The North Atlantic Oscillation (NAO) refers to a low-frequency mode of atmospheric variability with two centres of action in the pressure field. The present study examines various atmospheric fields associated with the NAO. The NCEP/NCAR reanalysis data set from 1958--1998 is used. We find that the NAO-positive (NAO-negative) phase induces: (i) anomalously high (low) winter mean sea-level pressure (SLP) in the Azores High region and anomalously low (high) SLP in the Icelandic Low region; (ii) anomalously high (low) winter mean temperatures over northern Eurasia and anomalously low (high) temperatures over northwestern Africa, Greenland, and segments of northeastern North America (i.e., primarily northern Quebec, Labrador, and Baffin Island); and (iii) anomalously high (low) winter mean preapitation centred over the Norwegian Sea/Barents Sea region and anomalously low (high) precipitation over southern Europe, western Greenland, and the above-mentioned region in northeastern North America. Also, it is shown that the variability in storm track position and intensity due to the NAO is largely responsible for the European NAO-induced precipitation anomalies.

Paleo-evidence suggests that the mean state of the atmosphere in the early Holocene resembles the positive phase of the modern North Atlantic/Arctic Oscillation (NAO). To test this theory we use a coupled ice-ocean model to study the effects of atmospheric and ocean forcing from the positive phase of the NAO and altered surface radiation on sea-ice concentration and thickness in the Arctic during the early Holocene. The resulting sea-ice concentrations are compatible with reconstructions based on dinoflagellate cysts assemblages. Sensitivity studies show that local winds dominate the changes in sea-ice concentration from the East Siberian Sea, whereas ocean currents and sea surface and air temperatures all contribute significantly to anomalous sea-ice concentrations in the Barents Sea. Patterns in sea-ice thickness anomalies show similar results but are spread out over larger areas. The net radiation is shown to have little effect on the Arctic sea-ice cover during the early Holocene. / Les données paléocéanographiques suggèrent un régime atmosphérique semblable à la phase positive de l'oscillation de l'Arctique/Atlantique Nord (NAO) moderne durant l'Holocène inférieur (9.5 ka). Pour examiner cette hypothèse, on utilise un modèle dynamique-thermodynamique de la glace de mer, couplé avec un modèle de circulation de l'océan, afin d'analyser les effets atmosphériques et océaniques de la phase positive de la NAO et du bilan radiatif modifié sur la concentration et l'épaisseur de la glace de mer durant l'Holocène inférieur dans l'Arctique. La concentration de la glace de mer simulée est compatible avec les reconstructions de glace de mer derivée des assemblages des kystes de dinoflagellés. Des analyses de sensibilité montrent que les vents locaux dominent les changements de glace dans la mer de Sibérie Orientale mais les courants marins et les températures de surface océaniques et atmosphériques contribuent aux anomalies dans la mer de Barents. Le patron des anomalies d'épaisseur de la glace de mer sont semblables, mais plus étalés dans l'espace. La rayonement net est un facteur négligeable dans les changements de la couverture de la glace de mer durant l'Holocène inférieur.

The McGill University S-Band radar has been upgraded in 1999 to dual-polarization capability. The transmission and reception of horizontally and vertically polarized electromagnetic waves provide new parameters that are sensitive to the hydrometeor shapes and other characteristics in the radar resolution volume. As a result, the retrieval of microphysical properties of weather events can be improved. / In this work, we studied the reliability of the polarization data obtained by the fast scanning radar. Measurements from different precipitation systems are analyzed to study how this new information can be exploited for operational purposes. The work done shows a great potential for particle typing and microphysical retrievals. However, the possible use of orthogonal polarization for measuring precipitation remains ambiguous.

