Modelling time-varying gravity fields from Level-1B GRACE data using mascons

Abstract

As an alternative to spherical harmonics, mass concentration (mascon) parameters have been successfully applied to the recovery of time-varying gravity (TVG) fields from the GRACE satellite mission. However, before meaningful mass anomalies can be estimated, the noise and errors inherent in the solutions needs to be quantified and appropriate procedures adopted for mitigation. The uniqueness of the mascon methodology is the capability to mitigate noise and errors using spatial and temporal constraints, which can be adapted and tailored to any geophysical signal of interest. Therefore, in the first instance, this work was motivated by the need to improve the accuracy of GRACE TVG fields by understanding the effect of noise and errors. This study then aims to validate mascons for recovery of basin scale inter-annual mass variability at a 10 day temporal resolution. Newcastle University’s orbit determination software, Faust, was modified to allow for estimation of mascon parameters including: modelling of accelerometer bias values; mascon parameterisation; and processing based on short-arc gravity field recovery and KBRR data. Accuracy assessments were undertaken using simulations in the presence of realistic noise facilitating the comparison of mascons and spherical harmonic coefficients, including an assessment of potential limitations associated with each technique. Comparisons with time-series derived from CSR RL05 Level-2 data validated the mascon TVG field recovery, before estimation of the mass change of Antarctica, Greenland and Alaska. Several hydrological basins, including the Amazon and Indus were also assessed before the GRACE trends resulting from the Sumatra earthquake of 2004 were investigated. While only provided for validation, these comparisons provide confidence in the mascon mass estimates. Between January 2003 and December 2013 a mass change of -83 ± 12 Gt/year and -242 ± 7 Gt/year were estimated for Antarctica and Greenland respectively by linear regression using mascons with a 10 day temporal resolution. Overall, the work undertaken in this thesis provides evidence of the improved accuracy achievable when using mascon parameters to estimating TVG fields from Level-1B GRACE data. ii As part of this work a processing methodology to estimate mascon parameters from Level-1B GRACE data using Newcastle University’s orbit determination software Faust has been established and documented. This leaves the University well placed to continue processing mascon solutions from Level-1B GRACE data and to estimate mascon solutions from the GRACE-FO mission. Through simulations, mascon parameters were found to offer advantages over spherical harmonics for the mitigation of noise and for improving the temporal and spatial recovery of the TVG field from GRACE. The mascon constraint matrix allowed more signal to be preserved up to degree ~35. Using basin constraints, simulation revealed that the constraint matric can be tuned to recovery the gravity changes resulting from any geophysical phenomena of interest. Basin constraints were found to optimise the signal recovery of GLDAS and a known mass change signal over Antarctica and Greenland. A novel way to create realistic noise and errors in the KBRR measurement was also documented. Generating monthly and 10 day mascon solutions using real data revealed that the noise and errors in mascon solutions is comparably lower than in CRS RL05 solution while also validating the mascon methodology established here. Comparison to published mass trends to those estimated using mascon parameters showed that the estimation of mascon parameters has application in the study of mass change in the cryosphere, hydrological applications and for the study of the co-seismic mass changes resulting from earthquakes.