Geoscience Reference
In-Depth Information
Adequacy of the Geographical Coverage of the Coastal Sea Level Gauge Network
Following the disastrous 2004 Indian Ocean tsunami, many additional global sea level
observing stations have become available for the purpose of tsunami detection and warning,
including those enabled in the United States by P.L. 109-424. Despite this increase in the num-
ber of near-real-time-reporting, rapid-sampling coastal sea level gauges, a map of the sea level
station coverage (e.g., Figure 4.4 for the Paciic Ocean) reveals that large regions with no cover-
age remain, such as Central America and southern Mexico, the Kuril Islands north of Japan,
and most of the Caribbean Islands, as pointed out previously (Bernard et al., 2007). In addition,
this dependence on data supplied by foreign agencies, although mitigated somewhat by the
redundancies and overlaps in coverage, exposes a vulnerability of the tsunami detection and
warning activities to potential losses in data availability.
A recent earthquake in the Caribbean illustrates the issue of coverage. On May 27, 2009,
a magnitude 7.3 earthquake occurred off the coast of northern Honduras. Eight minutes after
the earthquake, the PTWC issued a Tsunami Watch for Honduras, Belize, and Guatemala. Worst-
case-scenario tsunami forecast models suggested tsunami amplitudes up to nearly 1 m given
initial earthquake source parameters. No rapidly sampled, near-real-time sea level gauges exist
in the western Caribbean, so the PTWC could only wait for visual reports. After 74 minutes, the
PTWC canceled the watch based on the following, in the PTWC's own words:
“ . . . This center
does not have access to any real-time sea level gauges in the region that would be used to
quickly detect and evaluate the tsunami if one were present. However, enough time has passed
that any nearby areas should already have been impacted. Therefore, this center is canceling
the tsunami watch it issued earlier” (Paciic Tsunami Warning Center Message, May 27, 2009).
Gaps in the coastal sea level network exist, such as revealed by the Honduran earthquake
in May 2009. No analysis has been undertaken to evaluate critical coverage gaps with regards
to the tsunami warning decision process. Furthermore, no analysis has been undertaken to
determine the relative importance of each existing coastal sea level gauge to the tsunami
warning decision and evacuation decision processes. Although there is some degree of redun-
dancy in coverage in the current sea level gauge network for some purposes, there has been
no evaluation of the associated risk and the vulnerability of the warning process to failures of
single or multiple stations.
The spacing of sea level gauges for the purpose of tsunami detection is sparse, because it is
now known that tsunamis can be quite directional, focusing the majority of their energy within
a narrow sector, perpendicular to the sealoor rupture direction. For instance, Figure 4.3 dis-
plays the modeled beam pattern of a small tsunami generated by a large (magnitude 8.3) Kuril
Islands earthquake on November 15, 2006. Given the array of sea level gauges in Figure 4.4,
it is obvious that the maximum amplitudes of this tsunami were not observed in near-real
time. Because DART stations were not yet in place off the Kuril Islands, only the Midway Island
(28.2
o
N, 177.4
o
W) station at the far northwest end of the Hawaiian archipelago provided sig-
niicant advance notice to forecasters of the possible size of the tsunami at the main Hawaiian
Island to the southeast. Had the Midway Island station been temporarily inoperative, forecasters
