Predicting Comet Trajectories: Integrating Data Science with Physics

Qazalbash, Agha Sibt E Ali

Peacock Scholarship

Predicting Comet Trajectories: Integrating Data Science with Physics

Public Deposited

Download PDF

Abstract

This thesis develops and evaluates a hybrid framework for comet trajectory prediction that explicitly integrates data science with classical celestial mechanics. This project is motivated by a practical and methodological gap: purely data-driven forecasts can fit observational trends but may ignore physical structure, whereas purely physics-based treatments often assume complete orbital information and specialized computational infrastructure. To address this gap, an interactive Python application was designed around a real comet observation dataset from the COBS database. The system ingests observational variables including observation date, Right Ascension, Declination, heliocentric distance, geocentric distance, phase angle, magnitude, and motion metrics. After processing, the framework models the temporal evolution of the comet's apparent position using polynomial trend estimation and blends the resulting forecast with a Kepler-informed motion term derived from heliocentric distance. The application supports variable filtering, tabular prediction output, graph-by-graph exploration, and final integrated Kepler trajectory projection. The thesis argues that the most useful student-scale and research-oriented predictor is not a purely statistical model or a purely analytical orbital solver, but a hybrid system in which data science provides empirical adaptability and physics provides interpretability, constraint, and scientific coherence. Results from the implemented system show consistent positional prediction and coherent reconstructed trajectory, demonstrating the value of interdisciplinary modeling in computational astronomy.

Last modified