The generalized Lagrangian-mean (GLM) description extended by Noda (1988) is applied to mean meridional motions in the troposphere. A large number of air particles with different initial positions and initial times are traced for five days using a 30 day data set of velocities obtained from a perpetual January simulation of the MRI's tropospheric general circulation model. Meridional motions of the GLM are obtained by averaging the traced points over the initial times and over the Lagrangian longitudes with the fixed traced time. This GLM initially agrees with a conventional 25-day time averaged, zonal Eulerian mean. The initial hypersurfaces used for the space-time Lagrangian description are chosen as those determined by t0=constant, where t0 is the initial time, so that the averaging over the Lagrangian longitudes is equivalent to a zonal GLM but the averaging over the initial times is a time-ensemble GLM, in which the initial time is an ensemble parameter. The well-known feature of the initial Eulerian mean meridional circulation with a three cell structure (in one hemisphere) disappears within a day in middle and high latitudes in the model troposphere, while Hadley circulations continue to dominate in the tropics. In each hemisphere, the GLM meridional circulation outside the tropics is characterized by the motions converging, along mean isentropes, to the periphery of the subtropical jet and to the storm track zone in the lower troposphere. The doubling time of the GLM density is about two days. The mean meridional velocity field of the GLM circulation is very different from both the Eulerian mean meridional circulation and the residual mean meridional circulation. The GLM circulation in the troposphere is not induced by a single eddy but by numerous eddies. The eddy diffusion coefficients which are useful to the (zonally averaged) two-dimensional Eulerian models are approximated by the maximum values of the correlation coefficients between the velocity at traced points and the displacement from the GLM points. The diffusion along isentropes is dominant outside the tropics, resulting in the antisymmetric diffusion coefficient being proportional to the heat flux along the isentropes. This in turn results in the effective transport circulation being rather well approximated by the residual mean circulation, even in the troposphere. Particles need to be traced for more than five days in the tropics in order to evaluate the eddy diffusion coefficients by the present method.
The transient atmospheric response to midlatitude heating is examined analytically by means of a time dependent, zonally periodic, two level model (with and without surface friction). A Fourier series technique is used to find the solutions corresponding to segmented cosine and segmented sine zonal heating structures, both with cosly meridional dependence. The height at which the thermodynamic equation is applied is found to be crucial in determining the response of the model. When the heating is entered into the model near to the surface (Model 2), unstable modes are prevalent sooner than they would be when the heat forcing is applied at a higher level (Model 1). For both Models 1 and 2, two cases dependent on the meridional wavenumber 1 and independent of the tonal width of the heating structure x0, emerge. For small scale meridional heating structures (case 1; large 1), the response consists of an upper level high and a lower level low which propagate eastward with time. For large scale meridional heating structures (case 2; small 1) the response essentially consists of a zonal wavenumber 3-4 perturbation superimposed on the solution for case 1. An analysis of the baroclinic instability observed in the solutions is presented for each model and case described above.
A numerical calculation of the atmospheric 16-day wave mode under a realistic zonal mean wind distribution and a linearr model simulation of transient waves during January 1979 were performed. The results were compared with the analysis of the FGGEIIIb data set for the same period. It was found that the 16-day wave mode solution using a zonal mean field during January exhibits barotropic features of the normal mode in the troposphere and lower stratosphere. In the upper atmosphere, it bears more resemblance to a forced baroclinic wave mode due to the significant influences of the zonal mean wind and damping. The linear model simulation was performed by using the time-dependent wavenumber 1 component of the geopotential height at the 500mb level of the FGGEIIIb data set as the lower boundary condition. The behavior of the transient wave was found to be similar to that of the 16-day wave mode solution. The results of the numerical simulation and the structure of the 16-day wave mode solution are in good agreement with the analysis of the FGGEIIIb data set. This implies that the transient waves of January 1979 were dominated by the atmospheric 16-day wave mode.
The dynamic normal mode initialization scheme (DNI) proposed by Sugi is applied to a limited-area model. Performance of two integration schemes for DNI is examined, and it is shown that Okamura's time integration scheme is inefficient and a backward implicit scheme is appropriate for DNI applied to the limited-area model. All physical processes can be included in the integration cycle of DNI. Horizontal wind and precipitation rate are compared using a control forecast, a forecast with the nonlinear normal mode initialization (NMI), and with DNI. The deformation of horizontal wind near the mountains is reconstructed by DNI better than by NMI. The area-averaged precipitation is not largely improved by DNI, but the rms error of the precipitation from the control forecast is smaller in DNI than in NMI in the first three hours.