The objective of this study is to detect zones of supercooled drizzle using data from a vertically pointing radar (VPR). This radar was used in the AIRS (Alliance Icing Research Study) experiment held at Mirabel Airport from November 1999 to February 2000. / For cases where conditions favor the formation of drizzle drops with a diameter sufficiently large, a second mode with a distinct speed and reflectivity can be separated from that received from snow. That the second mode corresponds to supercooled drizzle is verified by comparing reflectivities and speeds with aircraft data taken during AIRS. / For cases without bimodality, regions of icing may be localized by a computation of the mass flux gradient. Accretion of supercooled water on snow (riming) increases both its density and fall speed. Thus, a sudden increase in snow fall speed indicates the presence of supercooled water. / A conceptual model describing necessary conditions for the formation of supercooled drizzle is proposed. This model is based on the analysis of a synoptic situation present for two days of intense icing.

This thesis investigates the interaction between slantwise convection and large scale marine cyclogenesis. The major aspects of the work are: (1) to understand the physics of convective-cyclogenetic interaction using a two layer primitive equation model with a parameterization for slantwise convection, (2) to perform numerous sensitivity experiments with a three layer model to test the robustness of the proposed interaction, and (3) to quantify the contribution of convection to marine cyclogenesis using a high resolution model simulation. / The major conclusions are: (a) The two and three layer models produce realistic explosive cyclones with a simple parameterization for slantwise convection. (b) The explosive deepening coincides with the formation of the "bent-back" warm front. (c) The frontogenetic calculations and quasi-Lagrangian heat and vorticity budget calculations performed with the two layer model indicate the following sequence of events: Convection leads to rapid warm frontogenesis through "tilting" and the formation of the bent-back warm front. The sudden surge of cold advection in the regions of the bent-back warm front then forces the upper level heights over the cyclone center to fall in a rather dramatic way. Increased upper level vorticity advection interacts with the low-level system leading to explosive cyclogenesis. (d) The sensitivity study performed with the three layer model indicates that the formation of the bent-back warm front and the proposed interaction are quite robust for a range of parameters. However, there are quantitative differences when the low-level cloud mass fraction, surface drag force, the air-sea thermodynamic disequilibrium, vertical shear and static stability are varied. Shallow convection is insignificant in affecting the dynamics of cyclogenesis. (e) The high resolution simulation reveals that the descending branch of the slantwise convection may be responsible for the formation of the dry slot over the cyclone center in the early stages. The stratification in both the warm and the bent-back warm fronts exhibits slantwise neutrality after the explosively deepening period. (f) The PV (Potential Vorticity) inversion technique indicates that convection contributes to marine cyclogenesis. The inverted wind fields show a small scale cyclonic vortex and the presence of cold advection in the bent-back warm frontal region. (g) The PV diagnostics further reveal that the cold advection in the bent-back warm front decreases the surface potential temperature anoma

By combining numerical simulations with different diagnostic tools, this thesis examines the various aspects of two explosively deepening cyclones--the superstorm of March 12-14 1993 and a storm that occurred during the Intensive Observation Period 14 (IOP-14) of the Canadian Atlantic Storm Program (CASP). Using conventional observations, the general aspects of the storms are documented and the dynamical and physical mechanisms are discussed. Then the life cycles are simulated with the Canadian Regional Finite-Element model. To improve the model initial conditions, a methodology is proposed on the basis of potential vorticity thinking, and is tested to be successful in the simulation of the March 1993 superstorm. Using the successful simulations as control runs, a series of numerical sensitivity experiments are conducted to study the impacts of model physics on the development of the two rapidly deepening cyclones. / The deepening mechanisms of both storms are examined within the context of PV thinking, i.e., using piecewise potential vorticity inversion diagnostics. In both cases, the upper-level PV anomalies contribute the most to the surface cyclone, followed by the lower-level thermal anomalies and diabatic heating related moist PV anomaly. It is found that a favorable phase tilt between the upper- and lower-level PV anomalies allows a mutual interaction between them, in which the circulations associated with the upper-level anomalies enhance the lower-level anomalies, which in turn feedback positively into the upper-level PV anomalies. In addition to the vertical interactions, there also exist lateral interactions between the upper-level PV anomalies for the March 1993 superstorm. The upper-level PV features (troughs) are isolated with the piecewise PV inversion. By removing or changing the intensity of the trough in the initial conditions, the RFE model is integrated to examine the impact of each trough and its interaction with the other trough on the superstorm development.

