<p>Respiratory diseases have been and still are one of the major challenges in health research. While many studies have demonstrated that respiratory health varies by socio-economic status and environmental exposure, consensus among the scientific community is still lacking regarding the sufficiency of evidence to infer a causal relationship therefore highlighting the need for improved assessment of both exposure and outcome. This thesis addresses these issues through a reanalysis of the Hamilton Children's Cohort (HCC) study, undertaken in the 1980's in Hamilton, Ontario, Canada. The HCC study was the first one that provided important evidence linking adverse respiratory health with smoking, hospitalization in infancy and air pollution. However, given the limited development of spatial analysis and GIS at the time of study, the spatial dimensions of respiratory diseases and air pollution were not fully explored. The objective of this thesis is to re-analyse the HCC data in order to investigate the spatial variation of air pollution and children's adverse respiratory health as well as the relative importance of other characteristics that may determine the burden of respiratory incidences. Children's exposure to air pollution is first re-estimated using kriging and land-use regression. Based on exposure assessment analysis, compared to kriging, the land-use regression model performs better in capturing local variation of particulate pollution (TSP). The results of the land-use regression model are then used in multivariate linear and logistic regression analysis for the assessment of children's respiratory health. Findings of the multivariate analysis verify and strengthen the results of the HCC study indicating that in addition to their findings, children's respiratory health is associated with chest illness in</p> <p>siblings, use of gas for cooking, low income and household crowding. In conclusion, this thesis demonstrates the usefulness of spatial analysis and GIS in assessing children's exposure to air pollution and also strengthens the HCC study findings revealing statistically stronger associations between children's respiratory health and a number of covariates. In general, the results hold promise and in combination with space-time analysis may lead to the development of advanced exposure assessment models in order to improve our understanding of the potential determinants of respiratory health.</p> / Master of Science (MSc)

<p>This study presents statistical land use regression models to characterize the spatial distribution of two air pollutants: nitrogen dioxide (NO₂) and sulphur dioxide (SO₂). While nitrogen dioxide (NO₂) is considered to be a marker of vehicular pollution, sulphur dioxide (SO₂) is a major industrial emission. The research uses mobile field- monitored pollution measurement data at over 2000 locations across the Hamilton Census Metropolitan area (CMA) during November 2005-March 2006. Various land use characteristics, transportation attributes and meteorological characteristics are used in the model to explain the spatial variations in the concentration of the pollutants. A key contribution of the paper is that it explicitly includes wind directions (i.e. north-east, north-west, south-west and south-east) into the models.</p> <p>The results reveal that proximity to the major roads and highways, total length of roads nearby, traffic volume, nearby park and commercial areas, distance to the lakeshore industries and wind speed have major impact on NO₂ concentration. While proximity to the roads and highways and higher total length of roads nearby increases the concentration of NO₂, higher amount of nearby park areas and wind speed decreases the level of concentration of the pollutant. For north-east direction distance to major roads and highways, lakeshore industries, and wind speed has significant impact on the concentration level. While for the north-west direction, commercial area within 350 meter, distance to QEW and Centennial Parkway North intersection contribute significantly to the concentration of NO₂, total length of the roads within 50 meter from the observation points and distance to the railway are significantly associated with the concentration level in case of south-east wind direction model. Lastly, the significance of a dummy variable indicates that the upper city has lower concentration of NO₂ compared to the lower city according to the south-west wind direction model.</p> <p>For the SO₂ models, the results suggest that proximity to downwind locations from the industries, proximity to commercial and park areas, and proximity to the Bay Front Area play a role to the concentration of the pollutant. While downwind location with closeness to the industries increases the level of concentration of SO_2,higher wind speed and proximity to the Bay Front Area reduces the level of pollution in general.</p> <p>All models for NO₂ and SO₂ provide reasonable model fit in terms of coefficient of determination (R²) of the estimated land use regression models for different major wind directions. Almost all parameters are statistically significant at least at the 5% level of significance. The estimated models are used to generate pollution surfaces for the Hamilton CMA. The resulting spatial distribution of pollution concentration can be useful for informing policy.</p> / Master of Arts (MA)

<p>Population synthesis techniques are used to create lists of population members, where each member is endowed with attributes of interest. Aggregating these attributes across the synthetic members yields distributions which conform to known aggregate tabulations. Population synthesis is used when disaggregate population information is desired, and only aggregate and sample data is available. In this work, population synthesis techniques are discussed and compared, using a small, complete test population of firms. Given the results of these comparisons, populations of individuals and households are synthesized for the City of Hamilton, Ontario. These populations are then linked together to form a hierarchically ordered 'Comprehensive' population, where individuals belong to households, which in turn occupy dwellings over space. The synthesized comprehensive population is created specifically to meet the data input needs of URM-Microsim, a state of the art residential mobility microsimulation model. Originally calibrated for use in Europe, URM-Microsim is adapted for use in the Canadian context via the aforementioned comprehensive population. Some background on residential mobility modelling, as well as the URM-Microsim model is also presented.</p> / Master of Arts (MA)