A statistical investigation is made into the characteristics of different circulation regimes during winter over the Pacific and North America and processes of transitions between these regimes. Using 5-day mean 500mb height data during the winters of 40 years (1946-1985), a principal component analysis was performed. Taking account of component stores, persistent circulation regimes of four types (P1+, P1-, P2+ and P2-) were derived. Investigation reveals that P1+ and P1-regimes have no relation to the Southern Oscilation Index (SOI) and are formed and maintained by the stationary wave activity sources along the jet in mid-latitudes. On the other hand, P2+ and P2 regimes have a close relation to the SOI and are consistent with the stationary wave propagation from low latitudes over the western Pacific to North America. On the basis of a composite analysis, it is shown that wave trains appear from over the Atlantic to Eurasia, preceeding the occurrences of circulation regimes. However, features of the wave train preceeding the occurrences of P1+ and P1- are different from those preceeding P2+ and P2-. In addition, two wave train indices (la and lb) are proposed as measures for detecting transitions between these regimes.
Seasonal and interannual changes of cloud amount over the western Pacific were studied using GMS cloud-amount data from April in 1978 to January in 1985, paying attention to the change in the variation with periods of about 30 days. The following results were attained. The variation with period of about 30 days shows a large amplitude in every season in the neighborhood of the tropical cloudy regions and in spring to autumn near large cloud-amount regions in the northern middle latitudes. The large amplitude regions are enlarged and extended to the middle latitudes in the summer Hemisphere, besides the seasonal northward or southward shift of large-amplitude regions, while the amplitude is small even in the subtropical and tropical regions in the winter Hemisphere. The contribution of 30-day period variation to total variation in the subtropics is larger in summer than in winter. The seasonal changes of cloud amount and the amplitude of 30-day period variations show an out-of-phase relationship in the middle latitude region and an in-phase relationship in the subtropics. interannual change in the amplitude of 30-day period variations, as well as cloud amount, is much influenced by the El Nino event. In the El Nino year 1982 to 1983 the amplitude was smaller than in other years in the Northern Hemisphere, while it was larger around (l5°S, 180°). In early 1980 the amplitude was large over a very wide region, in strong contrast to the small amplitude in early 1983. Out-of-phase or in-phase relationships between cloud amount and the amplitude can be seen in their interannual changes, too.
A large-domain, two-dimensional cloud convection model was used for the purpose of examining the natural properties of cloud convections under idealized conditions: the atmosphere received heat and moisture from an underlying uniformly warm water surface, and, at the same time, the atmosphere was cooled at a constant rate. Five experiments were carried out with different sets of microphysical processes. Quasi-steady states naturally attained in the experiments showed different spatial and temporal structures of convection. In the case where rain was not generated, a cellular structure similar to the Benard convection appeared. In the case where rain was generated but did not evaporate, there appeared only a single deep narrow cloud whose lifetime was unrealistically long. In the case with the full set of the standard microphysics, there appeared a 'double-scale' structure. That is, a number of deep convective clouds having a horizontal scale of O(1km) and a lifetime of O(1hour) were generated, and these clouds were spontaneously organized to form several cloud systems having a lifetime longer than 10 hours. Each of the cloud systems induced a rainfall over a region having a width of 30-100km during its life cycle. The principal mechanism for the generation of the double-scale structure in the last case is the formation of a cold air pool at the foot of each deep cloud by rain water evaporation. The formation of cold air limits the lifetime of individual cloud and thus determines the characteristic time scale of the shorter-lived, smaller-scale structure, i.e., individual cloud. At the same time, the cold air pool spreads out in the form of a density current and triggers new clouds at the edges of the pool. In this manner it produces and maintains the longer-lived larger-scale structure, i.e., cloud system. The double-scale structure which was naturally obtained in the case with the full set of microphysics resembles the double-scale structure of the convection over the earth's tropical ocean; i.e., short-lived, small-scale cumulonimbi are organized to form longer-lived, larger-scale clusters of clouds. Furthermore, the role of the cold air pools in the maintenance of the cloud systems is consistent with that observed in the cloud clusters in the atmosphere. These strongly suggest that the double-scale structure in the earth's tropical atmosphere is the natural form of precipitating cloud convection driven by vertical differential heating. The results of the experiments also show that the origin of the double-scale structure of the tropical convection is the existence of cloud microphysical processes.