A boundary-layer wind profiler has been operated continuously in downtown Montreal as a collaborative project of McGill University and the NOAA Aeronomy Laboratory. The records contain hundreds of hours of data on the melting layer of snow, consisting of Doppler spectra in the vertical profiler beam with a typical height resolution of 100 m and a time resolution of about 1 minute. A particular case study day was selected for analysis of precipitation processes associated with the melting of snow to become rain. We have examined the Doppler spectra just above and below the melting layer for evidence of aggregation or breakup during melting. Apart from the complete spectrum, we also show that the product of the mean Doppler velocity and the reflectivity factor in snow should be equal to that in the rain if aggregation and breakup are negligible. Both approaches indicate that there are times when snow merely melts to become rain, each snowflake forming one drop, but other times when aggregation or breakup are evidently important. It appears breakup is associated with regions of higher reflectivity.

The study of radiative transfer in multifractal clouds is of great interest, an important application being to Global Climate Models. In this work we develop a formalism analogous to the multifractal singularity formalism for understanding photon scattering statistics in radiative transfer in multifractals, and test the results numerically on lognormal multifractals. Although the results are only exactly valid in the thick cloud limit, the approximation is found to be quite accurate down to optical thickness of $ tau approx1$-10, so the results may be widely applicable. Furthermore we show the possibility of "renormalizing" the multifractal by replacing it with a near equivalent homogeneous medium but with a "renormalized" optical thickness $ tau sp{1/(1+C sb1)}$ where C$ sb1$ is the codimension of the mean singularity of the cloud. We argue that this approximation is likely to continue to be valid for multiple scattering, and is also compatible with recent results for diffusion on multifractals. Finally we analyze cloud liquid water content data and estimate the universal multifractal indices. We find that the scaling is respected over the whole range 5m-330km and that the cloud can in fact be reasonably described by a lognormal multifractal.

It is now well established that aerosols, by scattering and absorbing solar radiation, affect the terrestrial heat engine. Depending on the net relfectivity and absorptivity of the haze column and on the surface albedo, the net effect could be a cooling or a warming. But better understanding of the influence of aerosols is handicapped by lack of data and by the fact that their composition and distribution is highly variable in space and time. For the same reasons, the incorporation of aerosols in climate models is usually rudimentary. / We have in-situ measurements of the vertical profile and size distribution of aerosols ($>$0.09 $ mu$m in radius) taken over several places in eastern Canada at different times of the year. These data have been incorporated in a simple multi-layer radiative transfer model to estimate the effects of aerosols on the local radiation budget. A broad range of refractive indexes has been used and the increase of the particles size with the relative humidity is included.

Analysis of high resolution observational data gathered during a mesoscale field experiment in central Alberta (LIMEX-85) has led to a conceptual model of severe convective outbreaks in Alberta. It is proposed that most severe convective events result when upper-level cooling, associated with an advancing, synoptic-scale trough, occurs in phase with strong surface heating over the Alberta foothills. The deep destabilization over the elevated topography acts to amplify the mountain-plain circulation and to generate mesoscale upslope moisture transport. Concurrently, the surface synoptic pressure gradient gives rise to east-northeasterly winds which advect the moisture-rich air of the eastern plains into the lower-branch of the mountain-plain circulation. In this manner, the plains moisture is permitted to reach the convectively active foothills through underrunning of the capping lid. The end product of the synoptic-mesoscale interactions is the initiation of well-organized, severe convective storms which move eastward with the westerly component of the mid-tropospheric winds. A statistical analysis based on archived hail data provides additional evidence for the key synoptic-scale features of the conceptual model.