Environmental atmospheric conditions under which a tornado formed over Hokkaido Island, Japan on 7 Nov. 2006, detected from a supercell reproduced by a cloud-resolving model

Environmental atmospheric conditions under which a tornado was observed in the northeastern part of Hokkaido Island, the north part of the Japan Islands, at 1325JST (JST=UTC+9hours) on 7 November 2006 are examined from observed data, objective analysis data and successful simulation results by a cloud-resolving model with the horizontal resolution of 1 km (1km-CRM). The structures of supercell causing the tornado and the formation factors of the tornado are also investigated using a cloud-resolving model with the horizontal resolution of 250 m (250m-CRM). The initial and boundary conditions of 1km-CRM and 250m-CRM were produced from the predictions of a nonhydrostatic model with the horizontal resolution of 5 km (5km-NHM) and the 1km-CRM, respectively. Those of 5km-NHM are produced from 6-houly available mesoscale objective analysis data of the Japan Meteorological Agency (JMA). All numerical models used in this study are the JMA-NHM, and an additional data-assimilation is never made for the present simulations.

Around 1100JST, a supercell that caused the tornado formed on the eastern side of the Hidaka Mountains, the middle-south part of Hokkaido Island. At that time, a cold front extended from a developed low pressure, located north of Hokkaido Island, to the western side of the Hidaka Mountains. Over the formation area of the supercell, strong southerly winds dominated, because there belonged to a warm sector region of the cold front. Further, since the cold front passed around 1500JST across the observed point of the tornado, the super cell traveled in the warm sector region of the cold front.

The detail vertical wind profile over the formation area of the supercell shows that low-level winds were easterly due to the effect of mountains, while winds rotated clockwise and became stronger with height (i.e., east-southeast winds with about 10 m/s at a 500m-height, north-northwest winds with about 60 m/s at a 8km-height). The easterly component of winds disappeared at a height of about 3 km. The storm relative environmental helicity (SREH), estimated from the 1km-CRM results, is larger than 500 m² s^-2 around the formation area of the supercell. The traveling speed and direction of the super cell were determined from radar observations. The thermodynamical indexes around the formation area of the supercell were estimated as follows; the levels of free convection and neutral buoyancy (LFC and LNB) were several tens meters from the originating level and about 8 km, respectively. The CAPE was smaller than 400 J/kg, and this value is considerably smaller than those in the cases of supercells observed over the middle-western part of United States (~ 2000 J/kg).

The supercell traveled north-northeastward at a speed of about 80 km/h (~ 20 m/s). Ahead of the supercell, a low-pressure area was continuously found in the low layers. The pressure gradient force induced by this low pressure accelerated the easterly component of low-level winds. This modification of environmental wind conditions maintained large SREH ahead of the super cell. It is also necessary that a humid air is continuously supplied into the supercell to maintain convective activities. The humid air flowed into the supercell from the east direction of Hokkaido Island, but not the south direction. In other words, the pre-existence of a humid air ahead of the super cell was significant to maintain convective activities in the supercell. Moreover, foehn phenomena, caused by southerly winds across mountains, affected atmospheric conditions around the area where the tornado was observed. Consequently, the atmosphere below a height of 1 km became warmer and drier (i.e., temperature was raised by 2 ~ 3 K, and relative humidity was 70 %).

The supercell reproduced by the 250m-CRM has a height of about 10 km and a width of 20 ~ 30 km. It well agrees with radar observations (noted that the radar does not have any Doppler observation system), although it is simulated about 10 km east of the observations when it arrived at the observed area of the tornado. The simulated wind field shows that a gust front forms along the convergence zone between gusts produced by downdrafts in the supercell and a southerly humid inflow into the supercell. The gusts are enhanced under the atmospheric conditions affected by foehn phenomena. A sensitive experiment without the evaporation of raindrop below a height of 1.5 km shows that the gusts are weakened by several m/s. Many anti-clockwise circulations are found in the areas with strong updrafts over the gust front. One of them could be stretched upward by strong updrafts to become a tornado.

The structures of the simulated supercell show that southerly winds flow into the southeastern side of the supercell and they produce an area with strong updrafts. These strong updrafts, in which the maximum vorticity larger than 0.01/s is found, disturb the falling of rainwater and graupel to form the structure of vault. In the upper layers over the melting level, most of hydrometeors are not snow but graupel (noted that hail has not be yet introduced as a predicted value in the JMA-NHM). Meanwhile, in the western side of the area with strong updrafts remarkable downdraft areas are produced by the water loading and evaporation effects of raindrops. The above-mentioned analyses indicate that the simulated supercell has characteristic features of classic supercells.

The target supercell caused only one tornado about two hours after its formation. One reason could be the above-mentioned effect of foehn phenomena. Another reason could be the effect of complicated topography around the observed area of the tornado. The observed point of the tornado is located just in the open area of two valleys with a depth of higher than 500 m. The effect of topography should be examined in the trial of the tornado reproduction by using a cloud-resolving model with the further fine resolution.