Investigation of the effects of correlated measurement errors in time series analysis techniques applied to nuclear material accountancy data. [Program COVAR]

It has been shown in previous work that the Kalman Filter and Linear Smoother produces optimal estimates of inventory and loss from a material balance area. The assumptions of the Kalman Filter/Linear Smoother approach assume no correlation between inventory measurement error nor does it allow for serial correlation in these measurement errors. The purpose of this report is to extend the previous results by relaxing these assumptions to allow for correlation of measurement errors. The results show how to account for correlated measurement errors in the linear system model of the Kalman Filter/Linear Smoother. An algorithm is also included for calculating the required error covariance matrices.