A convergent Ekman flow is made by a differential rotation between a rotating annulus and a rigid upper lid, and the instability of a convergent Ekman flow is studied experimentally. It is shown that unstable waves with different characteristics from well known class A and class B waves exist in a convergent Ekman flow. The unstable waves are destabilized in the flow with the very low Reynolds number (about 1.5) and they extend vertically throughout the fluid layer. These characteristics are similar to those of unstable waves found by Arons et al. (1961).
A forecast experiment has been conducted using an initialization scheme including the topography of the Himalayas. Based on an idea that the large scale atmospheric motion is quasi-horizontal and tends to detour the huge mountain, the scheme uses a manual modification of the initial mass field to result in airflow moving around the mountain rather than crossing over it. In this initialization procedure, the break of an isobaric surface in the lower atmosphere due to the lump of the mountain is taken into account. The initial wind field is assumed to be rotational based on the results of the numerical experiment conducted by Kondo (1978) for the divergent barotropic fluid flowing around a circular column. The balance equation is solved to derive the rotational component of wind taking account of the internal boundary condition along the edge of the break. Two numerical time integrations have been carried out up to 48 hours with the initial data of 12 GMT 6 February 1978. One uses an excessively smoothed topography which is currently in operational use at the Japan Meteorological Agency and the other uses the more realistic topographical pattern of the Himalayas. The numerical prediction model adopted is the operational 4-level northern hemispheric primitive equation model (4L-NHM-2). These two forecasts are compared with that of the operational NWP model which uses an initialization scheme without the topography. The forecast using the modified initial wind with the more realistic topography (Experiment 2) yields the best result in that the temperature field in the lower troposphere is the most accurate. The forecast using the modified initial wind with the excessively smoothed topography (Experiment 1) is the second and the operational forecast is the worst in that fictitious warming is conspicuous in the lee side of the mountain. It should be further remarked that the temperature pattern not only near the Himalayas but also far downstream near the Japanese Islands is improved considerably in the present experiment, particularly in Experiment 2.
The operational performance of the JMA regional numerical weather prediction (NWP) model (6L-FLM) is summarized from operational and modelling aspects. The formulation of the computational model is first described. The finite difference scheme of the model is designed along the idea proposed by Okamura (1975). The orography is taken into account only in a limited way. As to the physical processess in the model atmosphere, the following treatments are introduced: (1) parameterization of transfer process in the boundary layer, (2) parameterization of the ensemble effect of convective activities, (3) condensation, (4) orographic effect, (5) sub-grid scale lateral diffusion. No radiative cooling is considered. Concerning the initialization procedure, combination of the balanced wind and the quasi-geostrophic divergent wind is used as the initial wind. Several examples are shown for particular cases like the cyclone development, the formation of a frontal disturbance and the outbreak of a Siberian high as well as the statistical score and the verification of the precipitation forecast. The model has shown acceptable skill for simulating evolution of the synoptic scale motion, but failed to yield a clear cut pattern of the medium scale disturbance. Nevertheless, convection activities due to the moist instability associated with the smaller scale motion are well controlled in the predicted vertical ρ-velocity field and accumulated rainfall distribution. Remaining problems to be solved in near future are discussed. Among other things, the following items particularly attract our attentions, i.e., (1) minimum predictable scab of the model, (2) adequate simulation of the baroclinic instability, (3) prediction in lower latitudes, (4) specification of the lateral boundary condition, (5) fictitious development of the surface anticyclone, (6) treatment of mountain and (7) improvement of the planetary boundary layer and convection parameterization schemes. It may be concluded that, regardless to these unsolved problems, the 6L-FLM is in general well capable of realistically depicting evolution of the synoptic scale motion and yields useful information to be interpreted to the local weather forecasting by the statistical method such as the Model Output Statistics (MOS).
An investigation of the atmospheric wind structure over Antarctica from surface upto about 80km is made using data from about 52 successful M-100 meteorological rocket soundings at Molodezhnaya, Antarctica where the author worked during 1971-73. The Antarctic summer was found to be characterised by light easterly winds which increased in strength with altitude. The South Polar winter was marked by strong westerly winds which had jet speeds of about 90m/s. The winter westerly jet increased in strength as the season progressed. The autumn reversal of winds from easterly to westerly occurred in the first week of February in the stratosphere and in February end in the troposphere. The vernal wind reversal from westerlies to easterlies occurred in the 3rd week of November in the stratosphere. It was found that the reversal first occurred in the upper layers and subsequently in the lower layers showing a downward propagation of the disturbance. The summer to winter shift was a relatively rapid change, while the winter to summer shift was slow and gradual. The winter and early spring period was marked by large wind perturbations. The easterly flow in the upper mesosphere was found to be predominant for about 8-9 months.
In order to clarify the process of cumulus development on the windward side of a mountain range in convectively unstable air mass, an orographic heavy rainfall over Kyushu (southwestern Japan) on August 29-30, 1974 is studied using satellite, radar, raingauge and radiosonde data. The heavy rainfall occurred under the situation that convectively unstable subtropical air mass was transported to a mountain area of 1, 004-1, 500 m high by strong lower tropospheric winds. The influence of the vertical motion of large-scale disturbances (typhoon and frontal system) on the heavy rainfall is supposed to have been very small since they were far from the analyzed domain. The cumulus development began at about 150km off the coast line and maximum precipitation was found over the windward slope of the mountain range. The distribution of rainfall reflected the small-scale topography of the range. The predominant peak of precipitation amount in 24 hours (-170mm) was located at the leeward foot of an isolated mountain (-1, 400m) near the coast. Though the radiosonde data suggested the existence of a shallow convective cloud layer over the windward slope of the mountain range, the raingauge and radar data indicated that the nature of the rainfall was highly convective and that the heaviest rainfall mainly resulted from deep convective clouds (-8km high) triggered by the orographic ascent.
Reconnaissance aircraft data on tropical storms and typhoons in the western North Pacific for the period 1974-78 were examined. Regression equations for maximum surface wind with central pressure and squareroot of the central pressure-depth as independent variables have been obtained and compared.