Evaluating non-tidal atmospheric products by measuring GRACE K-band range rate residuals

In order to reduce high-frequency non-tidal mass changes, while inverting for the Earth’s time-variable gravity fields from the Gravity Recovery And Climate Experiment (GRACE) measurements, it is usual to apply the Atmospheric and Oceanic De-aliasing (AOD1B) products. However, limitations in these products count as a potential threat to the accuracy of time-variable gravity fields derived from GRACE, as well as its follow-on mission(s). Therefore, in this study, we show to what extent the GRACE-type gravity recovery procedure is sensitive to different non-tidal atmospheric background models. For this, we evaluate the atmospheric parts of the GeoForschungsZentrum’s AOD1B RL05 and RL06, as well as those computed as a part of the European Space Agency Earth System Model ESA-ESM and the ITG3D model. These data products employ different atmosphere fields (operational and reanalysis data or their combination) as inputs, and they are also computed by implementing different 2-D or 3-D integration methods. The accuracy of these products is assessed by comparing the resulting GRACE K-Band Range Rate (KBRR) residuals computed for time-variable gravity field inversions, while using each of them separately as a background model. Our investigations during 2006 indicate that: (i) Applying ESA-ESM and ITG3D decreases averaged KBRR residuals by 2.8 and 3.4 nm s−1 compared to those reduced by the official RL05 products. (ii) Projecting these residuals onto the spatial domain indicates that the improvement covers 78.4 and 78.9  per cent of the globe, respectively. (iii) We find that, compared to ESA-ESM, ITG3D can further reduce the KBRR residuals by 1.8 nm s−1 at regions of high latitudes, which likely improve the uncertainty of ice mass estimations. Our investigation of the AOD1B RL06 products covers 2006–2010, which indicates the advantage of using the higher temporal sampling, that is, 3-hourly reanalysis data. Applying the RL06 reduces the averaged KBRR residuals by 44.2 nm s−1 with respect to the use of the RL05 for gravity field inversion. We, therefore, conclude that the integration method of ITG3D and utilizing reanalysis data with higher (than 6-hourly) temporal sampling rate are beneficial for GRACE-like gravity inversion such as the GRACE Follow-On mission with laser interferometric ranging system.