Gravity field recovery using co-estimation of background model errors to improve de-aliasing capabilities of the MAGIC double-pair constellation

GFZ has performed various full-scale simulations within the ESA NGGM/MAGIC Science Support Study, including instrument noise and background model error assumptions. The focus was set on developing and applying extended parameterization techniques for improved de-aliasing of short-term mass variations. The impact of using model uncertainties was investigated for ocean tide and non-tidal atmospheric and oceanic background models. As part of the DFG Research Unit NEROGRAV covariances of model uncertainties were computed and during gravity field retrieval model corrections were co-estimated using this prior covariance information. In principle, model errors are absorbed by the additional co-estimated parameters, and gravity field estimation is thereby improved. First, simulations with only ocean tide errors and only non-tidal background model errors were performed separately for one month to assess the error reduction obtained for each. Finally, ocean tide and non-tidal errors were included together in a full noise simulation and compared to the processing strategy that did not include co-estimation of background model errors. The novel optimized method was then also applied for monthly gravity field retrieval over one year, showing improvements for each month. The estimated residual ocean tides from this 1-year simulation were then used to calculate an improved ocean tide model optimized for gravity field recovery. In addition, monthly solutions, obtained with co-estimation of background model errors, have been calculated for the inclined pair alone using regularization for the not-covered polar regions and compared to double pair solutions for latitudes lower than +/- 70 degrees.