A glimpse at the GOCE satellite gravity gradient observations

Since 30 September 2009, following the launch and in-orbit testing of the most sophisticated gravity mission ever built, the European Space Agency (ESA) GOCE satellite is in 'measurement mode', providing continuous time series of satellite gravity gradient (SGG) observations and GPS satellite-to-satellite tracking (SST) observations. The availability of GPS SST observations allows the precise reconstruction of the GOCE position and thus the precise geolocation of the SGG observations. The SGG observations are based on the differences between observations taken by pairs of accelerometers, which need to be corrected first by applying a so-called calibration matrix and second by subtracting rotational terms (centrifugal and angular accelerations). The GOCE mission is designed to provide SGG observations that are most precise in a bandwidth ranging from 0.005 to 0.1 Hz, equivalent to spatial scales along the satellite flight path of about 80-1600 km. Therefore, the SGG observations need to be bandpass filtered. Two months of GOCE SGG observations covering November and December 2009 (first release by ESA) have been processed and analyzed in the spatial domain by comparison with gradients predicted with state-of-the-art gravity field models (EIGEN-5C and ITG-Grace2010s). The ITG-Grace2010s model displays a high level of consistency with the four SGG components that are designed to be observed precisely, except for the cross-track diagonal component at the low frequency end of the measurement bandwidth (close to 0.005 Hz) for which the differences are about a factor 3 higher. For the EIGEN-5C model, a similar consistency is observed, except for geographical areas where this prior model is considered to be less accurate, such as the Himalayan and Andes mountain ranges, the Indonesian Archipelago and Antarctica. The root-mean-square (RMS) of differences between 1 Hz GOCE SGG observations and those predicted by the prior models is below the signal size for frequencies up to at least 0.04 Hz, indicating the high quality of these observations, and also the models, over a large part of the measurement bandwidth. (C) 2010 COSPAR. Published by Elsevier Ltd. All rights reserved.