Network-adjustment drift correction includes polynomial coefficients of a time-dependent function for drift in the network adjustment solution. With this method, a linear model (1st, 2nd, or 3rd order) is fit to the data:
where D(t) is the cumulative drift at time t, (t-t0) is elapsed time, dpare the model coefficients, and a is the degree of the polynomial. The coefficients dpare parameters in the network adjustment, determined as part of the least-square solution.
Linear drift models can be used with either the network-adjustment drift correction option, or the continuous model option. When a linear model is suitable, the network-adjustment drift correction may be preferable, because uncertainty in the drift-model parameters is propagated into the uncertainty estimate of the final station values. One advantage of the continuous model, however, is the possibility to extend the drift model as a constant value, as opposed to extrapolating the linear model, at the beginning and end of the loop (see "Behavior at start/end" under Plotting options). The suitability of a linear drift model can be evaluated using the continuous model option, after which the drift method can be changed to the network-adjustment option before adjusting the network.