<p>Shopping is an essential aspect of our day to day lives, as it is necessary in order to fulfill numerous needs and desires. Evidence suggests that the proportion of travel for retail and service is increasing, such that traffic congestion can no longer simply be attributed to work related travel. The restructuring of the commercial sector and automobility has increased competition between outlets, such that stores are now competing against merchants across a large spatial region. As a consequence, many consumers are required to travel long durations to accomplish shopping activities.</p> <p>Discrete/continuous models can determine the likelihood that an individual will engage in a shopping activity, followed by the analysis of the travel duration. The models can overcome the sample-selectivity bias, since shopping is only accomplished by a subsample. Traditionally, the models have been estimated disjointly, however, they are increasingly being estimated jointly.</p> <p>Using the General Social Survey, the objectives of this study are twofold. First,<br />the study aims to analyze the shopping frequency and travel duration of Canadians by<br />comparing the one day behaviour of residents of non-Census Metropolitan Areas (non-CMA) and Census Metropolitan Areas (CMA) for 1998 and 2005. Second, it will<br />investigate the potential of a newly developed discrete/continuous model for the joint<br />analysis of shopping travel behaviour. The results of the analysis suggest that shopping<br />travel behaviour is similar regardless of region, and that the joint model provided<br />consistent and realistic estimates.</p> / Master of Arts (MA)

<p>Thecamoebians (testate amoebae) and foraminifera have never been systematically investigated in a phreatic cave system. Foraminifera (marine to brackish water protists) and thecamoebians (brackish to freshwater protists) are cosmopolitan and have been studied in virtually all environments on Earth. Although foraminifera-bearing sediment has been observed from vadose cave environments in the last 20 years, there is no mention of their ecology - or mere existence - in flooded cave passages in any micropaleontology or microbiology textbooks. Only one published manuscript briefly mentions foraminifera in preliminary test samples from phreatic caves in Bermuda. The unprecedented ability of these organisms to colonize any environment on Earth led to the hypothesis that they are also exploiting global phreatic caves.</p> <p>The karst landscape of the Yucatan Peninsula, Mexico, is an ideal setting for the formation of anastomosing, anchialine, phreatic cave systems. The stratified aquifer contains passages in both the superior freshwater lens (thecamoebian habitat} and marine water (foraminifera habitat) intruding below, creating an ideal location to test the hypothesis. Several important contributions have resulted from investigating surface samples and cores in this region. (1) Using thirty-three surface samples from three different cenotes, thecamoebians and foraminifera were found living in the sinkhole (cenote) environment, which are the physical entrance into the caves. An ecological boundary of -3.5 ppt was determined for thecamoebians, with Centropyxis constricta "aerophilia" determined as the most euryhaline thecamoebian taxa. (2) Foraminifera and thecamoebians have been colonizing Carwash Cave since the Sangamon Interglacial (MIS 5e), based on seventy-five surface samples collected throughout the system. The recovered taxa have responded to changes in cave environment similarly to other coastal systems, thereby indicating their potential as paleoenvironmental proxies in phreatic caves. (3) In conjunction with stable isotopes (δ¹³C_org and δ¹⁵N_org), thecamoebians and foraminifera were successful recorders of mid-late Holocene paleoenvironmental evolution in Carwash Cave. An important salinity transition in the freshwater lens (from >2 ppt to 1.5 ppt) of the aquifer at ~2760 Cal yrs BP is coincident with changes in the local Maya populations, indicating a possible environmental linkage between ancient freshwater resources and Maya cultural genesis. This thesis demonstrates unequivocally that thecamoebians and foraminifera exist in phreatic cave systems and hold wider potential as paleoenvironmental proxies in other global phreatic caves.</p> / Master of Science (MSc)

<p>This study was undertaken to further understand the interrelationships between the endobenthos and sedimentation. Quantification of ichnospecies abundance and diversity are used to interpret their significanceand the ecologicalconditions two sedimentologically different incised valleyfills. This study is based on cores from two incised valley-fills from the Cretaceous Western Interior Seaway (Alberta). The first, the Waskahigan Channel (Shingle Dl) in the Dunvegan Allofonnation (Cenomanian) interpreted as a migrating delta, contains mud and fine-grained sand. The second, the South Willesden Green Channel (Allomember C) in the Viking Allofonnation (Albian) interpreted as a straight incised valley-fill, contains coarser, reworked marine sand and conglomerates. These differences provide a comparative test for the innovative techniques used herein. TInsresearch provided a tool to interpret changes in sea level and the depositional environment by subtle changes observed in the abundance and diversity of ichnofossil communities and ichnofacies. ill the Waskahigan Bottleneck Channel (7-13-64-24W5), estuarine sand flats are distinguished by the such changes. The assemblages' abundance and diversity increased, then decreased and increased again as the channel infilled. In the South Willesden Green Channel, core IO-35-40-7W5 is interpreted as estuarine bay fill deposit. Without quantitative ichnology, the five successive fills recognized herein would be considered one deposit. In both channels, shells and shell fragments are absent as a result of physical weathering and chemical dissolution. This study contributes methods of ichnological quantification and comparison in subsurface core. TI,e subtle details reveal more information about the ecological conditions during the infilling of these Cretaceous depositional environments.</p> / Master of Science (MS)

