Les calcrètes de fissures sont des dépôts de carbonates secondaires que l'on retrouve accolés à l'une des surfaces internes des fissures de la roche calcaire. Si les fissures de la roche où elles se sont développées en sont très riches, elles n'ont jusqu'à maintenant été observées que dans les régions froides et sèches du lac Centrum situé au nord-est du Groënland, ainsi que dans les massifs calcaires du nord du Yukon. Il existe peu d'informations disponibles sur le sujet précis des dépôts de carbonates secondaires localisés dans les fissures de la roche calcaire des régions subarctiques. Par conséquent, l'objectif de la thèse est de déterminer quelles sont les conditions favorables à leur genèse et à leur croissance, ainsi que de dresser un portrait général de l'expression du temps sur leurs micro-structures au cours de leur développement. Pour atteindre cet objectif, une analyse des composantes macroscopiques et microscopiques des calcrètes de fissures a été réalisée sur une série d'échantillons provenant de trois régions calcaires du nord du Yukon: la région de Joe Creek située dans les Monts British, la montagne de Bear Cave ainsi que les calcaires du Dempster Highway situés dans les Monts Ogilvie. (Abstract shortened by UMI.)

Variations in the global hydrological cycle occur at many time and space scales. An important source of evidence about past variations in the hydrological balance are records of lake level fluctuations. The spatial extent and depth of a lake varies in response to changes In the water balance over its surface and catchment. These variations may result from local factors or climate change. However, spatial climatically-induced patterns can be identified when most lakes in a region change in a similar way. A global lake-level database was constructed from three previously developed databases, and used to investigate large-scale patterns present in the lake-level data during the Holocene. Using the geostatistical approach, regional patterns were objectively identified by examining the spatial correlation between sites and gridded using ordinary and indicator kriging at 3,000 year intervals. These maps were produced using two assumptions. First, the data were used under the assumption that climate is continuous in space, and hence, the data were treated as spatial continuous interval data. Then, the data were used under the assumption that the data is spatial categorical and that the classes are mutually exclusive. (Abstract shortened by UMI.)

Seventy-two samples of cavernous concretions collected in 10 caves from the South-West of Quebec, were analyzed in order to (1) establish if there was spatiotemporal bonds between moonmilk, popcorn and rimstones, and (2) determine if moonmilk is a speleothem or simply a state of calcite. The results of our analyses clearly show the presence of transition zones, which translate in a space-time continuum between the sedimentary characteristics of the divers ccncretions. The presence of layers showing moonmilk particularities in popcorn and rimstone samples, and the morphological similitude between moonmilk deposits and common speleothems, suggest that moonmilk is not a speleothem by itself, but an occasional component of speleothems.

A multimedia geochemical data set, collected in SW Cape Breton Island, was spatially and statistically analyzed to enhance the geochemical understanding and to evaluate the mineral potential of the area. The data sets, which include stream water and balsam fir twig surveys, along with the bedrock geology and the topography with drainage, were integrated and spatially analyzed in a geographic information system. Although stream sediment geochemistry is a well-established sample media in mineral exploration, little work on the statistical analysis of stream water has been done to determine the controls of the geochemistry of the water and to aid mineral exploration. Yule's coefficient was used in two attempts, each with different zones of influence, to compare the stream water and the balsam fir twig survey. The weights of evidence technique is a probability based approach for locating mineral potential using the spatial distribution of known mineral occurrences. The explanatory variables used to produce a mineral potential map included the underlying bedrock geology, copper residuals from the stream water geochemical survey and the proximity to structural faults. This approach calculates a weight for each class in the multi-class catchment basin maps for copper. (Abstract shortened by UMI.)

The purpose of this study is to determine the origin of the sediments that lie on the treads of the cryoplanation terraces of the Richardson Mountains in the Northwest Territories, Canada. A series of 15 bore holes were done on the treads of the terraces, at the bottom of which organic material was found and dated by radiocarbon method. Petrographic, granulometric, exoscopic and geochemical analyses were done on the sandstone of the Aklavik Formation of the Bug Creek Group, on the surface sediments and on the core sediments. The objectives of these analyses were to compare the petrology of the local sandstone to the sediments found on the treads of the terraces, to determine the granulometric distribution of the sediments, to determine the scarp's rate of weathering and the chronology of the events that took place. The results show that the sediments are produced by disintegration of the local sandstone of the Aklavik Formation. The weathering of the rock causes an increase in the intergranular porosity and fractures in the grains of quartz. The weathering of the sandstone releases sands, silts and clay. This material constitutes a potential source of loess in glacial times. The accumulation of sediments that lie currently on the treads of the terraces which are derived from the weathering of the sandstone, could have begun to accumulate about 5397 +/- 84 BP or even earlier at about 9420 +/- 90 BP. A weathering rate of the rock scarp was calculated at about 20 mm/1000 years and would seem to indicate that the terraces could have started to develop in the Tertiary period.

