The temporal variability of North Pacific Ocean surface cyclones is presented, based upon a nine cold-season dataset from the National Meteorological Center. Our results show that: (1) January is the most active month for oceanic cyclone activity; the most active zone of the cyclonic characteristics is at its southernmost location in January; (2) Interannual variability of cyclone activity is pronounced. The interannual variability of cyclone activity is predominantly stronger than the seasonal variability; (3) Analyses of objectively defined regimes, defined on the basis of a 30-day clustering of surface cyclone activity, reveal that regional climatological anomalies of surface cyclone frequency, significant at the 95% level of confidence, can be identified as precursors to the onset of these 30-day circulation regimes.

Our study is concerned with the winter climatology over the Northern Hemisphere for the years 1980 to 1986. / The local E-P flux divergence is divided into barotropic and baroclinic components. The 300 mb level is chosen to represent the upper troposphere. The three dominant convergence centres of the baroclinic component are found on the east coast of Asia, near the eastern Pacific and on the east coast of North America. In most areas the barotropic component of the eddy forcing accelerates the mean flow. Near the dateline the barotropic and baroclinic components of the eddy forcing tend to enhance each other, while on the east coast of North American the barotropic and baroclinic components partially cancel each other. / The three-dimensional structure of some general circulation statistics are discussed, and the results are compared with those of earlier studies. / The possible link between the interannual variation of the seasonal mean flow and the transient eddies forcing is discussed. The results show that no obvious correlation exists in our data set, so that no firm conclusions can be reached.

Non-linear phenomena exhibit extreme variability over a wide range of scales and intensities. In multifractal processes, variability increases algebraically with resolution: as we approach the small scale limit, we develop a highly singular field of diverging and vanishing densities. Even over a finite range of scales, the variability can readily exceed the finite signal-response range capabilities of measuring devices. In face of such extreme behavior, one can no longer consider the problems of observing such processes as "merely" experimental ones. / Detectors will rarely be capable of handling the full dynamic range of intensities, missing either the extreme events or the small input signal. Therefore it is of fundamental importance to understand what multifractals "look like" when observed through a detector having only a finite dynamic range. / Limitations on the observable dynamic range affect intensities at nearby scales, breaking the scale invariance and imposing a limit on the range of scales over which scaling behaviour can be observed. / A simple model of a threshold-type problem, in which a detector has a (finite) minimum detectable signal level, is solved in the multifractal framework. Results include a breaking of the scaling symmetry for scales particularly close to the scale corresponding to the resolution of the detector. The scaling improves as we as we degrade further to lower resolutions. It also improves as we move to higher moment statistics. / Rainfall time series from time scales of 180 years to 5 minutes are analysed, revealing, in particular, a break in the spectral scaling behavior near 2.4 hours. Some of the theoretical results are used to show that this break is likely to be caused by instrumental problems at low signal intensities. The correct scaling behavior is successfully recovered from the low resolution information.

In Doppler radar analysis of the atmosphere the spectrum width is rarely used but it contains information about turbulence. Turbulence is not the only effect that contributes to the broadening of the spectrum. Another effect is the cross-beam wind, which can be dominant in broad-beam radars such as wind profilers. Once this effect is removed, the so-called residual width then serves as an indication of turbulence. A large snowstorm is used in this study for the computation of the residual width. Strong wind and wind shear were observed during the storm. The time-height pattern of residual width bears a close resemblance to that of wind shear. This supports the interpretation of the residual width as being an indication of turbulence induced by wind shear. Energy dissipation rates are also estimated for the snowstorm. In some regions values as large as 800 cm$ rm sp2 s sp{-3}$ are observed. These are large, but within the range of what has been reported by others. The same techniques were applied to the study of clear-air turbulence to relate radar reflectivity with turbulence.

Accurate forecasting of hurricane motion is required to prepare for hurricane landfall. The underpredicted acceleration of Hurricane Opal by the National Centers for Environmental Prediction's (NCEP) operational eta model during landfall and over-land passage is investigated using quasi-geostrophic potential vorticity (QGPV) inversion. We identify and explain model errors in the evironmental steering flow using a systematic procedure including removal of Opal's cyclonic circulation that builds upon previous work. / The eastern ridge associated with the anticyclonic QGPV was persistently forecasted too weak and too distant from Opal, though the forecast of this feature improved at shorter ranges. The anticyclonic contribution to the retrieved flow increased from $-$11% of the observed vector (inhibiting the motion) to 15% (aiding the motion). This increase of 3.8 m s$ sp{-1}$ is alone sufficient to explain the slow forecast. / The eta forecasts likely did not sufficiently resolve the advection downstream of storm outflow and subsequent ridge building. This error propagated through the forecast cycle and prevented phase-locking with the ridge and increased flow. Representation of the ridge improved following advection of the outflow over the more dense U.S. upper air network. / This study emphasizes the need for accurate upper-air analyses and offers a real-time application of QGPV inversion that decomposes the steering flow. The need for further research into the intimate relationship between storm intensity and subsequent storm track is stated.

