machine learning techniques to classify earthquake severity and estimate seismic risk levels.

Using earthquake records from the USGS catalog (1900–2026) and NEMRC catalog (1994-2025), the project combines exploratory data analysis, geospatial visualization, and supervised machine learning to identify seismic patterns and support data-driven risk assessment.

The project focuses on risk estimation and pattern analysis, not exact earthquake prediction.

---

Objectives

• Analyze long-term earthquake trends in Nepal.
• Study spatial distribution of earthquakes.
• Classify earthquakes based on severity levels.
• Use classification results to estimate seismic risk.
• Visualize earthquake distribution using geospatial tools.

---

Dataset

• Source: United States Geological Survey (USGS) & National Earthquake Monitoring and Research Center (NEMRC)
• Region: Nepal and surrounding areas
• Time Range: 1900 – 2026
• Typical Features:
  • Date and time
  • Latitude
  • Longitude
  • Depth
  • Magnitude
  • Location description

---

Project Components

1. Trend Analysis (EDA)

• Earthquake frequency over time
• Magnitude distribution analysis
• Depth vs magnitude relationships
• Pre and post major earthquake comparisons

2. Geospatial Visualization

• Earthquake distribution maps
• Severity-based mapping
• Regional seismic concentration analysis

3. Machine Learning Model

A supervised learning model is used to classify earthquake severity.

Example severity classes:

• Low Severity
• Moderate Severity
• High Severity

Models evaluated may include:

• Logistic Regression
• Decision Tree
• Random Forest

4. Seismic Risk Estimation

Risk levels are derived from severity and frequency patterns to identify relatively high-risk regions or periods.

---

Tools & Technologies

• Python
• Pandas & NumPy
• Matplotlib
• GeoPandas
• Scikit-learn
• Jupyter Notebook

---

Project Workflow

1. Data collection
2. Data cleaning and preprocessing
3. Exploratory data analysis
4. Feature engineering
5. Model training and evaluation
6. Risk interpretation and visualization

---

Limitations

• Exact earthquake prediction is scientifically impossible.
• Results represent statistical patterns and risk estimation only.

---

Expected Outcome

• Understanding of Nepal’s earthquake trends
• Severity classification model
• Visual seismic risk insights
• ML pipeline demonstration for seismic data

---

Future Improvements

• Incorporate real-time seismic feeds
• Improve spatial risk modeling
• Apply time-series forecasting techniques
• Build an interactive dashboard

---

Author

Roman Shrestha
Madhav Tharu
Electronics, Communication & Information Engineering
Tribhuvan University, IoE Pashchimanchal Campus