We have analyzed data from the Limb Infrared Monitor of the Stratosphere (LIMS) instrument on board the Nimbus 7 to investigate equatorial Kelvin waves and dynamically wave-induced variations in the tracer fields such as ozone and water vapor in the lower stratosphere. Temperature anomalies of zonal wavenumber one, progressing eastward and having a period of 12-15 days, are clearly observed below 7hPa for April and May 1979. Vertical phase structure with a wavelength of about 15km progresses downward in the zonal flow being about- 16ms-1 averaged from 50 to 10hPa. At the same time, disturbances in the ozone field due to the Kelvin waves are observed: the amplitude is maximum at 30hPa where vertical gradient of zonal mean ozone is largest and the longitudinal structure is in-phase with that of temperature. These features are in good agreement with linear Kelvin wave theory. In the water vapor field, at 50hPa, we have also detected wavenumber one eastward progressing waves being in-phase with temperature though the spatial structure is not as clear as in the ozone field.
It is assumed that low- and high-frequency tropical waves are generated by the united mechanisms consisting of the evaporation-wind feedback (EWF), saturation-triggering (ST), and lateral-triggering mechanisms. Through the EWF mechanism, some waves become unstable owing to evaporation-wind feedback. Through the ST mechanism, other waves are triggered by the intermittent onset of moist convection, upon saturation, to neutralize any pre-existing conditionally unstable stratification. These mechanisms are theoretically interpreted by partitioning moist convective adjustment into two consecutive processes of diagnostic and prognostic adjustments. The two processes respectively restore and maintain convective equilibrium, and are crucial to the ST and EWF mechanisms. As a step to toward a unified theory, EWF instability is examined by the use of a theoretical Kelvin-wave model, which incorporates only the prognostic-adjustment process in the linearized perturbation equations, thereby excluding the ST mechanism. The solutions indicate that wave instability results from the EWF mechanism and not from the wave-CISK mechanism. For a plausible choice of adjustable parameters, one strongly unstable mode corresponds to the observed 40-50-day oscillation, while two weakly unstable modes correspond to the observed 25-30-day and 10-20-day oscillations. These results are compared with those from the numerical experiments conducted in Part I, using a nonlinear model incorporating the original moist convective adjustment scheme. It is then speculated that the 40-50- and 25-30-day modes can strongly grow through the linear and nonlinear EWF mechanisms respectively, while the 10-20-day mode can strongly amplify through the ST mechanism.
The Tropical Rainfall Measuring Mission (TRMM)'s ‘day-1’ combined radar/radiometer algorithm uses a rain-profiling approach which gives as much importance to the measurements of the TRMM satellite's precipitation radar (PR) and the TRMM microwave imager (TMI) as their respective intrinsic ambiguities warrant, which avoids any ad hoc shortcuts that might introduce large biases in the rain estimates, yet which is simple enough to be operational when TRMM is launched in 1997. The algorithm is based on the idea of estimating the rain profile using the radar reflectivities, while constraining this inversion to be consistent with the radiometer-derived estimate of the total attenuation. To perform the data fusion, the problem is expressed in terms of drop-size-distribution variables. Starting with an a priori probability density function (pdf) for these variables, a Bayesian approach is used to condition the pdf successively on the radar and the radiometer measurements. The resulting algorithm is mathematically consistent and physically reasonable. The conditional variances which it calculates serve to quantify the accuracy of its estimates: small variances indicate that the TRMM observations can indeed be explained by the models used; large variances imply that the models are not sufficiently consistent with the measurements.
Free-fall patterns and the variations in the vertical and horizontal velocities of rimed plate-like snow crystals were analysed by means of a stereo-photogrammetric method. The following main results were obtained: 1) Unstable fall patterns were classified into three types: nonrotation, swing, and rotation or spiral. 2) Considering the riming amount of crystals, the distinction among these types was approximately determined by the combination of the nondimensional moment of inertia (stability number) and the Reynolds number, which involves the mean vertical velocity. 3) The standard deviation of variation in vertical velocity was very small (less than about 3% of the mean vertical velocity). On the other hand, the standard deviation of variation in horizontal velocity was considerably large (4 to 15% of the mean vertical velocity). Accordingly, it is likely that the variation of horizontal velocity plays an important role in the random aggregation of rimed plate-like snow crystals having almost the same shape and size.
