Technical Comments

Comment on "The Great Sumatra-Andaman Earthquake of 26 December 2004"

See allHide authors and affiliations

Science  02 Dec 2005:
Vol. 310, Issue 5753, pp. 1431
DOI: 10.1126/science.1118950


Lay et al. (Research Articles, 20 May 2005, p. 1127) estimated a 600-km length for the tsunami source region. Adding tide-gauge data from Paradip, the northernmost of the Indian east-coast stations and therefore the most critical constraint on the northern extent of the source, we estimate that its length was greater by ∼30%.

Determining the length of the tsunami source region is one of the keys to understanding the complex nature of the 2004 Sumatra-Andaman earthquake. Assuming an instantaneous rupture and total slip on the fault, Lay et al. (1) used backward ray tracing to estimate a source region extending up to 600 km (∼9°N) north-northwest of the epicenter. The extent increased to 10°N on considering the time delay due to finite rupture propagation and slip rise.

In the analysis of Lay et al., the northern extent of the tsunami source region was constrained by the tsunami travel times reported from the tide gauges at Chennai, Visakhapatnam, and Port Blair (Fig. 1). At the time of the tsunami, however, the clock of the Port Blair tide gauge was 46 min ahead of the actual time (2). A data gap of 24 min (35 to 59 min after the earthquake) in this record adds to the uncertainty about the tsunami arrival time at Port Blair (2). Hence, we excluded Port Blair from our analysis and reestimated the length of the tsunami source by including the newly available tide-gauge data from Paradip.

Fig. 1.

Backward wavefronts for Paradip (red), Visakhapatnam (green), and Chennai (blue) corresponding to their respective observed travel times. The asterisk marks the epicenter of the earthquake. The northern extent (∼9°N) reported by Lay et al. is indicated by arrow A. Arrow B marks the new estimate (∼11°N).

The tide gauges along the east coast of India [maintained by Survey of India (SOI)] show that the entire coast was struck by tsunami waves at almost the same time (3). That Paradip, the northernmost of the stations (86.70°E, 20.26°N), was hit at almost the same time (3, 4) as Chennai and Visakhapatnam, and that the tsunami amplitude here was comparable to that at these stations (Fig. 2), motivated us to examine the compatibility of the tsunami source region reported by Lay et al. with the recorded tsunami travel time to Paradip (Paradip travel time was not included in their analysis). A forward ray tracing simulation shows that, although the rays starting from the estimated tsunami source region reach Chennai and Visakhapatnam in the recorded travel times, this is not the case with Paradip, implying that the estimated source region is not consistent with the travel times of all three stations. The tsunami source region must have extended farther north for Paradip to be hit at the recorded travel time.

Fig. 2.

The residual (cm) of the tide-gauge record. The tsunami arrival time (UTC) is shown in the bottom right corner.

Including the Paradip travel time as a constraint (in addition to Visakhapatnam and Chennai) and assuming instantaneous rupture propagation and total slip on the fault, we used backward ray tracing (5) to reestimate the northern extent of the tsunami source region. The backward wavefront for Paradip is in the admissible regions of those of Visakhapatnam and Chennai, only ∼11°N (Fig. 1), implying a source region extending into the Andaman Islands and ∼900 km long (from the epicenter). This is ∼200 km (∼30%) longer than the estimate of Lay et al. Including the time delay (1) due to tsunami excitation may extend the effective tsunami source region farther north.

This new northern extent constrains the tsunami travel time to Port Blair to ∼30 min (Fig. 3), close to the data gap in the tide-gauge record. The tsunami had already struck Port Blair by the time the tide gauge started recording again (2). Thus, an implication of the constraint imposed by the Paradip travel time is that tsunamigenic slip must have occurred over a longer arc than estimated earlier (1). The increase in length also indicates a much more destructive event than previously thought. Furthermore, our revised estimate has direct implications (6) for slip distribution on the fault and should help constrain the set of possible geophysical solutions, leading to a better understanding of the processes involved in the 2004 Sumatra-Andaman mega-event.

Fig. 3.

The approximate 30-min backward wavefront of Port Blair (brown) is consistent with those of Paradip (red), Visakhapatnam (green), and Chennai (blue).

Supporting Online Material



References and Notes

View Abstract

Navigate This Article