A zonally averaged coupled atmosphere-ocean model for climate studies is developed. The ocean component is the Wright and Stocker two-dimensional thermohaline circulation (THC) model, and the atmospheric component is a zonally averaged energy-moisture balance model for the atmosphere. Both single- and multi-basin configurations of the model are considered. / The results obtained with this coupled model are compared with those from an ocean-only model that employs mixed boundary conditions. The differences in the steady states of the two models and their linear stability are examined over a wide range of parameters, for both one- and two-basin ocean models. The presence of additional feedbacks between the ocean circulation and the atmosphere and hydrological cycle in the coupled model produces significant differences between the latter and the ocean-only model. The two models generally have different (though similar) equilibria, but, more importantly for the issue of climate change, the variability in the models near similar steady states is quite different. These differences indicate that to perform relevant investigations of long-term climatic variability, a coupled model is necessary. / Next the coupled model with three-ocean basins is applied to last glacial maximum (LGM) conditions. It is found that to achieve realistic results, it is necessary to add a thermodynamic sea ice model into the coupled atmosphere-ocean model. The variability of the LGM conveyor circulation in the coupled ocean-sea ice-atmosphere model is then examined, and the model is subjected to a range of freshwater perturbation experiments. The conveyor state circulation is quite sensitive to the interbasin atmospheric transport of water vapour from the Atlantic to the Pacific. In particular, increasing this transport makes the conveyor state more robust. The LGM model circulation does not exhibit internal century-to-millennial scale variability, nor can the latter be excited by steady freshwater forcing. However, rapid climatic change on a timescale of decades can be generated through transient freshwater forcing of the northern North Atlantic. Perturbations in the ocean circulation are also found to propagate from the Atlantic Ocean to the Pacific Ocean in a few decades. Stochastic, white noise forcing of the model results in a mainly red noise response but also excites a natural mode of THC variability with a timescale of about 150 years.

Three-dimensional radar data for three Florida storms are used with a radiative transfer model to simulate observations at 19 GHz by a nadir pointing, satellite bourne microwave radiometer. Estimates were made of spatial sampling errors due to both horizontal and vertical variability of the precipitation. Calibrated radar data were taken as realistic representations of rainfall fields. / The optimal conversion between microwave brightness temperature and rainfall rate was highly sensitive to the spatial resolution of observations. Retrievals were made from the simulated microwave measurements using rainfall retrieval functions optimized for each resolution and for each storm case. / There is potential for microwave radiometer measurements from the planned TRMM satellite to provide better 'snapshot' estimates than area-threshold VIS/IR methods. Variability of the vertical profile of precipitation did not seriously reduce accuracy. However, it is crucial that calibration of retrieval methods be done with ground truth of the same spatial resolution.

The Acid Deposition and Oxidant Model (ADOM) is an Eulerian long-range transport and deposition model. One of the most highly parametrized and least well established parts of the model is the cloud module that describes cloud formation, pollutant scavenging, aqueous-phase chemistry and wet deposition. As a means of gaining insight into the cloud module, results from simulations with the module are compared with the results of simulations for equivalent conditions with a three-dimensional dynamic cloud chemistry model. / Comparisons of results for a variety of initial conditions show that wet-deposition of sulphate, nitrate and ammonium ions tend to be underpredicted by the cloud module and that the pH of the rain is overpredicted. However, the differences are for the most part not large. Concentrations of hydrogen peroxide deposited at the surface are significantly smaller in the ADOM module than in the cloud chemistry model. The results of the cloud module do seem to be sensitive to the model cloud top height.

Sediment cores in the North Atlantic show evidence of periodic large-scale ice discharge events between 60 kyr and 10 kyr before present. These so-called Heinrich Events (HEs) occurred with a typical period between 5 kyr and 10 kyr. During each HE, a significant amount of ice was discharged from the Laurentide ice sheet into the North Atlantic. This input of freshwater through the melting of icebergs is thought to have strongly reduced the strength of the Atlantic thermohaline circulation. / One theory for HEs is that they are due to an internal oscillation of the ice sheet under constant forcing. This implies that past multi-millennial scale climate variations in the North Atlantic region may have been driven by variations in the Laurentide ice sheet and were not externally forced by climate change or sea level variations. A second theory for HEs requires some variable external forcing on an unstable ice sheet to produce a discharge event. / Using the 3-D ice sheet model of Marshall and Clarke (1997a,b), which includes ice sheet dynamics and thermodynamics, the possibility of internal oscillations within an ice sheet is verified and an analysis of the mechanisms associated with these oscillations is performed. The bed topography and strain heating are found to be critical for the formation and development of fast moving ice streams, which lead to large iceberg calving. In addition, the calving parameterization is found to be an important factor in the ability of the model to periodically discharge large amounts of ice.

The interannual variability of the sea surface temperature (SST) in the western South Atlantic Ocean is explored by applying the composite analysis method to a 40-year period of COADS data (1953-1992). Interactions between the ocean and the overlying atmosphere, as well as oceanic processes, are investigated in order to explain the variability in the SST associated with the two converging western boundary currents in this region: the Brazil and the Malvinas Currents. / The pattern of interannual variability of SST displays a dipole-like structure in the vicinity of the confluence of the two currents. That is, the SST fluctuations in both currents are roughly out-of-phase: warm episodes in one current accompany cold episodes in the other. It is suggested that the variability in the oceanic advection of the currents accounts for part of the observed variability in SST in the confluence region, the dipole pattern being associated with the strength of the temperature gradient in the confluence. / The pattern of variability of the air-sea heat exchange associated with the observed SST fluctuations displays a distribution of ocean-to-atmosphere flux anomalies roughly out-of-phase with those of SST. It is concluded that the SST fluctuations associated with the Brazil and Malvinas Currents are related to the variability in both the air-sea heat exchange and the oceanic advection of the currents. / The warm and cold events observed in the currents are associated with distinct atmospheric circulation patterns. Thus, it is suggested that changes in the atmospheric circulation on interannual timescales induce anomalies in the air-sea heat exchange, which have been found to be partly responsible for the variability in SST of the Brazil and Malvinas Currents.

