Simultaneous radiosonde observations were made at Funadomari on Rebun Island and at Soya Cape in order to study the kinematic and thermodynamic structure of a cold air mass that penetrates into the Sea of Japan from the Sea of Okhotsk through the Soya Straits when broad band cloud appears on the west coast of Hokkaido. The time interval between sonde observations was ∼ 2 hours. In the period from 26 to 27 Feb. in 1990, we were able to observe the developing stage of the broad band cloud and obtain the kinematic and thermodynamic structure of a meso-β-scale vortex which landed on Rebun Island just prior to the development of the cloud. At altitudes below 1 km, the static stability was higher at Soya Cape than at Rebun Island. It is confirmed that the cold air mass which blew over Rebun Island was closely related to the development and movement of the broad band cloud. The depth of the meso-β-scale vortex defined from the wind field was ∼ 4 km. The level of the cloud top of the vortex, defined as the region where the relative humidity was greater than 70 %, was ∼ 2.5 km at Rebun Island. There was a transient layer whose kinematic and thermodynamic structures were different from those inside the "cloud region" and the environmental air, at least on the leading side of the "cloud region". There appeared a meso-β-scale high temperature region with a strong wind (more than 50 m s-1) just below the tropopause and right above the "cloud region" of the low-level vortex. This upper-level meso-scale disturbance moved with the low-level vortex, and its structure changed in accordance with the developing stage of the low-level vortex.
The present study describes the mesoscale trough that often occurs over the Korean peninsula during the excursion of the Siberian High and investigates the mechanisms of its formation using both observations and a three-dimensional atmospheric numerical model. The mesoscale trough begins to form in the morning hours and shows a well-defined structure through-out the peninsula in the afternoon according to the case observed on 14-15 February 1986, during which the air temperature over the peninsula is higher than normal. It shows a weakening tendency in the early morning hours of the second day, but develops again during the daytime until it dissipates as the high pressure system over China moves eastward. A numerical simulation of this mesoscale trough produces fairly well the major observed features such as the temporal and spatial distribution of sea-level pressure. The results from various numerical experiments indicate that the mesoscale trough observed on 14-15 February 1986 forms as a result of the dynamical and thermal impacts of the mountain ranges on the Korean peninsula and the thermal contrast between the air over the peninsula and the neighboring seas under a relatively warmer winter day condition, with which surface sensible heat flux into the atmosphere during daytime is significantly larger over the peninsula than that over the neighboring seas. The trough over the northern half of the peninsula is mainly formed by the dynamical and thermal effects of the mountains in the northern peninsula, while the trough over the southern half of the peninsula is primarily formed by the thermal effects of the elevated area and the thermal contrast between air over the peninsula and the neighboring seas.
The wind in a limited region can be separated into the external and internal winds, which depend upon the vorticity and divergence only outside and inside the region, respectively. This is the basic property of the wind separation. A given case that the vorticity and divergence are given within the region and the wind is prescribed at the boundary is used to represent a common feature that is often produced in the prediction of a limited-area model and its initialization. If the estimated external wind for a given case satisfies the consistency condition at the boundary, the given case is a consistent case; otherwise, it is an inconsistent case. For an inconsistent case, it is impossible to find a continuous wind that satisfies the wind given at the boundary and yields the vorticity and divergence in agreement with those given within the region. A direct method with two Poisson equations of reconstructing the wind for a given case can only be used if the given case is a consistent case. The implicit normal-mode initialization (Temperton, 1988) is further illustrated from the wind reconstruction and wind separation. The goal of the initialization is to modify the analyzed ageostrophic wind to become the balanced ageostrophic wind. Only the imbalanced ageostrophic wind causes the fast gravity-inertia waves, and needs to be filtered in the initialization. From one computed example, the balanced ageostrophic wind is about 68 % of the ageostrophic wind, thus it is a good approximation to the ageostrophic wind. It is also found that the rotational component of the ageostrophic wind (3.38 ms-1) is larger than that of the divergent component (1.64 ms-1). The balanced wind is the sum of the geostrophic wind and balanced ageostrophic wind, and it is a good approximation to the observed wind. It is proved that the fixed wind components at the boundary cannot be used in the initialization because the initialized case in this situation is an inconsistent case. Based on the basic property of the wind separation, it is very natural to use the fixed external wind as the lateral boundary condition in the initialization. The proposed initialization and boundary condition have been tested and the results show that the meteorological noise can be effectively removed at the start of the forecast.