The evolution process and fine structure of a long-lived MαCS(meso-α-scale convective system) in the subtropical East Asia was studied for 14-15 July 1979 using Geostationary Meteorological Satellite IR observation and radar observation data. This MαCS was generated in the Baiu frontal zone under the influence of a weak short-wave trough and developed into a weak Baiu frontal depression. The MαCS consisted of a few MβCSs (meso-β-scale convective systems). The evolution and propagation of the MαCS were related to both development and displacement of MβCSs. The successive generation of MβCSs over the western coast of Kyushu (western part of Japan) indicated the orographic influence on the formation of convection. The successive development of MβCSs to the west of the preexisting one over Kyushu resulted in the slow moving of the MαCS over Kyushu. The environmental conditions for development of MβCS and the strong response in the upper-level wind field to MβCSs are also discussed.
Part I of the present study has shown that a long lived meso-α-scale convective system (MαCS) during 14-15 July 1979, which was accompanied by a weak Baiu frontal depression consisted of meso-β-scale convective systems (MβCSs) and the development/decay and movement of MβCSs had strong influence on the behavior of the MαCS. In Part II, the fine structure of precipitation was studied using dense surface observation data over western Japan (Kyushu). MβCSs consisted of a few meso-γ-scale convective systems (MγCSs), which had the life-time of∼3 hours and brought precipitation of 1-5×108 tonne. Although MγCSs developed in the eastern, central and trailing portions of the depression, the large part of the precipitation over Kyushu was brought about by MγCSs generated around the trailing portion. These MγCSs were generated over the western coast of Kyushu. Under the condition of strong vertical wind shear in the trailing portion, MγCSs showed remarkable band structures generated over the western coast of Kyushu. The intense precipitation (maximum 100mm/hour) in this area was not associated with strong gusts nor with temperature fall. The orographic influence of Kyushu on the generation of MγCSs and the possible role of symmetric instability in the strong vertical shear flow on the formation of precipitation bands were discussed.
Weekday-weekend differences of air temperature in cities in Japan are examined by the use of data covering a nine-year period. 121 cities are classified into five groups according to their populations. The temperature is found to be lower on Sundays than on weekdays for all the groups, including the group which consists of cities with population of 30000 to 100000. The relation between weekday-Sunday temperature difference and population roughly agrees with what is expected from a simple advective model. On the other hand, relatively large temperature differences are found for cities in Hokkaido in winter.
The surface heat island and the consequent wind system around Tokyo are investigated on the basis of weekday-weekend differences of air temperature and wind vector. By using data covering a nine-year period, temperature and wind fields on weekend days are expressed in the form of departure from those on weekdays. The result reveals the existence of a cool island (i.e. area with lower temperature than on weekdays) centered on the city from Saturday night to Sunday night. The temperature drop at the center of the city is 0.2-0.3°C.In the daytime on Sundays, the cool island covers most of the area within a radius of 50km from the center and is accompanied by divergent wind (i.e. smaller convergence than on weekdays) with order of 0.1ms-1. The divergent wind is detectable under mean wind of 1.5-3ms-1 as well as in near-calm cases. The existence of divergent wind accompanying a cool island on Sundays is detected also for the area around Osaka.
Life cycle effects of cloud clusters on the moisture distribution are studied using GATE data. As clusters develop, moisture supply exceeds precipitation, the excess being transported upward by convective clouds and stored in the midlevel outflow divergence region. The fraction of moisture supply going into storage decreases during growth. Total moisture content and moisture supply attain their maxima when the radar echo intensity is strongest. When the upper level is cloudiest, both the moisture supply and the radar echo intensity weaken. The middle level divergence is replaced by convergence that maintains the upper level stratiform clouds. Precipitation exceeds moisture supply and the moisture content is depleted. Depletion of moisture below 600mb is caused primarily by the drying effect of mesoscale sinking beneath the upper level cloud decks. However, moisture continues to increase in the upper troposphere. As clusters dissipate moisture is depleted throughout the entire troposphere.
Supercooled raindrops were observed for several hours at Yellowknife, N.W.T., Canada on January 10, 1986. The characteristics of supercooled raindrops, such as the size distribution and the precipitation intensity, were examined from replicas obtained at about 10 or 15 minute intervals. The following main results were obtained: 1) The size distribution of supercooled raindrops was similar to the Marshall-Palmer distribution type. 2) Because a layer of air temperature higher than 0°C was not detected at Yellowknife from available idata, the melting of snow crystals is impossible. Accordingly, it seems reasonable to conclude that the supercooled raindrops were formed by the warm rain forming process. 3) The precipitation intensity, the water content, and the total number concentration of supercooled raindrops in general resemble the features of drizzle type rain.