Mean Density Estimation derived from Satellite Constellations

Abstract:

With the advent of nanosatellite constellations, we define here a new method to derive neutral densities of the lower thermosphere from multiple similar platforms travelling through same regions of space. Because of similar orbits, the satellites are expected to encounter similar mean neutral densities and hence experience similar drag if their drag coefficients are equivalent. Utilizing free molecular flow theory to bound the minimum possible drag coefficient possible and order statistics to give a statistical picture of the distribution, we are able to estimate the neutral density alongside its associated error bounds. Data sources for this methodology can either be from already established Two Line Elements (TLEs) or from raw data sources, in which an additional filtering step needs to be performed to estimate relevant parameters. The effects of error in the filtering step of the methodology are also discussed and can be removed if the error distribution is Gaussian in nature. This method does not depend on prior models of the atmosphere, but instead is based upon physics models of simple shapes in free molecular flow. With a constellation of 10 satellites, we can achieve a standard deviation of roughly 4% on the estimated mean neutral density. As additional satellites are included in the estimation scheme, the result converges towards the lower limit of the achievable drag coefficient, and accuracy becomes limited by the quality of the ranging measurements and the probability of the accommodation coefficient. Data is provided courtesy of Planet Labs and comparisons are made to existing atmospheric models such as NRLMSISE-00 and JB2006.