Removal of arbitrary discontinuities in atmospheric density modeling

The use of discontinuous a P values, for the simultaneous sequential estimation of six spacecraft LEO parameters, atmospheric density, ballistic coefficient, and other observable paramters, results in the failure of McReynolds' Filter-Smoother Consistency Test. But this test is satisfied when input values of a P are splined (smoothed) before use. The three-hourly discontinuities in a P are always wrong in the sense that three-hourly discontinuities are never produced by the magnetometers from which a P is derived. Our purpose herein is two-fold: (i) Present graphical results for McReynolds' Filter-Smoother Consistency Tests (ii) Record and present the development of the osculating polynomial splines we have used to smooth the discontinuous a P function. © Analytical Graphics, Inc., 2004.