Gravity Estimation from a Simulated GRACE Mission: Short vs. Long Arcs

We present simulations of gravity estimation from a GRACE-like satellite mission: low-low intersatellite tracking with a precision of order 1 μ m s^-1 yielding gravity fields of degree and order 120. We employ a unique parameterization of the intersatellite baseline vector which allows the gravity estimation to be performed (relatively) independently of the GPS tracking data once sufficiently accurate orbits are obtained. This considerably simplifies data processing during the gravity estimation. During that process only certain components of the baseline parameterization need be adjusted; other components are uncorrelated with gravity and may be adopted unchaged from the initial GPS orbits. The technique is also amenable to very short arcs of data. We present comparisons of gravity estimation from 30 days of observations with arcs of length 15 minutes vs. arcs of one day. Our `truth' field is the EGM96 model; our prior field is a degree-70 clone of EGM96, perturbed from it by amounts comparable to the standard errors of EGM96 (and identically zero for degrees 71--120). For a high inclination orbit, the short-arc analysis recovers low-order gravity coefficients remarkably well, although higher order terms, especially sectorial terms, are understandably less accurate. The simulations suggest that either long or short arcs of GRACE data are likely to improve parts of the geopotential spectrum by several orders of magnitude. This is especially so for low-order coefficients, which are markedly improved for all degrees through 120.