Proceedings of the Sixth International Conference on Asian and Pacific Coasts (APAC 2011) December 14 16, 2011, Hong Kong, China PROJECTION OF FUTURE STORM SURGE DUE TO CLIMATE CHANGE AND ITS UNCERTAINTY A CASE STUDY IN THE TOKYO BAY TOMOHIRO YASUDA, NOBUHITO MORI, SOTA NAKAJO, HAJIME MASE Disaster Prevention Research Institute, Kyoto University Gokasho, Uji, Kyoto 611-0011, Japan YUTA HAYASHI Graduate School of Engineering, Kyoto University Kyotodaigaku-Katsura, Nishikyo-ku, Kyoto 615-8540, Japan YUICHIRO OKU Osaka City Institute of Public Health and Environmental Sciences 8-34, Tojo-cho, Tennoji-ku, Osaka 543-0026, Japan It has been discussed that intensity of typhoon may increase in the future due to climate change. It is important to estimate the probable maximum magnitude of storm surges under the future climate for coastal disaster mitigation. In the present study, sensitivity of storm surge projection due to storm track in Tokyo Bay is discussed by employing the hundreds of different severe meteorological conditions generated by the potential vorticity inversion method. Ensemble numerical results show the most hazardous tropical cyclone track is different from most intense tropical cyclone track in Tokyo Bay. Estimated maximum storm surge in the Tokyo Bay is found to be 1.4 m which is the same level as the historical highest record. 1. Introduction Crown height of seawall is generally designed by the past highest storm water level or an estimated storm surge by a specific (standard) tropical cyclone (TC) plus the astronomical spring tide level. It is likely expected to increase intensity of TC in the future due to climate change. It is important to estimate the probable maximum magnitude of storm surges in the future climate for coastal disaster mitigation. However, future climate projections contain uncertainty of models and scenario especially for the TC intensity projection due to difficulty of TC modeling. In the present study, we examined the uncertainty of storm surge projection as a case study for the Tokyo Bay, Japan. The ensemble storm 369
370 surge simulations are conducted combining the storm surge model and the perturbative meteorological conditions generated by the Potential Vorticity Inversion (PVI) method (Davis and Emanuel, 1991; Yoshino et al., 2007) from the extreme TC in the GCM outputs of the future climate experiment. 2. Future extreme TC Spatial resolution of GCM outputs is not fine enough to assess quantitative impacts of TC activity. Therefore, GCM outputs should be downscaled by Regional Climate Model (RCM). However, given the limitation of simulation cost, it is not realistic to carry out downscale GCM outputs for all periods dynamically. In the present study, future extreme TC is identified from GCM outputs of the future climate by using the Extreme Event Index (EEI) (Oku et al., 2009). The EEI is the risk index estimated from grid climate data such as wind speed or precipitation data. The JMA/MRI (Japan Metrological Agency/ Meteorological Research Institute) Atmospheric GCM outputs (Kitoh et al., 2009) were employed as grid climate data. The EEI is defined as i j I p( i) q( i, j) u( i, j), (1) where i is the place and j is the rank, p(i) is the weight coefficient depending on place and varies with disaster accounted, q (i, j) is the weight coefficient depending on rank and u (i, j) is the climate data (wind speed in this study). The index I represents strength of extreme event. The track of future extreme TC which has the maximum value of EEI (target place i is the Tokyo Bay) is shown in Fig. 1. Figure 1. Future extreme TC track by AGCM.
