1983年4月27日東北地方で発生した乾燥強風の解析と数値予報実験

Abstract

Detailed observational study and numerical prediction experiment are made on the extremely strong dry wind event over the northeastern Japan in 27 April 1983, which causedbig forest fires of -9000ha.This extremely strong dry wind took place under the synoptic situation of strong pressure gradient between an extratropical cyclone over the Japan Sea and an anticyclone over the Pacific. The observational study reveals that the abrupt increase of SW-WSW wind with the decrease of relative humidity and temperature rise took place around noon over the lee-side of mountains. The distinct low-level inversion layer observed at 09LST 27 over Sendai (the Pacific coast of the northeastern Japan) was not observed at 21LST 27.The thick dry neutral layer observed at 21LST indicates that the inversion layer was destroyed by convective motions generated over the ground heated by insolation. At the peak period (15LST) of the strong wind, the wind speed at the surface reached to that at 900mb. These observed features suggest that the convective mixing of the momentum is one of the important factors to cause the strong surface wind in 27 April 1983. The prediction experiment of this strong wind event was made by using a 16-layer 64 km-mesh primitive equation model which includes parameterization schemes of various boundary layer physical processes. While a model including both diurnal variation of insolation and mountains (S-M model) simulates the strong dry wind event fairly accurately, a model including only the diurnal variation of insolation (S model) does not simulate. The analysis of the predicted result indicates that the orographically induced downward motion in the S-M model suppresses the formation of cloud against the thermally generated convection and maintains the strong radiative heating and concomitant convective mixing throughout the afternoon. This prediction experiment demonstrates that even the localized strong wind event can be predicted by the high resolution primitive equation model.