RADARSAT-1 images of the northern part of Axel Heiberg Island were acquired to define a selection of preferable beam modes for geological mapping in this cold and arid environment with locally rugged topography. Fine, Standard, and Extended High beam mode images with different resolutions, areas of coverage, look directions, incidence angles, and dates of acquisition were analyzed in this study. Major geological contacts in this part of the Sverdrup Basin and the Franklinian Mobile Belt are well identified in these images. Numerous post-Paleozoic tabular intrusive bodies were also identified. Although shadowing and other geometric effects were accentuated, structural features and lineaments were best highlighted in RADARSAT-1 images with higher incidence angles where less topographic displacement is present. Significant seasonal changes in radar image tone associated with localized melting of the active layer were clearly visible between winter and summer images. RADARSAT-1 beam modes were successfully used to make stereo-pairs. (Abstract shortened by UMI.)

The application of fuzzy logic in a GIS framework is a valuable method to assist in mineral resource assessments (MRA) in areas where data are sparse. This study uses a digital geological map, backed by a digital geological data model, derived from published legends and reports. Together they function as a 'spatial-attribute relational data model' that provides evidence, in the form of derivative maps, to support mineral potential according to deposit model criteria. A knowledge-base is created with fuzzy membership functions linked to the classes of each derivative map that indicate favourability between geological features present in the database with those required by model criteria. A fuzzy-logic-based 'inference net', as implemented in the GIS modelling language, is used to combine spatial evidence to determine mineral resource potential for three mineral deposit sub-types: (1) MVT Pb-Zn, (2) Sedimentary Cu and (3) Sediment-Hosted Sulphides. This method is shown to be valuable for providing an 'audit trial' for the complex decision-making process associated with resource assessment; it provides a means for experimenting and testing various hypotheses and viewpoints associated with mineral deposit models, and mimics some aspect of how geologists determine mineral potential for a region using information provided in geological maps and mineral deposit model literature.

A detailed analyses of Little Dyke Lake basal sediments revealed two environmental disturbances occurred during the late-glacial. These disturbances were correlated to the previously reported Killarney and Younger Dryas climatic oscillations of the Maritimes. Organic accumulation commenced at about 11 500 yrs BP. Macrofossil and pollen evidence indicate that a forest-tundra including abundant juniper and spruce krummholz had developed 300-600 years after deglaciation. Changes in the vegetation composition were followed by an increased inwash of coarser, siltier sediment. With climate amelioration, a spruce woodland grew during the time when paleo-Indians occupied the nearby Debert site. The deposition of clayey and organic rich sediments are associated with the spruce woodland. Plants found in the understory, especially herbs and grasses, and those most apt to grow under cooler, drier and disturbed conditions then became more important in the landscape. This shift in the vegetation cover is accompanied with the sudden replacement of dark clayey sediment by a reddish coarse silt. The termination of the Killarney cooling is reflected by an increase abundance of coarse mineral sediment. However, high Juniperus pollen percentages accompany a shrub Betula maxima when maximum erosion of sands arrives to the lake basin. Changes in the character of the sediment appear to coincide with pollen changes resulting from climatic cooling. (Abstract shortened by UMI.)

This study is more in the nature of "exploratory" than "confirmatory" in its exploration of the temporal and spatial patterns of lightning-caused forest fires in Quebec, for the fifteen year period 1978 to 1992. Economic returns garnered from forest products, as well as species and habitat loss, make lightning-caused forest fires an important area for investigation. Natural forest fires are affected by many factors. For the purposes of this study two classes of variables are used in the investigation of the location and timing of lightning forest fires; these are species and weather data. The species data is gathered at the zone forest level and forest age is used at a mature or young level. Weather variables are limited to temperature, precipitation and thunderstorms. These are selected as a result of agreement in related literature that they are valid variables to work with. Lightning-caused fires are mapped at various temporal scales and these are compared with the species and weather data. The study reveals a dichotomous relationship, where fuel characteristics, or forest species, determine the location of fires (or spatial component), and weather dominates the temporal component. Finally, an attempt was made to make generalisations based on the findings so as to identify issues for further research. Understanding the mechanisms which drive the occurrence and distribution of lightning-caused forest fires, is one of the first steps in creating lightning-caused forest fire models.

The primary objective of this thesis was to study climatic variations that occurred in the Grassland Ecoclimatic Province (Canadian Prairies), between 1921 and 1990, using a geographic information system (GIS). DBASE IV was used to establish the Koppen classification. The annual classification showed important climatic variations within the study period. No station could be associated to only one climatic classification, but to many categories. For example, the climate for Medicine Hat cannot be considered as purely semiarid (31 years classified as BSk) because it also has many years classified as Dvb (12) and Dsb (11). The climatic maps, representing each decade, showed much more information, concerning the spatial variations of the Prairies climate. The 1981-90 period had the most important surface classified as semiarid (BSk), this classification representing 7.73% of the total area covered by this study (579 800 km$\sp2$). This semiarid area, generally situated around Medicine Hat, completely disappeared if not almost between 1931 and 60. The 1931-40 map did not show the presence of very dry conditions reported for the Prairies during the Dust Bowl period (1936-38). Yearly maps created for this important dry spell, agreed with some of the information published by Phillips (1990) and Jones (1991), concerning the areas affected by the Dust Bowl droughts. However, the map for 1936 showed no station classified as BSk, despite the fact that it was very dry that year. The use of the Koppen's classification permitted a climatic analysis based on critical thresholds of temperatures and precipitation, which favor or limit vegetation growth. The use of a GIS, helped to easily process large amounts of data, in order to create climatic maps according to this climatic classification. (Abstract shortened by UMI.)