Environ. Res. Lett. 10 (2015) 064011 I Takayabu et al Figure 2. Predicted tracks of Typhoon Haiyan, of the down-scaling done by NHRCM05 from WEP (in broken lines). The thick line indicates the best track data with the dots in the interval of six hours. Figure 3. Time history of Typhoon Haiyan in ALL simulations (16 ensemble experiments) with the 1 km WRF model. (a) Central pressure; (b) maximum wind speed. The westward propagation speed of the typhoon is slightly slower in the WEP simulations compared with the observed best track, so the changes in pressure and wind are shown in relation to longitude. The position of Taclobane is around 125 E. day in the model simulations. This phase lag of the simulated typhoon is not attributable to the WRF model itself, because when Typhoon Haiyan was simulated with the same WRF 3 km/1 km model system with reanalysis data as the parent model, the result showed no phase lag (Mori et al 2014). This result suggests that the phase lag in this ensemble downscaling experiment is attributable to the performance of WEP. Because of this phase lag, however, we compared the evolution of the typhoon in relation to its longitudinal position (figure 3). The comparison indicates that in the ALL simulations both 4 central surface pressure and maximum wind speed agree well with the best-track data, though the intensification of the TC that occurred before its landfall on Leyte and Samar Island was slightly delayed. The minimum central pressure (MCP) of the simulated typhoon was as high as 906 hPa (in experiment m02 (#1001)), which corresponds well to the intensity of the actual TC of 895 hPa as estimated by the Dvorak method (Dvorak 1975). The rapid decline of the typhoon strength after its landfall on Leyte and Samar Island was also reproduced by all of the ensemble simulations.