Novel considerations about the error budget of the LAGEOS-based tests of frame-dragging with GRACE geopotential models

A realistic assessment of the uncertainties in the even zonals of a given geopotential model must be made by directly comparing its coefficients with those of a wholly independent solution of superior formal accuracy. Otherwise, a favorable selective bias is introduced in the evaluation of the total error budget of the LAGEOS-based Lense-Thirring tests yielding likely too optimistic figures for it. By applying a novel approach which recently appeared in the literature, the second (ℓ=4) and the third (ℓ=6) even zonals turn out to be uncertain at a 2-3×10-11(ℓ=4) and 3-4×10 -11(ℓ=6) level, respectively, yielding a total gravitational error of about 27-28%, with an upper bound of 37-39%. The results by Ries et al. themselves yield an upper bound for it of about 33%. The low-degree even zonals are not exclusively determined from the GRACE Satellite-to-Satellite Tracking (SST) range since they affect it with long-period, secular-like signatures over orbital arcs longer than one orbital period: GRACE SST is not accurately sensitive to such signals. Conversely, general relativity affects it with short-period effects as well. Thus, the issue of the a priori "imprinting" of general relativity itself in the GRACE-based models used so far remains open. © 2013 Elsevier Ltd. All rights reserved.