The temporal evolution and spatial structure of the local easterly wind 'Kiyokawa-dashi' in Japan Part II: Numerical simulations

A numerical simulation of the strong southeasterly (SE) ''Kiyokawa-dashi'' wind in Yamagata, Japan on August30, 2004, is examined and compared to the Coherent Doppler Lidar (CDL) observation. Three-dimensionalnumerical simulations were performed using a non-hydrostatic meso-scale model developed by the Japan MeteorologicalAgency. The sensitivity of the numerical simulation was examined with respect to the resolution of thehorizontal and vertical grid, surface roughness, and ground surface temperature. For the case of a 1-km grid with85 vertical layers including realistic surface settings, the observed characteristics of Kiyokawa-dashi were well reproduced:the strong SE wind (10 m s_1) was extremely low, about 0.1-0.5 km AGL, and the maximum windspeed over 12 m s_1 was observed at 0.1-0.2 km, AGL, under the low-level stable layer. Strong winds appearedat the foot of the final lee slope facing the Shonai Plain and 1 km south of the valley; this result was in goodagreement with the CDL observation. On the basis of the stream line analysis, Kiyokawa-dashi was strongly affectedby the upper SE wind, i.e., upper air on the eastern side of the mountains blew down to the middle andlower layer on the western side. The strongest wind field located 1 km south of the valley was a{currency sign}ected by themountain at the southwestern end of the valley. It was observed from the sensitivity experiments that the heightof the jet-like flow was found to be closely related to the low-level stable layer, and the critical layer was not veryimportant in reinforcing the low-level SE wind. This case of Kiyokawa-dashi was explained on the basis of thehydraulic theory.