The temporal gravity solutions estimated from NASA/DLR’s Gravity Recovery And Climate Experiment (GRACE) mission, and its successor, NASA/GFZ’s GRACE Follow-On (GRACE-FO), manifested as mass transports within the Earth system, have been used for a wide variety of Earth Science and climate change studies since 2002. However, there is an around one-year gap between the two satellite gravity missions (2017-2018). ESA’s fifth Earth Explorer Mission, the Swarm 3-satellite constellation, equipped with geodetic quality GNSS tracking system, was proposed to fill the gravimetry observation climate record data gap, at a moderate spatial resolution. Here, I applied a modified decorrelated acceleration approach to recover temporal gravity field using the 3-satellite Swarm constellation GPS tracking data. This approach is based on the simple linear relation between the second time derivative of the orbit and the gravitational acceleration. However, the time derivative could highly amplify the noise and make the noise correlated. In addtion, GPS positioning also involves correlation noise. Therefore, two linear transformations were introduced to decorrelate the observation noise. Next, two adjustment methods were studied to optimally combine the three gravity components, namely along-track, cross-track, and radial direction, along with introducing relative weights among orbital arcs for the final optimal gravity field estimation. The Swarm-only temporal gravity solutions have a good to excellent agreement with the overlapping GRACE/GRACE-FO solutions at least up to spherical harmonics degree around 13 (~1500 km, half-wavelength).
Swarm-only temporal gravity solutions were then used to fill the mass change data gap over Greenland and West Antarctica ice-sheets during 2017-2018. Over Greenland, Swarm observed mass anomalies agreed well within the time epochs that overlaped with GRACE (correlation coefficient (CC) = 0.62), and GRACE-FO (CC=0.78). Within the data gap year, Swarm observed mass anomalies were relatively small suggesting that the Greenland mass loss slowed down, where the estimated short-term linear trend dropped from -54.3 ± 1.9 mm/yr (2013-2016 from GRACE) to -13.3 ± 7.5 mm/yr (2016-2018 from Swarm). In addition, as compared with the relatively quiet 2015-2017 at 13.5 ± 14.7 mm/yr, Swarm observed a fast ice mass loss episode at -89.2 ± 9.4 mm/yr during the gap year over West Antarctica, which agreed well with the estimate from GRACE and GRACE-FO without considering the gap at -92.8 ± 2.8 mm/yr during 2017-2019. This fast mass loss episode observed by Swarm also supports that the offset between GRACE and GRACE-FO time series is indeed due to mass loss but not a systematic bias.
The official GRACE/GRACE-FO gravity products are derived from K-/Ka Band range (KBR) rate observations. Alternatively, the range acceleration observations could be used to estimate temporal gravity based on the so-called acceleration approach. In this study, by means of satellite orbit refinement, novel error mitigation schemes, and proper stochastic model estimation, the representation of range accelration was significantly improved in the acceleration equation (admittance spectrum dropped from up to 7 to around 1), and the in-situ line-of-sight gravity difference (LOSGD) was estimated with a high fidelity (CC = 0.96 with Level 2 data predicted LOSGD). For the first time, the improved acceleration approach was implemented for global temporal gravity recovery using GRACE and GRACE-FO observed range accelerations. The temporal gravity solutions recovered using this approach are, in general, in good agreement with the GRACE official Level 2 data products, based on the comparisons of the global mass variation trends, and basin-scale mass anomalies times series. Particularly, the gravity solution correlations between solutions in this study and other solutions are higher during the GRACE-FO time span. Despite the loss of an accelerometer onboard one of the GRACE-FO satellites, this closer comparison could be attributable to the improved range observation quality and the reduced noise level, which is clearly shown in the gravity inversion formal error. Because the high-low GPS tracking data were not used in this study, the low degree sectoral coefficients are believed to be slightly degraded compared to other solutions.
The conventional GRACE/GRACE-FO temporal gravity solutions are at monthly sampling, which cannot easily be used to study sub-monthly mass transport events. However, the satellite ground track coverage varies from time to time. For the denser coverage time, a sub-monthly temporal resolution could be reached. A shorter solution data span, less than half of the nominal monthly data span, would enable observing signals which propagates quicker than a month. I employed the improved acceleration approach developed in this study to estimate solutions for every 13 days with one day sliding windows, which gives a daily sampling rate. The daily mass anomalies estimated from these solutions are shown to have a high correlation with the Morakot Typhoon (2009) induced precipitation evolutions (CC=0.87). It is shown that GRACE data is able to monitor the Morakot Typhoon induced massive rainfall during its landfall over Taiwan, which lasted only several days, though left a vast destruction on human lives and properties.
In addition to the conventional spherical harmonic solutions, the GRACE/GRACE-FO Data Centers also deliver alternative data products called the “mascon solution”. Constraints are applied during the inversion so that it is free from the conventional GRACE post-processing. This advantage makes it a better candidate for coastal sediment deposition studies. Here, I used the University of Texas Center for Space Research (CSR) RL06 mascon data product to quantify the sediment deposition in the Bay of Bengal. By subtracting the Glacial Isostatic Adjustment (GIA) forward model predicted mass anomalies, ocean mass anomalies and the early Holocene Sediment Isostatic Adjustment (SIA) forward model predicted mass anomalies from the total mass change observed by GRACE (2002-2017), I obtained the mass anomalies estimation induced by the sediment discharge and transport in the Bay area. The corresponding sediment deposition rate estimate is 0.5± 0.2 Gt/yr, which is only half of the Brahmaputra river annual sediment discharge. This study also suggested the current SIA model tended to underestimate the SIA induced subsidence approximately by a factor of 2.
In conclusion, the gravity solutions estimated from Swarm GPS tracking data using the modified decorrelation acceleration approach are capable to capture temporal gravity signals up to around degree 13. The Swarm-only solutions are shown to be able to fill the data gap between GRACE and GRACE-FO over West Antarctica and directly observe a fast mass loss episode.
For GRACE and GRACE-FO, the improved acceleration approach has estimated the in-situ LOSGD with a high quality as indicated by the high correlation (CC=0.96) with L2 product predicted values and the monthly gravity solutions estimated from LOSGD have a good to excellent agreement with the official L2 products. The resulting GRACE daily sampled 13-day gravity solutions are capable to observe and quantify the evolution of an example abrupt weather episode, the landfall of the 2009 Morakot Typhoon over Taiwan. The demonstration of this novel monitoring of cyclone, for the first time, allows feasibility of using gravimetry data for possible disaster management.