Environ. Res. Lett. 10 (2015) 064011 I Takayabu et al We estimated the effect of changes in SST and atmospheric temperatures (ATM) on the MPI separately (supplementary table S1). The total contribution of SST and ATM to the MPI difference was −12 hPa (supplementary table S2). The contribution from SST changes alone ranged from −17 to −16 hPa, and that from ATM alone ranged from +3 to +4 hPa. The difference in the sign of the contributions from SST and ATM is consistent with the results of a previous study (Knutson and Tuleya 2004). Our result suggests that the SST change is the main contributor to the MPI difference between NAT and ALL. In the oceanic region around the Philippines, the SST was from +0.2 to +0.8 K higher in the ALL simulations compared to the NAT simulations (supplementary figure S2), and these higher SSTs can account for the MPI difference. Figure 5. Comparison of (a) MCP and (b) maximum surface wind speed between the NAT and ALL experiments. 3.2. Difference in the development of the Typhoon between NAT and ALL conditions Next, we compared the ALL results with the NAT results obtained by the cascade downscaling method and found that the influence of the present climate state in comparison to the hypothetical natural condition on the intensity of the worst case storm was robustly indicated. In 15 of the 16 WRF experiments, the MCP was lower in the ALL simulations (mean difference and standard deviation; −6.44 and 4.98 hPa) than in the NAT simulations (figure 5(a)). In addition, the maximum surface wind speed was stronger in 15 of 16 ALL simulations (mean difference and standard deviation; 2.89 m s−1 and 2.06 m s−1) (figure 5(b)). This intensification has a 1% significant level. To investigate whether the intensity of the worst case storm was accounted for by the differences of climate states between the real climate in November 2013 and the hypothetical natural climate, we used the maximum potential intensity (MPI) following Emanuel (1986) to examine the differences in the potential development of the typhoon between NAT and ALL. 6 3.3. Storm surge estimation Finally, we used the nonlinear shallow-water longwave model SuWAT (Kim et al 2008) to estimate increases in the water surface elevation due to the worst case storm. We focused on the maximum water elevation, the storm surge height at Tacloban, where Typhoon Haiyan caused catastrophic damage. To estimate a disaster at such a specific point, the largest permissible bias in the track compared to the best track is 50 km (empirically estimated from the maximum wind speed radius of the typhoon). Because only ten ensemble members satisfied this condition for each of ALL and NAT experiments, we applied SuWAT to estimate the surge at Tacloban to only 2 × 10 experiments. The average maximum surge height in the ten D3 experiments was 2.60 m (standard deviation 1.36 m) and 2.19 m (standard deviation 1.00 m) in ALL and NAT, respectively, an amplification of around 20% in ALL (supplementary table S3). The maximum surge height along the coast, simulated in experiment m02 (#1001), was 4.27 m in ALL, whereas in NAT it decreased to 3.80 m (figure 6, shown with land inundation survey data (Mori et al 2014, Tajima et al 2014) which influenced local bathymetry and wind waves). Although there is not real surge data, the reanalysis of maximum surge height shows 5.15 m (Mori et al 2014). While the surge height in ALL is underestimated in comparison with the reanalysis, the spatial pattern of maximum surge height is similar to reanalysis by Mori et al (2014). As the first seiche mode in Leyte Gulf increased the amplitude of the storm surge at Tacloban, the surge height at any specific point is very sensitive to the position of the TC track. We estimated the mean maximum surge height using the ALL surface wind amplification data (5.82%; the difference between NAT and ALL) with the NAT track data (ALL_Pseudo). The results showed that the difference in the track between the ALL and NAT simulations had limited influence on the mean maximum surge height. Thus, the difference in maximum surge height between them is explained mainly by the