A Transfer Function Between Line‐of‐Sight Gravity Difference and GRACE Intersatellite Ranging Data and an Application to Hydrological Surface Mass Variation

• Published in: Journal of geophysical research. Solid earth 2018-10, Vol.123 (10), p.9186-9201
• Main Authors: Ghobadi‐Far, Khosro, Han, Shin‐Chan, Weller, Steven, Loomis, Bryant D., Luthcke, Scott B., Mayer‐Gürr, Torsten, Behzadpour, Saniya
• Format: Article
• Language: English
• Subjects: Acceleration; admittance; Basins; Correlation; Correlation analysis; Data; Data processing; Electrical impedance; Estimation errors; Geophysics; GRACE; GRACE (experiment); Gravitation; Gravitational fields; Gravity; Gravity field; Hydrologic models; Hydrology; Modelling; Rangefinding; range‐acceleration; Runoff; Satellites; Spectral analysis; Surface water; time‐variable gravity; Transfer functions; Water storage
• ISSN: 2169-9313; 2169-9356
• DOI: 10.1029/2018JB016088

We develop a transfer function to determine in situ line‐of‐sight gravity difference (LGD) directly from Gravity Recovery and Climate Experiment (GRACE) range‐acceleration measurements. We first reduce GRACE data to form residual range‐acceleration referenced to dynamic orbit computed with a reference gravity field and nonconservative force data. Thus, the residuals and the corresponding LGD data reflect time‐variable gravity signals. A transfer function is designed based on correlation‐admittance spectral analysis. The correlation spectrum shows that residual range‐acceleration and LGD are near‐perfectly correlated for frequencies >5 cycles‐per‐revolution. The admittance spectrum quantifies that the LGD response to range‐acceleration is systematically larger at lower frequencies, due to the increased contribution of centrifugal acceleration. We find that the correlation and admittance spectra are stationary (i.e., are independent of time, satellite altitude, and gravity strength) and, therefore, can be determined a priori with high fidelity. We determine the spectral transfer function and the equivalent time domain filter. Using both synthetic and actual GRACE data, we demonstrate that in situ LGD can be estimated via the transfer function with an estimation error of 0.15 nm/s2, whereas the actual GRACE data error is around 1.0 nm/s2. We present an application of LGD data to surface water storage changes in large basins such as Amazon, Congo, Parana, and Mississippi by processing 11 years of GRACE data. Runoff routing models are calibrated directly using LGD data. Our technique demonstrates a new way of using GRACE data by forward modeling of various geophysical models and in‐orbit comparison with such GRACE in situ data. Key Points In situ gravimetric observables can be computed from combination of GPS and KBR data of the GRACE satellite mission We develop a transfer function for accurate estimation of LOS gravity difference directly from level‐1B residual KBR range‐acceleration data The in‐orbit comparison between in situ gravimetric data and models is a simple and powerful approach for hydrological model evaluations