The Southern Oscillation (SO) in Australasia and the western Pacific is studied for the January/February peak of the boreal winter monsoon. The high/cold SO phase is defined by anomalously high pressure at Tahiti and low pressure at Darwin and cold sea surface temperature (SST) anomalies in the equatorial Pacific. Composite surface analyses of contrasting ten-year ensembles during 1948-92 are complemented by case studies of the upper-air circulation during the high/cold SO year 1989 and the low/warm 1992. During the low/warm as compared to the high/cold SO phase, the Australasian low is weak, but the core of positive pressure departure is located somewhat farther North. Over the equatorial Pacific a weakened zonal pressure gradient entails slower easterlies which lead to anomalously warm surface waters. These maintain lower pressure, and enhanced cloudiness, convection, and rainfall over the central equatorial Pacific, where an increase in convective activity is accompanied by a statistically significant northeastward shift of the South Pacific Convergence Zone, enhanced lower-tropospheric convergence, upward motion, and upper-tropospheric divergence. The divergent component of the wind at 200mb is directed primarily northward into a broad band of upper-tropospheric convergence, which feeds subsidence, lower-tropospheric divergence, and clear skies, and contributes to positive surface pressure departures in the western North Pacific.
The relationships of sea surface temperature (SST) anomalies with atmospheric general circulation and with momentum and heat fluxes are investigated in order to show the dominant structures in the North Pacific in the ENSO cycle (ES: 2-5 years) and decadal (DC: longer than 5 years) time scales, using newly computed monthly 5°×5° (longitude × latitude) gridded datasets for SST, wind stress, and net surface heating fields and National Meteorological Center (NMC) geopotential height data for the 40-year period of 1951-1990 The Pacific/North American (PNA) like pattern that has three centers in the 150°W-170°W longitudinal band in the wintertime 500hPa height field is prominent in the decadal changes of wintertime SST anomalies. Stronger (weaker) westerlies are located south (north) of the normal jet stream position during periods of below-normal (above-normal) SST anomalies in most of the entire extratropical North Pacific. Those SST anomalies are caused by the enhanced (suppressed) heat release along the Kuroshio and its extension, and by stronger (weaker) southward Ekman transport. However the dominant pattern of the wind field on the ES time scale is quite different from that on the DC time scale. The westerly jet in the central North Pacific is stronger than normal, but shifts northward of its mean position in years of ENSO warm episodes. On the other hand, the northwesterly wind is weaker, hence heat release is suppressed in the western North Pacific. A Western Pacific (WP) like pattern then appears in the atmospheric geopotential height field over the ocean with a below-normal center in the Bering Sea and an above-normal center south of Japan. The situation is reversed in the years of ENSO cold episodes. The differences in the spatial patterns of SST anomalies on the DC and ES time scales are associated with differences in the atmospheric circulation, such as the predominance of PNA-like versus WP-like patterns. The corresponding change in the wind system then brings about a change of heat release and Ekman transport that form the SST anomalies. It appears that the distinctions between these two atmospheric patterns are responsible for the differences in the spatial structure of tropical SST anomalies on the DC and ES time scales.
Synoptic characteristics of precipitation over the HEIFE area in Northwestern China are investigated. The results show that rainfall in the HEIFE area is caused by disturbance on the polar frontal troughs, and the rain area covers the whole HEIFE area of 200×100km. However, the amount of rainfall depends on the altitude, e. g. 600mm/year in the Qilian Mountains at 3000mASL and only 100mm/year or less at the bottom flat of the valley called “Hexi Corridor” at around 1300mASL. During summer, the water vapor content is rather large and stratification of the air over the desert shows potential instability. However, because of the high lifting condensation level (LCL) and the free convection level (LFC) over the desert, isolated cumulus convections do not develop, in spite of the large buoyancy induced by high ground temperatures. As a result, rainfall over the HEIFE area occurs when the disturbances on the polar frontal trough move into the area. The total amount of rainfall observed in the lower land of the HEIFE area, however, seems to be significantly reduced due to the evaporation of raindrops within the deep and dry boundary layer. Meanwhile, in winter, the water vapor content is quite small and stability is nearly neutral, which accounts for the low precipitation in this area.
