Assessment of the Initial Response from Tsunami Monitoring Services Provided to the Northeastern Caribbean

A mag 6.4 earthquake offshore northern Puerto Rico earlier this year (1/13/14) is a reminder of the high risk of earthquakes and tsunamis in the northeastern Caribbean. Had the magnitude of this event been 0.1 larger (M 6.5) a tsunami warning would have been issued for the Puerto Rico-Virgin Islands (PRVI) region based on the West Coast Alaska Tsunami Warning Center (WCATWC) and Puerto Rico Seismic Network (PRSN) response procedures at the time. Such an alert level would have led local authorities to issue evacuation orders for all PRVI coastal areas. Since the number of deaths associated with tsunamis in the Caribbean region is greater than the total casualties from tsunamis in the entire US (including Hawaii and Alaska coasts) having an effective and redundant warning system is critical in order to save lives and to minimize false alarms that could result in significant economic costs and loss of confidence of Caribbean residents.

We are evaluating three fundamental components of tsunami monitoring protocols currently in place in the northeastern Caribbean: 1) preliminary earthquake parameters (used to determine the potential that a tsunami will be generated and the basis of tsunami alert levels), 2) adequacy of the tsunami alert levels, and 3) tsunami message dissemination. We compiled a catalog of earthquake locations (2007-2014) and dissemination times from the PTWC, WCATWC and NEIC (final locations). The events were classified into 3 categories: local [17°-20°N, 63.5°-69°W], regional (Caribbean basin) and distant/teleseismic (Atlantic basin). A total of 104 local earthquakes, 31 regional and 25 distant events were analyzed. We found that in general preliminary epicentral locations have an accuracy of 40 km. 64% of local events were located with an accuracy of 20 km. The depth accuracy of local events shallower than 50 km, regional and distant earthquakes is usually smaller than 30 km. For deeper local events the error distribution shows more variability (-32 to 81 km); preliminary locations tend to underestimate depth. A trade-off between epicentral location and depth was observed for several local events deeper than 50 km.