Consider Covariance Analysis

This chapter considers the effect on the state vector of errors in constant measurement or model parameters. The design of orbit determination algorithms usually begins with a definition of important error sources and a statistical description of these error sources. The effect of erroneous assumptions regarding the mathematical description of the force model or the measurement model, the statistical properties of the random errors, and the accuracy of the numerical values assigned to the unestimated measurement and force model parameters, as well as the round-off and truncation characteristics that occur in the computation process, can lead to reduced estimation accuracy and, on occasion, to filter divergence. Consider covariance analysis is a “design tool” that can be used for sensitivity analysis to determine the optimal parameter array for a given estimation problem or to structure an estimation algorithm to achieve a more robust performance in the presence of erroneous force and/or measurement model parameters. Covariance analysis is an outgrowth of the study of the effects of errors on an estimate of the state of a dynamical system.