Environ. Res. Lett. 10 (2015) 064011 I Takayabu et al Figure 1. Schematic diagram of multi-time ensemble PGWD method. Pall et al (2011) proposed a new approach for quantifying the roles of external drivers and natural variability on specific extreme events by performing two types of ensemble simulations with an atmospheric general circulation model (AGCM). One took into account anthropogenic changes in the well-mixed greenhouse gas (GHG) concentration and the other did not. The first was forced by historical anthropogenic and natural forcing factors and observed SSTs, whereas in the second; historical changes in GHG concentrations and estimates of their effect on SST were omitted. Then, by comparing the results, they inferred how human activity influenced the likelihood and intensities of events. This approach of Pall et al (2011) is called Probabilistic Event Attribution (PEA), and has been applied in some research of recent extreme weather and climate events, such as heat waves and river floods (Otto et al 2012, Christidis and Stott 2014, Shiogama et al 2014, Wolski et al 2014). The aim of this study was to estimate a robust signal of climate change in increasing the severity of a coastal hazard by Typhoon Haiyan, as an example of a worst case scenario in the present climate (e.g. Mori et al 2014, Lin et al 2014). Here we focus on the ‘worst case scenario’ in order to assess a physical upper bound that spawns disasters; we do not take into account the frequency of its occurrence. For this purpose, we performed two types of ensemble simulations of Typhoon Haiyan. The first used observed SSTs, atmospheric properties and GHG concentrations (ALL simulations), and the second ensemble used counterfactual natural external conditions (NAT simulations). Although the idea of our approach is partly based on that of PEA, it should be noted that we do not intend to investigate changes in the frequency of typhoons. To evaluate possible amplification effects 2 due to climate change, we adopted ensemble prediction (Saito et al 2010) and applied dynamical downscaling from a global model (WEP: Weekly Ensemble Prediction system of Japan Meteorological Agency with the equivalent grid point resolution, 60 km) (Sakai 2009, Saito 2011) to a high-resolution Weather Research and Forecasting (WRF) model (Skamarock et al 2008) at the horizontal grid spacing of 1 km that was able to represent the strength of a category 5 TC (Gentry and Lackmann 2010, Kanada et al 2012). With the dynamical downscaled WRF 1 km models, we selected 16 experiments out of all 51 ensemble members. When we discuss the activity of typhoons, frequency of genesis and tracks are important, as well as maximum intensity. To discuss the disaster prevention, we should handle these three metrics jointly. However, for the purpose of disaster mitigation, it is worth discussing at least the change in maximum intensity of the worst typhoon. Because Typhoon Haiyan was the most powerful typhoon to make landfall to date (Mori et al 2014), it is very important to discuss how the worst case event would change under the warming climate environment. The paper has the following structure. The dynamical and ensemble downscaling procedure has been introduced in section 2. Section 3 describes the representativeness of the model typhoon, and also a signal of climate change appeared in the downscaling integrations. In section 4 we present some discussions, and section 5 is the conclusion. 2. Methods The dynamical downscaling procedure adopted in this study is shown schematically in figure 1. The parent model was the Japan Meteorological Agency (JMA)