Long-period spectral features of the Sumatra-Andaman 2004 earthquake rupture process

Abstract

The goal of this study is to investigate the spatial variability of the seismic radiation spectral content of the Sumatra–Andaman 2004 earthquake. We determine the integral estimates of source geometry, duration and rupture propagation given by the stress glut moments of total degree 2 of different source models. These models are constructed from a single or a joint use of different observations including seismology, geodesy, altimetry and tide gauge data. The comparative analysis shows coherency among the different models and no strong contradictions are found between the integral estimates of geodetic and altimetric models, and those retrieved from very long period seismic records (up to 2000–3000 s). The comparison between these results and the integral estimates derived from observed surface wave spectra in period band from 500 to 650 s suggests that the northern part of the fault (to the north of 8◦N near Nicobar Islands) did not radiate long period seismic waves, that is, period shorter than 650 s at least. This conclusion is consistent with the existing composite short and long rise time tsunami model: with short rise time of slip in the southern part of the fault and very long rise time of slip at the northern part. This complex space-time slip evolution can be reproduced by a simple dynamic model of the rupture assuming a crude phenomenological mechanical behaviour of the rupture interface at the fault scales combining an effective slip-controlled exponential weakening effect, related to possible friction and damage breakdown processes of the fault zone, and an effective linear viscous strengthening effect, related to possible interface lubrication processes. While the rupture front speed remains unperturbed with initial short slip duration, a slow creep wave propagates behind the rupture front in the case of viscous effects accounting for the long slip duration and the radiation characteristics in the northern segment

A Bayesian source model for the 2004 great Sumatra‐Andaman earthquake

“…The standard deviation on the moment is very small −0.2 × 10 22 N m ( M w = 9.25 ± 0.01)—but it could vary more significantly depending on the assumed elastic structure. This moment magnitude is close to the magnitude ( M w 9.3) obtained by normal mode analyses [ Stein and Okal , ; Park et al , ; Clévédé et al , ], which also depend on the assumed elastic structure.…”

“…Such a long duration made conventional teleseismic body wave inversions for the distribution of fault slip difficult because of the interference between direct and reflected waves. Surface waves [ Ammon et al , ] and normal modes [ Stein and Okal , ; Park et al , ; Clévédé et al , ] were also used to analyze the source, but the information content of low‐frequency records is inherently much less than high‐frequency records and thus only provided low resolution source models.…”

“…An intriguing feature of this exceptional earthquake is the very slow rupture inferred at the northern end of the fault where low seismic radiation was detected [ Ammon et al , ; Tsai et al , ; Vallée , ] but where large geodetic offsets were measured [ Gahalaut et al , ] and intense aftershocks occurred [e.g., Lay et al , ]. The anomalously large low‐frequency content in seismic records associated with a northward migration of the centroïd compared to the centroïd obtained from surface waves [ Stein and Okal , ; Park et al , ; Clévédé et al , ] confirms that large coseismic moment was released in this Andaman section. The atypical behavior of one near‐field tide gauge at Port Blair (PB in Figure ) suggests that the rise time could be on the order of 30 min [ Singh et al , ], although this inference relies only on one observation.…”

“…Such a long rise time would produce a tsunami of smaller amplitude and would be difficult to model since we have very little constraint on the exact slip history. One interpretation of these observations is that the slow rupture had an early fast slip component and then continued slipping slowly [ Lay et al , ; Singh et al , ; Clévédé et al , ].…”

Structural and thermal control of seismic activity and megathrust rupture dynamics in subduction zones: Lessons from the Mw 9.0, 2011 Tohoku earthquake

“…A first example is North-Central Lesser Antilles (Laigle et al, 2013), where the subducting cold slab is associated with low rates of slab dehydration and mantle serpentinization, deep mantle corner flow, inducing an extended seismogenic slab-mantle interface and similar alongdip plate boundary segmentation. A second example is Central Chile (Comte et al, 1986;Beck et al, 1998) and Northern Chile (Peyrat et al, 2010;Fuenzalida et al, 2013) where recent sequences of M7-8-class interplate earthquakes suggest along-dip segment interactions. Other examples are the cold-slab subduction along the Alaska-Aleutian zone, where along-dip segment interactions bear some similarities with Tohoku (Yomogida et al, 2011;Wada and Wang, 2009), and the Sumatra-Andaman subduction zone, which features along-strike segments with strong variations of slab-age and dehydration rates (Moeremans et al, 2014) suggesting trench-parallel variations of the along-dip segmentation.…”