ANU RL03 Mascon Solution

The RL03 solution employs data-driven regularisation. We use the strength of the prefit range accelerations to determine which mascons must be adjusted (hence need weaker regularisation) and which have no signal and don't need to adjust (hence strong regularisation applied).

For the initial iterations, we estimate the best-fitting adjustment to each mascon - one at a time - and use the RMS reduction of the range acceleration residuals to scale the regularisation sigma used. Range acceleration residuals within a +/- 8 degree latitude box (and corresponding longitude width) are included when finding the best-fitting adjustment for each mascon.

In subsequent iterations we use the median residual within a +/- 2 degree box around each mascon to scale the regularisation sigma for each mascon.