Thermospheric Density Forecasting using AI/ML
One of the primary challenges in sustainable Low Earth Orbit (LEO) satellite operations is the accurate forecasting of thermospheric neutral densities, which exhibit significant variability with the solar cycle and during geomagnetic storms. Extreme space weather events, such as the Gannon storm of May 2024, can severely destabilize satellite orbits and potentially cause premature re-entry. Traditional forecasting approaches face inherent trade-offs: empirical models offer computational efficiency but limited accuracy, while physics-based models provide higher accuracy at the cost of computational intensity. This whitepaper describes Digantara’s proprietary Space Weather (SWx) solution, a novel AI-driven methodology that leverages machine learning techniques to identify complex relationships between space weather drivers and thermospheric densities. Digantara’s Thermospheric Density Forecasting Model utilizes a comprehensive training dataset spanning 24 years of density measurements derived from actual satellite observations, supplemented with synthetic data to enhance model robustness and coverage. The report presents the results of a testing framework performed at the company to assess the accuracy of the Thermospheric Density Forecasting Model.