Re-analysis/prediction of Typhoon Vera (1959) Project: ReVera -Isewan Typhoon like a Katrina hitting Japan 50 years ago-
Kotaro Bessho, MRI, Tsukuba, Ibaraki, Japan; and T. Nakazawa, H. Kamahori, E. Shindo, M. Hara, T. Kawabata, M. Kunii, and N. Kohno
50 years ago, Super Typhoon Vera (1959) made landfall in Kii peninsula in Japan around 0900 UTC on 26 September in 1959. It brought tremendous disasters such as storm surge to Japan islands, especially around the Ise Bay area. It was the most tragic meteorological disaster in Japan after the World War II. Indeed, total amount of death toll was more than 5,000. Because of its massive damage for the Japanese society, Vera is one of the memorable typhoons in Japan, given a special name as ‘Isewan (Ise Bay) Typhoon' from Japan Meteorological Agency (JMA). After Vera, Japanese government enacted a law of Disaster Countermeasure Basic Act, and Vera is regarded as a baseline for bank construction, wind resistant design, evacuation plan, and so on. In the view point of changing the national disaster planning drastically, Vera is rated on par with Katrina (2005) landfalling in United States. The track forecast of the time for Vera was very accurate. However, the forecast of moving speed of Vera at 50 years ago was much lower than the analysis. In addition, the forecast of sea-level height at the Ise Bay was at most 100 to 150 cm, while the observation was 389 cm that was mush higher than the forecast. Recent advances of objective numerical reanalysis system enable us to obtain long-term reanalysis data. JMA starts the project associated with the long-term re-analysis during the period from 1958 to 2012, called as the ‘JRA-55' project. This is a succeeding project for the ‘JRA-25' project focused on the period from 1979 to 2003. Using the reanalyzed dataset and sophisticated numerical models, we can simulate past remarkable meteorological phenomena such as typhoons. Here we performed numerical predictions for validating the Vera's predictability using the latest forecast technique. We used an interim version of JRA-55 as initial conditions for the track, intensity and storm surge predictions of Vera. We performed track predictions using the global model with a horizontal grid spacing of 60 km with the boundary conditions every 12 hours four days before Vera made landfall in Japan. In all cases, predicted Vera made landfall in Japan. Among them, the forecast case at the initial time, 0900 JST on 24 September in 1959 showed the closest to the best track. Then ensemble forecasts using 11 members were performed using the same initial condition. The forecasts indicate that all members can predict the realistic track making landfall in Japan. The landfall places are widely distributed from the Kyushu islands to the Kanto plain. On the other hand, the tracks are almost close to the best track until when Vera passed across 30°N. To predict the intensity of Vera and its induced storm surge more accurately, high resolution mesoscale model is needed to prepare the initial condition as close to the observation as possible. For this purpose, JNoVA (JMA Non-hydrostatic model Variational data Assimilation system) was used for implementing the mesoscale analysis for 24 hours from 0900 JST on 25 September in 1959 with a 3-hour assimilation window. Here we assimilate the dropsonde data into the analysis system. The data for Vera were obtained from US military aircraft reconnaissance and were archived in the JMA old report. We performed 36 hours forecast experiment using this analysis result and the JMA non-hydrostatic model with a grid spacing of 5 km from 0900 JST on 26 September in 1959. The result shows the results of numerical experiment. Vera's track made landfall in the Kii peninsula and precipitation amount caused by Vera are successfully simulated. In addition, the time lag between simulated landfall and analyzed one based on the best track is within one hour. Figure displays a pseudo-satellite image calculated by the output of the numerical simulation. This image is almost similar to an actual infrared satellite image. This kind of realistic image was not acquired 50 years ago because the first geostationary meteorological satellite (GMS) over the Pacific Ocean was launched in 1977. After conducting the numerical simulation by the mesoscale model, storm surge predictions were performed using the Princeton Ocean Model and the result of numerical simulation as an atmospheric forcing. The predicted sea-level height at the port of Nagoya was very close to the observation. From these above-mentioned experimental results, we can obtain highly accurate prediction for Vera using the latest forecast technique. One of the important points is that the numerical model used in the present numerical simulations is almost similar to the operational one in JMA. This suggests that we have high potential capability to predict intense typhoons like Vera using the JMA operational prediction system.
Extended Abstract (284K)
Session 16C, Tropical Cyclone Modeling: Structure and Intensity II
Friday, 14 May 2010, 10:15 AM-12:00 PM, Arizona Ballroom 10-12
Browse or search entire meeting
AMS Home Page