371 The future extreme TC around middle latitude of the Northwest Pacific of which central atmospheric pressure and maximum wind speed are 930 hpa and 30 m/s respectively; the TC passes near the Tokyo Bay without landfall. 3. Ensemble numerical experiments of future extreme TC The impact of TC is sensitive to the tracks. The influence of TC will be severer if the future extreme TC passes the worst course from the original track shown in Fig.1 to Tokyo area. It is necessary to control TC tracks in order to consider the possibility of taking other paths keeping the dynamical characters. In the previous method, meteorological field was directly moved parallel. Meteorological field controlled by previous method, however, do not satisfy laws of physics since wind field and temperature field are related each other. In the present study, the PVI method is employed to produce an initial TC condition without dynamical meteorological inconsistency. This procedure allowed that ensemble numerical simulations were conducted for mesoscale weather. 3.1. Downscaling of meteorological field The spatial resolution of JMA/MRI AGCM (TL959) used for future TC modeling in the global scale is about 20 km. Then, AGCM outputs were dynamically downscaled by using the Weather Research and Forecasting (WRF) model that is a practical mesoscale weather prediction system. The minimum horizontal resolution of WRF is 5 km and other simulation conditions of WRF are shown in Table 1. Table 1 Simulation conditions of WRF. WRF-ARW ver3.0.1.1 grid resolution x 5km grids 453 x 453 x 50 time step t 30sec. initial time 2093/08/30 06:00 analysis period 2093/08/31 06:00 ~ 09/02 00:00 boundary input JMA/MRI GSM (AGCM20) cumulus scheme Kain-Fritsch microphysics WSM6 PBL scheme MYJ Level 2.5 outputs wind speed at 10m (1min.) precipitation, surface pressure (5min.) others (60min.)
372 The meteorological field at start time of simulation is used as input data into WRF, and numerical simulation is conducted for 54 hours. TC tracks can be eventually controlled by inputting another initial condition into WRF. PVI method was employed in order to control initial condition with characteristic of future extreme TC. Potential vorticity is extracted from AGCM outputs of wind speed, temperature and atmospheric pressure field, and then the location and intensity of potential vorticity is changed. The modified potential vorticity is converted to meteorological field by using PVI method. This metrological field is input into WRF. PVI method enables to produce new initial conditions easily without dynamical meteorological inconsistency. The initial location of future extreme TC was perturbed every 5 km over 200 km x 100 km (189 cases) far from the Tokyo Bay. 3.2. Results of ensemble simulation A set of ensemble simulation results is shown in Figure 2. In the left hand panel of Fig. 2, the color solid lines show simulated TC tracks and their colors indicate the central atmospheric pressure of TC. In right hand figures of Fig. 2, the upper panel shows TC tracks with the maximum wind speed at the central Tokyo during this event indicated by colors. The lower panel shows the relationship between the maximum wind speed at the central Tokyo during this event and the longitude when the TC has passed at 34 N indicated by purple line in the upper panel. A red line indicates the longitude of the central Tokyo. Most of the modified TCs hit the eastern Japan, but the future extreme TC shown in Fig. 1 does not make landfall. Even though initial locations were shifted parallel, track and intensity of TC were not monotonically shifted due to topological and sea Figure 2. Results of ensemble simulation by PVI method.
373 surface thermal effects in the mesoscale weather simulation. The TCs passing through the area between two black lines are more hazardous for the central Tokyo and maximum wind speeds are 24-32 m/s. The worst TC made landfall and went northward at Sagami Bay located about 50 km west of Tokyo Bay. Its maximum wind speed was 31.8 m/s. 4. Projection of future storm surge 4.1. Storm surge simulation Surge-WAve-Tide coupled model (SuWAT) developed by Kim et al. (2009; 2010) was employed as the storm surge simulation model. The SuWAT consists of storm surge module and wave module and these modules are coupled together. The water elevations and currents calculated by storm surge module are given to wave module in the next time step, and the sea surface drag coefficient and radiation stress calculated by wave module are give to storm surge module in the next time step. The wave module in SuWAT is SWAN (Simulating WAve Nearshore) and the storm surge module in SuWAT is long wave equation with bottom friction. The computation time steps for the surge and wave modules were 1 s and 600 s, respectively. The time interval for exchanging information between the surge and wave modules was set as 600 s. SuWAT employs multinesting scheme. In this study, 4 domains were used from computation as shown in Fig. 3. Figure 3. Simulation domains with mean water depth. The spatial resolution of the coarsest domain is 8,100 m and the one of the finest domain (right panel of Fig. 3) is 300 m. The numbers in right panel indicates locations where storm surge simulation results are assessed following