A steady-state primitive equations model is used to study the structure of stationary planetary waves in the Northern Hemisphere stratosphere. The zonal mean circulation is specified using observed January mean data, as is the wave structure at the lower boundary of 100 hPa. Experiments are performed using data from four years, 1982-84 and 1986. Numerical solutions are found for the structures of zonal wavenumbers 1 to 3 throughout the Northern Hemisphere stratosphere. / In one series of experiments, a linear model is employed, with forcing only by stationary waves at the lower boundary. This model reproduces the gross features of the stratospheric stationary waves, but the wave amplitudes are much less than those observed in nature. / In a second series of experiments, the January means of the transient vorticity and heat flux divergences are calculated using observed data, and applied as an additional forcing term in the model, along with the lower boundary forcing. The model yields significantly larger wave amplitudes, closer to those observed in nature, when forcing by transients is included. / In a third series of experiments, a model which includes the nonlinear interactions among the stationary waves is employed, with forcing only by the waves at the lower boundary. This yields somewhat better results than the linear model, but the change in the wave structure due to the inclusion of nonlinear interactions among the stationary waves is smaller than that due to the inclusion of forcing by transients. / The ability of the model to reproduce the observed features of the monthly mean stratospheric circulation varies greatly, depending on the year studied.

The role of turbulence in the process of collision and coalescence of small cloud droplets is still an outstanding problem in the area of cloud physics. In particular, the growth of droplets in the radius range for 10 to 15 $ mu$m is not well understood. The present research has been motivated by the curiosity whether or not turbulence affects the growth rate of such small drops. / We developed a method to calculate collision rates of small hydrodynamically interacting drops embedded in an external flow field; we call it the flux method. Then, the method was tested for simple cases of laminar flows such as linear shear and a two-dimensional deformation field. The tests were designed not only to validate the method but also to examine the mechanisms associated with the simplified types of external flows which may be equally important for real turbulent flows. / In order to obtain estimates of collision rates for turbulent flows, the flux method was used in conjunction with a probabilistic approach. Numerous simulations of trajectories of two hydrodynamically interacting droplets in a turbulent field were carried out. The ratio of the number of collisions to the total number of simulations gave the probability of collision for different relative positions of the drops. Because the Reynolds number of the flow around droplets (based on the drop radius and terminal velocity) is small, the trajectories were calculated with the help of a model based on the linear Stokes hydrodynamics. Turbulence was modelled in the form of random Fourier modes with both the space and the time spectrum prescribed. Both spectra were characterized by Kolmogorov scaling. The space spectrum was modelled in the inertial and dissipation subranges. On the basis of scale analysis, only small scale time variations were allowed, and, the so called Eulerian-Lagrangrian time spectrum was applied. / The results show that most collision rates increase moderately in a turbulent flow characterized by a rate of energy dissipation of the order of 1, 10, and 100 cm$ sp2$ sec$ sp{-3}.$ The estimated increase in collision efficiencies, however, is not uniform, and a rather complicated relation between the increase in the collision efficiency and the parameters--the drop radii, and the rate of energy dissipation--can be observed.

Clouds and aerosols impact on climate, both independently and through their interaction with one another. Thus, in climate models, it is important to make cloud schemes and in-cloud chemistry as realistic as possible. To this end, the UQAM sub-grid scale cloud scheme was added to LCM11, a one-dimensional version of the Northern Aerosol Regional Climate Model (NARCM). The goal of this project is to assess whether it is necessary to complete parameterization of sub-grid scale processes in LCM11 by implementing sub-grid scale parcel chemistry. / Parallel runs of LCM 11 with and without parcel chemistry were performed under various chemical scenarios. The production of sulphate, the main anthropogenic constituent of tropospheric aerosols, was compared within parallel runs. Sulphate production was not significantly changed by the implementation of parcel chemistry. Thus, for the scenarios assayed in this project, the addition of sub-grid scale parcel chemistry to LCM11 is deemed unnecessary.

Supercooled cloud liquid water (SCW), which causes aircraft icing, can be detected indirectly by observing characteristics of the bright band with a vertically pointing radar. Accretion of supercooled water droplets on precipitating snow (riming) increases the density and fall speed of snow and decreases the intensity of the bright band. For stratiform precipitation with a melting layer, a small peak-to-rain reflectivity difference and a high snow-to-rain velocity ratio indicate high-density snow. A numerical model of the bright band gives a semi-quantitative relationship between snow density and bright band parameters. SCW content is computed theoretically and depends on the gradient of mass flux with height above the melting layer. A large mass flux gradient is an indication of growth by riming.

Montreal's heavy precipitation event of 8--9 November 1996 was noteworthy for its all-time record-breaking 24-hour accumulation of 134.0 mm. We analyse this case through a study of the synoptic-scale and mesoscale states. / Searches for the best analogues to this event are conducted. Fields of the best mass analogue, for 1977, are compared to those of the 1996 event. / It is found that, in the Montreal area, the 1996 event had persistently larger precipitable water content and synoptic-scale ascent values. Averaging the storm-total precipitation values of five randomly-chosen Montreal-area stations yields a value of 74.8 mm for this event. The analogue case, despite its excellent synoptic-scale similarity, was characterised by an average value of only 20.2 mm.