Reconstruction of GRACE Mass Change Time Series Using a Bayesian Framework. Issue 7 (7th July 2022)

• Record Type: Journal Article
• Authors: Rateb, Ashraf; Sun, Alexander; Scanlon, Bridget R.; Save, Himanshu; Hasan, Emad
• Is Part Of: Earth and space science. Volume 9:Issue 7(2022)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2022-07-07
• Subjects: Geodesy -- GRACE (‐FO) -- Bayesian inference -- MCMC -- Mass change -- Hydrology; Space sciences -- Periodicals; Geophysics -- Periodicals; 500.5
• DOI: 10.1029/2021EA002162
• Languages: English
• ISSNs: 2333-5084

Abstract

Abstract: Gravity Recovery and Climate Experiment and its Follow On (GRACE (‐FO)) missions have resulted in a paradigm shift in understanding the temporal changes in the Earth's gravity field and its drivers. To provide continuous observations to the user community, missing monthly solutions within and between GRACE (‐FO) missions (33 solutions) need to be imputed. Here, we modeled GRACE (‐FO) data (196 solutions) between 04/2002–04/2021 to infer missing solutions and derive uncertainties in the existing and missing observations using Bayesian inference. First, we parametrized the GRACE (‐FO) time series using an additive generative model comprising long‐term variability (secular trend + interannual to decadal variations), annual, and semi‐annual cycles. Informative priors for each component were used and Markov Chain Monte Carlo (MCMC) was applied to generate 2, 000 samples for each component to quantify the posterior distributions. Second, we reconstructed the new data (229 solutions) by joining medians of posterior distributions of all components and adding back the residuals to secure the variability of the original data. Results show that the reconstructed solutions explain 99% of the variability of the original data at the basin scale and 78% at the one‐degree grid scale. The results outperform other reconstructed data in terms of accuracy relative to land surface modeling. Our data‐driven approach relies only on GRACE (‐FO) observations and provides a total uncertainty over GRACE (‐FO) data from the data‐generation process perspective. Moreover, the predictive posterior distribution can be potentially used for "nowcasting" in GRACE (‐FO) near‐real‐time applications (e.g., data assimilations), which minimize the current mission data latency (40–60 days). Plain Language Summary: A complete‐time series of Gravity Recovery and Climate Experiment and its Follow On (GRACE (‐FO)) satellite data without gaps is critical for learning about the impact of climate change and human interventions on the Earth's systems. However, maintaining operational satellites is challenging. The GRACE (‐FO) record has 33 missing monthly solutions. To develop estimates of these missing solutions, we evaluated the processes that generate these data using probability and the information we knew about these processes, such as distributions, ranges of intercepts, slopes, frequencies, and variability. We then used Markov Chain Monte Carlo to generate 2, 000 samples for each of these processes. These samples were then used to infer the complete time series using their medians. We added a level of uncertainty around these medians using these samples. We compared these results with land surface modeling and other reconstructed data. Our method compared favorably with the model outputs and the original data. Then, we estimated a similar probability of the near future as surrogates for unseen or undelivered GRACE (‐FO) data. We recommend to researchers interested in GRACE's near real‐time data application (e.g., data assimilation) to incorporate these surrogates to overcome the GRACE data latency. Key Points: Uncertainties in existing and missing solutions of Gravity Recovery and Climate Experiment and its Follow On (GRACE (‐FO)) missions need to be informed Bayesian inference is used to decompose and model temporal GRACE (‐FO) signals The new data explain the variability in original observations and can be used for nowcasting