THE SEYCHELLES ARCHIPELAGO
The Republic of Seychelles and the archipelago of the same name comprise some 155 islands lying in the western Indian Ocean north-northeast of Madagascar off the East African coast (Fig. 3). The northern group of islands is predominantly composed of Precambrian granitic rock. The remainder of the islands are scattered coral atolls that extend more than 1000 km to the southwest. Aldabra and the nearby island of Assumption, the most southwesterly islands in the Seychelles archipelago, lie just 300 km north of Madagascar. The granitic islands in the northern group are the subaerial parts of a micro-continent that was formed by rifting from Madagascar in the mid-Cretaceous (Collier et al., 2004). These islands sit on a broad (300 km x 150 km) continental shelf with water depths less than 200 m (Fig. 4). The many coral islands, on the other hand, have no such shelf (see Series DOS 604, Sheet Seychelles, published by the Government of the United Kingdom, Directorate of Overseas Surveys, 1983) and rise, in most cases, abruptly from abyssal depths.
Fig. 3: Indian Ocean showing location of granitic islands of the Seychelles archipelago (in box). Red star shows the epicentre of the 26 December 2004 Sumatra earthquake. Westernmost islands of Seychelles, including Aldabra, lie outside lower left corner of box north of Madagascar.
Fig. 4: Locations of the two largest granitic islands and bathymetry of Seychelles Bank, a continental fragment (Collier et al., 2004). Brown areas are <500 m depth.
Investigation of the tsunami involved collection of eyewitness accounts and documentary evidence while carrying out field surveys of tsunami run-up and inundation at representative sites along the coasts of Mah é and Praslin (Fig. 4). Evidence was obtained through interviews with eyewitnesses (including written accounts) and collection of digital still photographs and video imagery, along with compilation of documentary material supplied by government officials and the Seychelles Broadcasting Corporation. Information on impacts in other parts of the Seychelles archipelago was provided through other contacts.
Field methods included surveys of shore-normal profiles from the reef flat landward to the limits of tsunami inundation. These surveys were carried out using a pair of graduated rods fabricated on-site, with a tape measure for distance and an ocean horizon (or hand level if a horizon was not available) to measure the change in elevation between successive points along the profile (modified method of Emery, 1961; see also Krause, 2004). Notes on the surface material, vegetation, tsunamic deposits, and other features relevant to interpretation of tsunami impacts were made along the profile.
The profiles were positioned using dual-frequency, geodetic quality, Global Positioning System (GPS) instrumentation. Raw GPS observables (code and carrier phase) were recorded in either static or kinematic mode as required. A minimum of one static point was established along most profiles; water-level reference points were also occupied with GPS at some sites where no profiles were surveyed. Simultaneous GPS observations were made at reference stations established to support the surveys. These were located at Anse Forbans during surveys on Mahé (AF01) and at the Praslin airport (meteorological station) during surveys on that island (PRA1). In most cases, the distance between the base station and the survey site was less than 15 km (usually much less). A few points on Mahé (specifically survey sites in and north of Victoria) were more distant, and for these surveys an intermediate base station (BMG5) was established at the airport. However, due to data overlap gaps, BMG5 was not used in the final processing. AF01 proved adequate for these northern sites. Quality checks in the field were followed by post-processing using two different methods.
The raw data from the newly established GPS reference stations were processed using the on-line Precise Point Positioning (PPP) service at the Geodetic Survey Division of Natural Resources Canada. PPP utilizes precise orbital products from the IGS (International GPS Service) to provide coordinates consistent with the International Terrestrial Reference Frame (ITRF-IGb00). The precision of the PPP service is nearly consistent with that of phase-differential GPS processing, but there is no requirement to collect GPS data simultaneously at a reference site. Therefore PPP is ideally suited for an application such as this survey, where it was difficult to locate stable and secure reference stations with established coordinate values. The coordinates, as determined by the PPP processing were subsequently used to constrain the reference station positions in the phase-differential processing carried out using Ashtech Solutions TM version 2.7 software. This software permitted the processing of both static and kinematic data collected along the profiles. The GPS ellipsoid elevations (WGS-84) were converted to orthometric heights using the EGM96 geoid model in Ashtech Solutions TM . Additional conversions were carried out using the on-line calculator at UNAVCO ( http://sps.unavco.org/geoid/ ).
The GPS surveys were tied to surveyed control points on Mahé to determine the geoid undulation value appropriate for conversion to elevation above mean sea-level datum. The GPS surveys enabled the determination of elevations for the upper limits of inundation, as well as reference to the tide gauge during the tsunami event. Leveling data provided by the Seychelles Meteorological Service (Lands and Survey data) and by Shikiko Nakahara (SOEST, Hawaii) enabled us to determine the vertical offsets between tide-staff zero (TSZ, the datum for the water-level data), mean sea-level datum (MSLD, the vertical datum for land surveys), and Chart Datum (CD), as follows:
MSLD = TSZ + 0.97 m
TSZ = CD + 0.07 m
MSLD = CD + 1.04 m
GPS observations at TP4 (a tide-gauge reference pin set into the cap of the seawall on the north side of the tide-gauge jetty) and at BMG5 (a benchmark near the radiosonde facility) yielded a mean geoid undulation of -40.48 m) for conversion of elevations to MSLD (Table 1) for sites on Mahé. It is not known what vertical datum these existing survey points are referenced to. The mean geoidal undulation for this area derived using EGM96 Geoid Model is -41.175m, a difference of 0.695m.
Another control point, 1704, was established in 1998 on the glide-path at Seychelles International Airport, with an observed ITRF94 ellipsoid elevation of -38.21 m (MAPS geosystems, 1998). That project reported standard deviations of height ranging from 0.022 m to 0.029 m for the ITRF network on Mahé, Coétivy, Desroches, Assumption and Farquhar, using vectors between these five islands and to International GPS Service (IGS) stations at Diego Garcia, Mahé, and Malindi ( Kenya). Holding the ITRF network points fixed, they then established control to the remaining islands, including Praslin, with standard deviations ranging from 0.030 m to 0.085 m in height. Our elevation for point 1704, based on a 30 min occupation, was ‑38.0113 ± 0.2498 m (2 s ), referenced to the ITRF (IGb00) reference frame. For our surveys on Praslin, we were unable to relocate the 1998 control points because subsequent airport construction work destroyed them. Consequently, an autonomous control site was established and occupied over 2 days. Elevations were computed using the same approach adopted on Mahé, comparing EGM96 orthometric heights with elevations determined by reference to local sea level at the time of surveys.
|