Rainfall is essential for the design of many hydraulic structures. In particular, rainfall data are usually required as input to watershed models for simulating the runoff process. However, in practice rainfall records, even if they are available, contain only a finite amount of information regarding the historical patterns. A stochastic simulation of the rainfall process is thus needed to generate many sequences of synthetic rainfall series that have similar properties as those of the observed data. A large number of generated rainfall sequences would provide more adequate information for assessing the response of a watershed system. The present study is therefore concerned with finding the best rainfall model that could provide reliable synthetic rainfall series for practical application purposes. More specifically, two popular stochastic rainfall-modeling schemes were assessed for their accuracy and suitability in the description of the rainfall process. / The first model (MCME) represents a combination of the rainfall occurrence process and the distribution of rainfall amounts on wet days. / The second model is based on the Neyman-Scott rectangular pulses (NSRP) stochastic process. (Abstract shortened by UMI.)

In this thesis, an indirect effect of sulfuric acid particles on Arctic climate during the cold season, from November to May, is investigated. Specifically, the research is focused on the alteration of the air mass dehydration rate due to sulfuric acid aerosols. These anthropogenic aerosols differ from natural aerosols by their poor ability to nucleate ice, their strong solubility and their reduced homogeneous freezing temperature when internally mixed with other compounds. / Simulations performed with three column models and analysis of observed data at Alert (1991--94) are used to investigate an indirect effect of these aerosols on climate: the dehydration - greenhouse feedback. Each model covers different levels of physical basis and realism of their simulation. Two scenarios have been compared in the simulations: an acid aerosols scenario and a natural aerosols scenario. / Results show that aerosol acidification leads to a depletion of the ice crystal number concentration and an increase of their mean size. As a result, clear sky precipitation (CSP) occurs more frequently than ice fogs during Arctic haze episodes. This result is in agreement with observations that indicate an increase by more than 50% of the weekly mean CSP frequency, when the proportion of sulfuric acid is greater than the mean observed value of 20%. Consequently, the sedimentation flux of ice crystals and the dehydration rate of the lower troposphere are accelerated. The radiative effect is a weaker atmospheric emissivity in the boundary layer, up to the height corresponding approximately to the top of the ice crystal layer. As a result, the infrared flux reaching the surface and the greenhouse effect are decreased. Simulations performed for the period 1991 to 1994 at Alert show a negative radiative forcing of about --3 W/m2 at the surface between November and May. The net result is a strengthening of the surface-based temperature inversion of 1.3°C, with a surface cooling of 0.4°C and a warming of 0.9°C at 800 hpa. / The indirect effect of the dehydration - greenhouse feedback due to anthropogenic acid aerosols can explain in part the observed strengthening of the surface-based temperature inversion in the Arctic (Kahl et al., 1993).

Several numerical experiments are carried out using the Bryan-Cox Ocean General Circulation Model to investigate the variability of the North Atlantic thermohaline circulation under steady, non-zonal, surface forcing and realistic geometry. To this end the annual mean surface forcing fields were derived from the climatological data sets of Levitus (1982), Hellerman and Rosenstein (1983) and, Schmitt et al. (1989). Further, Arctic freshwater flux, an important part of the hydrological cycle within the North Atlantic Deep Water formation region, is taken into account. / It is found that under present-day climatological surface forcing the system may oscillate at interdecadal period. The mechanism driving the oscillations is linked to changes in both the horizontal and vertical extent of convection in the northern "Labrador Sea". The structure of the surface freshwater flux forcing plays a major role in both the initiation and sustenance of the interdecadal oscillations. Allowing for a freshwater flux into the northern region of the "Labrador Sea" inhibits the interdecadal variability. The oscillations, however, appear, relatively insensitive to Arctic fresh water transport into the "Greenland Sea". / A detailed three-dimensional discussion of the physics behind the interdecadal oscillations is presented.

This is a study of various methods for the analysis of the returned power measured by a laser ceilometer. The calibration constant of the ceilometer is computed, and used to solve for atmospheric optical extinction profiles, in various cases of haze and precipitation. Other methods of analysis, requiring simultaneous measurements with a UHF boundary layer profiler, are also used to find extinction profiles. A comparison between the extinction profiles found using the different analysis techniques is provided. The results indicate that the method of solution that uses the calibration constant of the ceilometer is the most accurate. This work is not a comprehensive study of all the different methods of solving for optical extinction profiles using ceilometer and radar measurements, although it provides solid groundwork for future research.

To better understand the role of mountains in lee cyclogenesis, two such cases which occurred during BASE (Beaufort and Arctic Storms Experiment) are simulated using the Mesoscale Compressible Community model (MC2). Both cases are shown to satisfy criteria for lee cyclogenesis, despite some ambiguity in its definition. The successful simulations reveal that lee cyclogenesis involves several processes: (1) formation of an upper-level short wave, (2) column stretching, (3) enhanced convergence and increased relative vorticity resulting from adiabatic warming, (4) latent heat release and, in one case, increased baroclinicity due to low-level blocking by topography. / The results of sensitivity experiments indicate that removal of topography (latent heat) produces a stronger (weaker) lee cyclone. Topography significantly influences the distribution of precipitation with climatological consequences for areas in the lee. It is found that cyclogenesis can still occur in the absence of mountains in the two cases studied, although mountains modify the cyclogenetic processes.