Aftershock Oracle

Overview

Earthquakes may seem to be just tremors that make buildings rattle and trees shake, but they can be very dangerous. Even though they were only 8% of the disasters from 2000 to 2019, they accounted for 58% of total fatalities! My project relies on AI models to analyze earthquake data and understand aftershock behavior. I used historic earthquake data from USGS and tectonic plates information, including mainshock magnitude, location, depth, and distance to nearby faults, etc. Next, my model learned from these features and observed patterns. The original goal was to forecast the probability that at least one aftershock (Mag 2.5+) would occur within 30 days and 50 miles of a mainshock (Mag 6+). I then extended it to estimate the maximum aftershock magnitude, when aftershocks would strike, and where aftershocks were most likely to cluster. I wrote my code using PyCharm, a Python IDE. Next, I built and tested three machine learning models, and compared them based on metrics such as precision, recall, and AUC (accuracy). For my purpose, XGBoost performed best, reaching an accuracy of 0.82 on the test data. With this model, I was able to get an AUC of up to 85% when testing magnitude and probability. For location forecasting, I divided the region into a 50×50 grid and generated a probability heatmap. My model passed my criteria of 80% AUC, and it was proven to be a useful aftershock forecasting system. For time forecasting, I created time bins that predicted the number of aftershocks that would occur in a half-day period. My results were comparable to Omori’s Law, a widely accepted time-based frequency formula. My project showed that AI models can learn patterns from past earthquakes and provide insights into aftershock behavior, giving valuable forecasts on location, time, magnitude, and probability.

Competition history

• CSEF 2026 Earth & Environmental Sciences (Track 2) (Junior Division) · Entry J-09-09

Related projects

Closest projects by meaning, across every fair and year in the corpus.

Browse more like this

• All Earth and Environmental Sciences projects
• CSEF Earth and Environmental Sciences projects
• Every CSEF 2026 project