374 section. In this study, tide was not considered to save computational cost although astronomical tide is possible to simulate in SuWAT. The numerical storm surge simulations were conducted for 26 TCs obtained from ensemble simulation of future extreme TC in Chapter 3. These selected TCs bring down stronger wind in the entire data set in the central Tokyo. 4.2. Results of storm surge simulations Figure 4 shows time series of simulated results of 26 TCs at location indicated by #6 in Fig. 3. The top panel is the atmospheric pressure, middle one is the wind speed and bottom one is the water surface elevation. Figure 5 shows maximum simulated storm surge each cases at each location (#1~12). Colors indicate different simulation cases in both Figs. 4 and 5. Figure 4 Simulated results at location #6. Figure 5. Maximum storm surges at each location in Tokyo Bay. The maximum magnitude of storm surge at location #6 was 1.4 m which is the same level as the historical highest record by Typhoon Kitty in 1949. Typhoon Kitty had great impacts on Tokyo bay area. The TC which brought
375 down the maximum storm surge did not have either minimum atmospheric pressure or maximum wind speed, and this characteristic was also presented in the results of other locations. The storm surge heights were found to be affected by the track more than by the intensity. It is seen that two cases brought down higher storm surge at most locations. Tracks of these two TCs and Typhoon Kitty are shown in Fig. 6. It was found that these two TC tracks are similar to Typhoon Kitty s one. When the future extreme TC takes these tracks, storm surges on Tokyo Bay become very high. Figure 7 shows the potential maximum storm surges estimated from all simulation results. The potential maximum storm surge in a closed-off region of the Tokyo Bay was estimated about 1.3 to 1.4 m. 5. Conclusions Figure 6. Tracks of worst two TCs and Typhoon Kitty. This study investigated the uncertainty of storm surge projection in the future climate. A series of ensemble dynamic downscaling of the future extreme TC from TC in MRI-AGCM were conducted by using PVI method. The worst TC brought maximum wind speed of 31.8 m/s by ensemble runs. The ensemble simulation of future storm surges was also conducted in Tokyo Bay, and the maximum storm surge was 1.4 m which is comparable with the historical highest records. Dynamic downscaling of TCs by PVI method is useful to assess the impact of TC track sensitivity for the severe storm surge disasters in the future climate.
376 Figure 7. Potential maximum magnitude of storm surges in Tokyo Bay by future extreme TCs. Acknowledgements The present study was conducted under the framework of the Projection of the change in future weather extremes using super-high-resolution atmospheric models, being supported by the KAKUSHIN Program and the Kakenhi Grantin-Aid of the Ministry of Education, Culture, Sports, Science, and Technology (MEXT). A part of this study was supported by Service Center of Port Engineering, Japan. References 1. Davis, C.A. and Emanuel, K.A. (1991). Potential vorticity diagnostics of cyclogenesis. Mon. Wea. Rev., Vol. 119, pp. 1929-1953. 2. Kim, S.Y., Yasuda, T. and Mase, H. (2009). Numerical Analysis of Effects of Tidal Variations on Storm Surges and Waves. Applied Ocean Res., Vol.30, pp.311-322, doi:10.1016/ j.apor.2009.02.003. 3. Kim, S.Y., Yasuda, T. and Mase H. (2010). Wave set-up in the storm surge along open coasts during Typhoon Anita. Coastal Eng., Vol.57, pp.631-642, doi:10.1016/j.coastaleng.2010.02.004. 4. Kitoh, A., Ose, T., Kurihara, K., Kusunoki, S., Sugi, M. and KAKUSHIN Team-3 Modeling Group. (2009). Projection of changes in future weather extremes using super-high-resolution global and regional atmospheric models in the KAKUSHIN Program: Results of preliminary experiments. Hydrological Research Letters, 3, pp.49-53, doi:10.3178/hrl.3.49. 5. Oku, Y., Yoshino, J., Ishikawa, H., Takemi T. and Nakakita E. (2010). Maximum Damage Estimation by Multi-Track Approach of Extreme Typhoon in Future Climate. Annuals of Disas. Prev. Res. Inst., Kyoto Univ., No. 53 B. (in Japanese)