Spatiotemporal Analysis of the GRACE-Derived Mass Variations in North America by Means of Multi-Channel Singular Spectru
We apply multi-channel singular spectrum analysis (MSSA) to infer the main spatiotemporal modes of mass variability in North America derived from GRACE monthly gravity field data. MSSA is a data-adaptive method for analyzing time lagged maps of variabilit
- PDF / 1,171,390 Bytes
- 8 Pages / 547.087 x 737.008 pts Page_size
- 43 Downloads / 151 Views
Spatiotemporal Analysis of the GRACE-Derived Mass Variations in North America by Means of Multi-Channel Singular Spectrum Analysis E. Rangelova, W. van der Wal, M.G. Sideris, and P. Wu
Abstract We apply multi-channel singular spectrum analysis (MSSA) to infer the main spatiotemporal modes of mass variability in North America derived from GRACE monthly gravity field data. MSSA is a data-adaptive method for analyzing time lagged maps of variability on regional and global scales. The method proves useful in studying the annual and longterm continental water mass variations and the secular deformation signal associated with glacial isostatic adjustment (GIA) of the Earth. We study the capabilities of the MSSA method using simulated spatiotemporal data series and address issues such as lag-window length, spectral mixing, and significance of the extracted modes. We investigate two cases using the GRACE RL-04 data. We analyze water mass variations derived from the GRACE data (corrected for GIA) and compare our results with the main modes extracted from the GLDAS/Noah and WGHM continental water storage models. Good agreement between the annual amplitudes exists in the Cordillera region, the WGHM model being closer to GRACE compared to GLDAS/Noah. Furthermore, we model the North American GIA signal using the GRACE data corrected for hydrology. Two peak signals are observed west (9–11 mm/year uplift rate) and southeast (11–13 mm/year) of Hudson Bay in agreement with the multi-dome geometry of the North American ice sheet in the ICE-5G model.
E. Rangelova () Department of Geomatics Engineering, Schulich School of Engineering, University of Calgary, Calgary, AB, Canada, T2N 1N4 e-mail: [email protected]
72.1 Introduction The Gravity Recovery and Climate Experiment (GRACE) project (Tapley et al., 2004) delivers models of the Earth’s gravity field every month to model continental water mass variations (e.g., Schmidt et al., 2006), deformation related to glacial isostatic adjustment (GIA) of the Earth (e.g., Tamisiea et al., 2007), and glacier/ice sheet mass changes (e.g., Luthcke et al., 2006), among others. Chambers (2006), Schrama et al. (2007) and Rangelova et al. (2007) have shown that the statistically-based orthogonal bases of the method of principal component/empirical orthogonal functions (PC/EOF) analysis can be successfully applied to model the GRACE-derived mass variations at regional and global scales. In this paper, we investigate the generalization of the PC/EOF analysis, i.e., the method of multi-channel singular spectrum analysis (MSSA) (e.g., Allen and Robertson, 1996), to analyze the mass variations in North America derived from GRACE monthly gravity field data. Working with maps of variability that are lagged in time, MSSA identifies coherent spatiotemporal patterns (empirical orthogonal functions) and principal components, in contrast to the conventional PC/EOF analysis, which identifies spatial patterns and their time evolutions. Similar to PC/EOF, MSSA allows for modelling inter-annual and long-term
Data Loading...