The circulation patterns induced by mobile global heating are investigated by using linearized shallow-water equations on the sphere when the Rayleigh friction rate (α) is different from the Newtonian cooling rate (c). The other parameters involved in this system are Lamb parameter (i. e., the parameter representing the effect of planetary rotation) and the velocity of the heat source. Numerical solutions for a wide range of these parameters are obtained directly from the simultaneous equations which are calculated from the shallow-water equations by expanding variables and the heating distribution in a series of normalized spherical harmonics. When the motion of the heat source is not fast, the circulation patterns obtained strongly depend on α rather than c. The circulation, which is zonally uniform only in the height field, appears only when both α and c are sufficiently small. Further, both the height and divergence fields become zonally uniform when α is sufficiently large, c is small. This pattern never appears for steady heating. The circulation patterns also become zonally uniform for rapidly mobile heating, unless c is large. These results suggest that the assumption of α=c is applicable to a planetary atmosphere of α≠c such as the earth's one, if α is neither extremely smaller nor larger than c. This investigation may be applied to the Venusian lower atmosphere where α is supposed to be considerably larger than c. When the relaxation time of the dynamical damping process (=1/α) is shorter than 100 days, it is suggested that the direct circulation between the day and night sides appears independent of c.
On the afternoon of 8 September 1994, a severe thunderstorm passed over the Gunma and Saitama prefectures, north of Tokyo. It lasted for more than 3 hours and produced a strong gusty wind associated with hail. The wind produced damage to window glass of the Misato Junior High School (MJHS), injuring 2 teachers and 71 students. The damage-producing wind and its parent thunderstorm were analyzed by using data from a damage survey, surface meteorological stations, upper air soundings, satellite images and a conventional radar. It is identified that at least three downbursts occurred in association with the storm. The storm moved toward the east-southeast direction at a speed of about 8ms-1. It was accompanied with a temperature drop of about 10K and a divergent wind, though not initially of damage-producing intensity, below its cloud base from its developing stage. Its cloud-top reached about 15km AGL in its mature stage. The radar reflectivity data showed that it had a marked overhang in the direction of the movement. The width of the region with damage-producing hail below the storm markedly increased twice. The times and locations of the increases coincided with those when and where descents of a reflectivity core were indicated from the radar. Furthermore, time-space-converted wind data at several surface stations showed that divergent wind fields, corresponding to Downbursts A and C, occurred near the times and locations of these events. Downburst A occurred right in the middle of Kodama environmental monitoring station (KD) and Kodama District Fire Station which are only 3km apart, and was clearly delineated from the wind record at these two stations. The temperature drop due to the passage of the storm was largest near the two stations and was more than 11K. The cool air originating from Downburst A later spread over a region of 40km in diameter. After the occurrence of Downburst A, the storm quickly dissipated. Downburst B occurred near KD, where another indication of a divergent wind field and a corresponding further temperature drop were recorded. The portion of the storm where Downburst B was generated passed over MJHS at about the time of the damage-producing wind. Though no observational data exist near MJHS, it is conjectured from the damage survey and these facts that the fourth downburst may have occurred near MJHS. All of the downbursts occurred 6-10km behind the gust front. CAPE of the storm environment decreased from more than 1800m-2s-2 to less than 700m-2s-2 during 3 hours of the passage of the storm. The difference of equivalent potential temperature between the low- and mid-levels similarly decreased from more than 26K to less than 16K.
In order to study the effect of surface friction on the properties of mesoscale convection which constitute the rainbands of tropical cyclones, numerical experiments are performed with an axi-symmetric non-hydrostatic model. As discussed by Yamasaki (1983), when the tangential wind velocity is not very strong (10ms-1 or less), the typical time scale of mesoscale convection is a few hours. When the tangential wind intensifies and frictional inflow becomes very significant, the lifetime of mesoscale convection becomes 11-12 hours. In contrast to the convection with time scale of a few hours, it does not decay owing to an outward shift of low-level convergence associated with the intensification of cold downdraft outflow. Its decay is caused by its inward shift due to strong frictional inflow. In this situation, rainwater falling from the cloud, which tilts outward with height at this stage, acts to prevent the low-level warm moist air from flowing into the cloud. A new mesoscale convection is formed at some distance from the old one. Structure, time evolution and the mechanism of the mesoscale convection with the longer lifetime are discussed in detail.
