Elizabeth George, University of Birmingham; Sean Elvidge, University of Birmingham; Matthew Brown, University of Birmingham
Keywords: thermosphere, neutral density, orbit propagation, Kalman filter
Abstract:
The last decade has seen a dramatic increase in the number of satellites in low Earth orbit (LEO). On top of the number of satellites there is also an increasing amount of orbital debris to contend with. In January 2019 there were estimated to be 34 000 debris objects larger than 10 cm in size, 900 000 objects between 1 cm and 10 cm and 128 million objects from 1 mm to 10 cm. This expansion is producing challenges for satellite operators due to the increase in collision warnings. The best way to reduce the number of possible collision warnings is to reduce the size of the error in a satellite’s position. The more accurately the position can be predicted the smaller the error will be. For satellites in LEO the largest uncertainties in position arise from difficulties predicting atmospheric density. Neutral densities within the thermosphere can be challenging to model and predict as they show complicated temporal and spatial variations. The temporal variations can be abrupt over the time scales of minutes to hours, there are also diurnal variations, variations in line with solar rotation and the solar cycle as well as longer term trends. Spatial variations include variations in latitude, longitude and altitude.
Orbit propagators are used to predict the tracks of satellites. Errors between the propagated and recorded position mainly arise from difficulties measuring and predicting density within the thermosphere. As these differences arise from errors in density predictions the approach can be reversed and density estimated from errors in orbit propagation. This study aims to use an ensemble Kalman filter to combine an orbit propagator with precise orbit information and an atmospheric model to produce improved estimates of density within the thermosphere leading to better predictions of satellite position.
Kalman filters present a data assimilation technique which is used widely in the fields of meteorology and oceanology. A Kalman filter is an algorithm used to obtain the optimal combination of observations with a model by minimising variances. It is ideal for use in systems which are continuously changing. Kalman filters are recursive; an estimate of the current state is updated to provide a predictive step. This predictive step then becomes the current state and the process is repeated. Here an estimate of the state is created using an atmospheric model. An ensemble Kalman filter is a Monte Carlo implementation of the Kalman filter. Here the input parameters of solar irradiance and magnetic activity will be perturbed to produce a range of ensemble members. For each member the density will be estimated and used to calculate the atmospheric drag felt by the satellite. This can then be used as input for the propagator. Propagated positions for each member can then be compared to recorded position to make an estimate of the density the satellite has passed through. By repeating this process for a range of satellites in different orbits a global view can be built up. This picture of the thermosphere can then be used as input to a data assimilation model to improve the predictions of density made into the future.
Date of Conference: September 19-22, 2023
Track: Atmospherics/Space Weather