The effect of breakup of melting snowflakes on the resulting size distribution of raindrops was discussed based on the breakup behavior of snowflakes as they melted in warm kerosene. The maximum diameter, cross-sectional area, and mass of 50 snowflakes were measured as well as the size distribution of the water drops resulting from their melting. The total number of resulting water drops correlated best with the original mass of the snowflake. The averaged number of water drops increased linearly with an increase in mass for masses less than 3.0 mg. Although the mass of each snowflake was similar, the size distribution of the resulting water drops varied greatly. On average, the size of the water drops formed from snowflakes with a mass less than 1.0 mg and greater than 2.0 mg showed exponential and Gaussian distributions in their percentage of original snowflake mass, respectively. Taking into account only the breakup of melting snowflakes, we calculated the size distribution of raindrops formed from snowflakes having a Gunn-Marshall distribution. The slope of the calculated size distribution of raindrops agrees well with that of Marshall-Palmer distribution.
Comparative experiments with real data using hydrostatic and non-hydrostatic models are performed for the torrential rain which occurred on 6 August 1993 in Kagoshima, the southern Kyushu, Japan. A modified version of the three-dimensional anelastic model (Saito, 1994; Kato and Saito, 1995) is used, which is nested with the operational hydrostatic model (the Japan Spectral Model) of the Japan Meteorological Agency. An explicit warm rain process predicting cloud water and rainwater and the moist convective adjustment are individually or conjunctionally employed in the model. The non-hydrostatic simulations of 5 and 10 km horizontal-resolution with a warm rain scheme reproduce continuously heavy rain which corresponds well with the observation. As Kato and Saito (1995) pointed out in a previous comparative experiment of ideal moist convection, the hydrostatic simulation tends to overestimate and overexpand precipitation in comparison with the non-hydrostatic counterpart, and the drag effect of hydrostatic water loading is more significant for convective development than the non-hydrostatic effect. Furthermore, in the 5 km simulations, the hydrostatic approximation greatly overestimates the total precipitation, more than in the simulation of ideal moist convection (Kato and Saito, 1995). From these results, a non-hydrostatic model with hydrostatic water loading is recommended for a high-resolution numerical prediction model.
Interannual variations of the winter stratosphere and troposphere in the Northern Hemisphere are investigated by focusing attention on the long-term variability for the period from 1958 to 1994. Two modes of variability are first extracted by applying an empirical orthogonal function (EOF) analysis to the winter mean 50 hPa height data. The connection of these modes of variability with the troposphere and oceans is then investigated by calculating correlation coefficients between the time coefficients of the EOF and (i) the 500 hPa geopotential height, and (ii) sea surface temperatures (SSTs). The first stratospheric EOF mode alternates between the polar regions and mid-latitudes, and can be traced into the troposphere. The tropospheric circulation pattern associated with this mode is identified as changes in the meridional propagation of stationary planetary waves. Comparison with general circulation model (GCM) results suggests that this mode is essentially produced by atmospheric dynamical processes. Time coefficients of the EOF 1 mode exhibit an increasing trend. This mode appears to be related to the recent tropospheric circulation changes that have occurred over the Euro-Atlantic region and the eastern Asian sectors. The second EOF, on the other hand, exhibits a quite high correlation with SSTs in the equatorial eastern Pacific, and its variability is identified as linked to the El Nino/Southern Oscillation (ENSO) cycle.
In 2-dimensional steady adiabatic unforced systems, both the Bernoulli function B and potential vorticity Q are functions of the stream function ψ and Q is the derivative of B with respect to ψ, i.e., Q = dB/dψ. In this note, this formula is extended to 3-dimensional steady adiabatic systems. Even if there exists dissipation, essentially the same formula holds as far as the inner product between the dissipative force and the velocity is everywhere negative. An application is presented to the problem of lee vorticity.
Concentrations of gaseous H2O2 and O3 have been measured at Ogasawara Hahajima Island located 1000 km south of Tokyo in April and July 1995. In April, when the Ogasawara Islands was covered with the continental air mass, O3 and H2O2 concentrations were high. Although O3 concentration in July was significantly lower than that in April, the mean H2O2 concentration in July was higher than in April. This may be due to a difference in the solar radiation. Hydrogen peroxide concentration was usually higher in daytime than at night. However, the increases of H2O2 concentrations were somtimes observed in the nighttime when the relative humidity was low. The nocturnal decrease of H2O2 concentration was strongly dependent on the relative humidity; in other words, H2O2 concentration is somewhat inversely proportional to the relative humidity. The loss of H2O2 at night is caused by a heterogeneous process in the marine atmosphere. The heterogeneous loss rate was estimated at 0.3 to 6.5 × 10-5 s-1 and was greatly influenced by the relative humidity. This process may significantly affect HOx concentrations in the marine atmosphere.