A statistical study is made of the interannual variability of the northern winter stratospheric circulation in connection with the equatorial quasi-biennial oscillation (QBO) and the solar cycle, by using the 37-year stratospheric dataset of the Freie Universität Berlin and the 31-year NMC global data. During the period 1962/63-1977/78, analyzed first by Holton and Tan (1980, referred to as HT), the polar-night jet is stronger in the W (westerly) than in the E (easterly), as was mentioned by Holton and Tan (1980). However, the difference between the W and the E is barely significant in ‘the latter half period’ (1978/79-1993/94). When the whole period is classified into two groups defined as ‘Min’ and ‘Max’ with respect to the intensity of the 10.7-cm solar flux, it is clearly shown that the late-winter jet in the W is much stronger than in the E (the value of Student's t test exceeds 6) in the Min group, whereas it is no stronger in the Max group. The reason why the result from the HT period resembles that from the Min is probably because the HT period includes two solar minima and one maximum. In early winter, the circulation seems to be correlated with the QBO irrespective of the solar cycle. This difference between early and late winter suggests that the equatorial QBO influences the extratropical circulation in early winter and that the solar cycle modifies it in late winter. An extensive analysis of wave components is also made. The result from the Min is similar to that of the HT period, and the difference between the W and the E is larger than in the HT period. In late winter, the result from the Max is the inverse of the Min result. Finally, the occurrence of major warmings is shown to be related significantly to the QBO and the solar cycle. Such a relationship is clearly illustrated by plotting the occurrence of the major warming onto a 2D phase space of the solar flux and the equatorial wind.
We propose a mechanism that may contribute to observed near-2-day variability in cloudiness observed over the equatorial Pacific Ocean during active phases of the Madden-Julian oscillation (MJO). Our hypothesis is motivated by the following already-established results: 1) Embedded within the MJO are eastward-propagating ‘superclusters’ (Nakazawa, 1988), which are made up of a collection of cloud clusters. 2) Emanating from the superclusters are westward-propagating patches of enhanced cloudiness (Nakazawa, 1988), which were identified as inertial-gravity waves with a spectral peak near two days by Takayabu (1994a). 3) Sub-synoptic scale variability is enhanced on diurnal (and other) time scales during active periods of the MJO (Hendon and Liebmann, 1994). We presume that inertial-gravity waves are initiated by convective activity within the envelope of the supercluster. Thus, since the eastward movement of the supercluster is at about the same speed as the phase speed of the westward-moving inertial gravity wave, diurnal forcing moving eastward with the supercluster would project onto two-day westward-propagating inertial-gravity waves. The essential dynamics of our argument are illustrated with a linear shallow-water model forced by stationary and eastward-propagating mass sources which are diurnally modulated. High-resolution satellite imagery is used to argue our hypothesis. Although observational results are less than conclusive, we believe that the data do suggest that the origin of observed near-2-day variability may at least partly lie in the mechanism we propose.
An equation of the β-gyre growth of a cyclone is derived in a barotropic nondivergent system. The environmental flow is assumed to be an arbitrary quadratic function of coordinates, and therefore non-uniformly straining. The cyclone flow is assumed to consist of a symmetric cyclonic circulation and a flow of β-gyres. Other components of higher wave-number are assumed small and are neglected. The derived equation says the following: The contribution of the β-effect to the β-gyre growth is proportional to βcosα. Here, α is the azimuthal angle of the anticyclonic β-gyre centre around the cyclone centre, measured counterclockwise from the east. The first spatial derivative of the environmental flow affects the β-gyre growth only through the deformation. This contribution is proportional to λcos(2λ-2α). Here, λ and γ are, respectively, the strain rate and azimuthal angle of the dilatation axis of the environmental deformation matrix at the cyclone centre. These results are consistent with the already known results in the case of an environmental flow with a constant shear. The second spatial derivative of the environmental flow affects the β-gyre growth only through the relative vorticity gradient. This contribution is proportional to |∂iΩ|sin(Θ-α). Here ∂iΩ and Θ are respectively the relative vorticity gradient and its azimuthal angle at the cyclone centre. While the β-effect induces the energy transfer between the β-gyres and symmetric cyclonic circulation, the relative vorticity gradient in addition to this induces energy transfer between the β-gyres and environment.