Rat Islands Tsunami
Observation data from the expanding tsunami observation network will continue to provide more data for model verification. NOAA's National Geophysical Data Center (http://www.ngdc.noaa.gov/seg/hazard/tsu.shtml), NOAA's Tsunami Warning Centers (http://wcatwc.arh.noaa.gov/), and NOAA's Center for Tsunami Research websites provide updated information on the latest tsunami data. Here, we present the data used for the first real-time model forecast test as an example of data use for model verification.
The magnitude earthquake was located on the shelf near Rat Islands, Alaska, on 17 November 2003 and generated a tsunami. This tsunami provided good data for testing operational models, since the tsunami was detected by three tsunameters located along the Aleutian Trench and was recorded at many coastal locations (Titov et al., 2005). This was the first real-time tsunami detection by the newly developed tsunameter system (DART). In addition, for the first time, tsunami model predictions were obtained during the tsunami propagation, before the waves had reached many coastlines. Here, the combined use of tsunami propagation and inundation models is required for simulation of tsunami dynamics from generation to inundation. The test requires matching the propagation model data with the DART recording to constrain the tsunami source model (Fig. 1). If a finite-difference method on a structured grid is used, several nested numerical grids would allow ``telescoping'' from a coarse-resolution propagation model into a high-resolution inundation model with a model grid of at least 50 m resolution. If an unstructured grid method is used, a single grid may include enough resolution near the coast. The data-constrained propagation model should drive the high-resolution inundation model of Hilo Harbor. The inundation model is to reproduce the tide gage record at Hilo (Fig. 2). Since this benchmarking is required for the forecasting models, it is essential to model 4 hr of Hilo Harbor tsunami dynamics in 10 min of computational time.