Extraction of GRACE/GRACE-FO observed mass change patterns across Antarctica via independent component analysis (ICA)

• Published in: Geophysical Journal International
• Main Authors: Shi, Tianyan, Fukuda, Yoichi, Doi, Koichiro, Okuno, Jun’ichi
• Other Authors: NASA, JPL, CSR, GFZ, ICA
• Format: Article in Journal/Newspaper
• Language: English
• Published: Oxford University Press (OUP) 2022
• Subjects: Antarc*; Antarctic; Antarctica; Ice Sheet
• Online Access: http://dx.doi.org/10.1093/gji/ggac033; https://academic.oup.com/gji/advance-article-pdf/doi/10.1093/gji/ggac033/42330943/ggac033.pdf; https://academic.oup.com/gji/article-pdf/229/3/1914/42644192/ggac033.pdf

SUMMARY Here we qualitatively analyse the mass change patterns across Antarctica via independent component analysis (ICA), a statistics-based blind source separation method to extract signals from complex data sets, in an attempt to reduce uncertainties in the glacial isostatic adjustment (GIA) effects and improve understanding of Antarctic Ice Sheet (AIS) mass-balance. We extract the six leading independent components from gravimetric data acquired during the Gravity Recovery and Climate Experiment (GRACE) and GRACE Follow-On (GRACE-FO) missions. The results reveal that the observed continental-scale mass changes can be effectively separated into several spatial patterns that may be dominated by different physical processes. Although the hidden independent physical processes cannot be completely isolated, some significant signals, such as glacier melt, snow accumulation, periodic climatic signals, and GIA effects, can be determined without introducing any external information. We also observe that the time period of the analysed data set has a direct impact on the ICA results, as the impacts of extreme events, such as the anomalously large snowfall events in the late 2000s, may cause dramatic spatial and temporal changes in the ICA results. ICA provides a unique and informative approach to obtain a better understanding of both AIS-scale mass changes and specific regional-scale spatiotemporal signal variations.