Development of an observational error model

In calculating the orbit of a minor planet with a least-squares algorithm, current practice is to assume that all observations of a given era have the same uncertainty, and that the errors in these observations are uncorrelated. These assumptions are unrealistic; and they lead to sub-optimal orbits. Our objective is to develop and validate an observational error model that provides realistic estimates of the uncertainties and correlations in asteroid observations. When used to populate the covariance matrix of the least-squares algorithm, the resulting orbits are shown to more accurately and precisely represent asteroid trajectories.