<p>The Lower Cretaceous Viking Formation is a complex stratigraphic unit containing a variety of geographically separated sand bodies of varying orientation, thickness and lithology. Many of these sand bodies are prolific hydrocarbon reservoirs which have been extensively drilled by industry. Despite this fact, their origins and interelationships are largely unknown. Establishment of a Viking Allostratigraphy in the Willesden Green area (1) allows the recognition of distinct allomembers based on bounding discontinuities and facies associations, and (2) determines the exact stratigraphic relationships and depositional histories of the Viking oil and gas fields as defined by basin-wide sea level fluctuations. Detailed core and log correlations indicate that in the Willesden Green area, the Viking Formation is made up of 4 distinct packages of sediment separated by 3 major stratigraphic breaks. The lower most package is regional in extent, and consists of three gradational coarsening-upward cyclic offshore mudstone/siltstone sequences (Members A and B). At Willesden Green, Member B is erosively incised by a major unconformable scour surface (VE2 - Viking Erosion surface 2) filled with conglomerate-rich channel/estuarine sediments (Member C). The VE2 incision was carved during an initial Viking lowstand, and infilled during a temporary stillstand in the ensuing transgression. The top of Member C is erosively truncated by a regionally tracable pebble-mantled ravinement surface (VE3). Member D, a sandying-upward sequence of storm-dominated lower shoreface to transitional offshore sediments, overlies VE3, and records a second major Viking progradational event. This unit thickens southward, and can be traced laterally into correlative upper shoreface and nonmarine sediments at Caroline and Harmattan. The top of Member D is bevelled by a second basin-wide ravinement surface (VE4) which rises in a step-like fashion towards the south. This surface is mantled by a thin veneer of pebbles which locally accumulate to form the elongate shaleencased, conglomeratic shelf sand bodies of Member E. Member E gradationally passes upward into the silty shales of the Colorado Group and records a final transgressive flooding of the Viking basin.</p> / Master of Science (MS)

published_or_final_version / abstract / Earth Sciences / Doctoral / Doctor of Philosophy

<p>BACKGROUND: Canada's aging, heterogeneous population presents a challenge with respect to maintaining a sufficient national blood supply. Patterns of donation and correlates of donor data will be identified through analysis of the data.</p> <p>DATA: Geo-coded blood donor and donor clinic data are provided by Canadian Blood Services. Blood donor data is provided for the fiscal year 2006-2007 indicating the total number of donors for each Canadian postal code, excluding the province of Quebec. Potential con-elates of blood donation are selected based on social and economic characteristics, as well as descriptors of city size and geographical location in the urban hierarchy measures of accessibility, and capacity of donor clinics.</p> <p>METHODS: Data is aggregated to n = 3,746 census tracts in 40 Census Metropolitan Areas (CMA) across the country and then to n = 992 census tracts for the Toronto CMA. The number of donors per population in each Canadian census tract is regressed against the set ofpotential donation con-elates. For the Toronto CMA model, the donor count in each census tract is regressed against similar potential correlates.</p> <p>RESULTS:A number of factors are found to influence blood donation in Canada including the propOliion of younger residents, English ability, proportion of people with immigrant status, higher education, and a population-based measure of accessibility. These findings are confirmed when a model involving the city of Toronto is created. The Toronto model achieves similar correlates as the national model with the addition of variables that are unique to the city of Toronto. These unique attributes involve travel, employment, and gender.</p> <p>CONCLUSION: While a number of correlates of blood donation are observed across Canada, important contextual effects across metropolitan areas are highlighted. These contextual effects are supported by the uniqueness ofthe Toronto model's secondary correlates. The thesis concludes by summarizing what these findings contribute to the field of blood donation in Canada. Further mention is also given regarding the role of spatial filters as a tool in regression analysis.</p> / Master of Arts (MA)

<p>The objective of this thesis is to validate the results of the Integrated Model of Urban LAnd use and Transportation for Environmental analysis (IMULATE), as well as the Integrated Model for Population Aging Consequences on Transportation (IMPACT) in the Census Metropolitan Area (CMA) of Hamilton, Ontario. The land use/demographic modules of these two models are validated using observed data from Statistics Canada, while data from the Transportation Tomorrow Survey (TTS) and the City of Hamilton are used to validate the transportation modules. Statistical, graphical and GIS visualization techniques are incorporated into this validation.</p> <p>This thesis illustrates some sub-modules in IMULATE and IMPACT can work very well, while the predictive ability of others is not as good. IMULATE considers more factors to simulate land use development. It generates accurate simulations of household dynamic using observed data as exogenous input. We used the "fmal demands" for economic sectors as the exogenous input to estimate the employment's distribution. After recalculating the "fmal demands" in its employment location model, the generated employment is also found to be close to the observed value. However, we found that its transportation module was not able to produce accurate predictions of inter-zonal trips and traffic flows over the Hamilton's road network.</p> <p>IMPACT can predict the growth of population by gender and age with good accuracy. The simulation results for males are better than females. The inter-zonal trips generated by IMPACT are found to be much closer to the observed value than the inter-zonal trips generated by IMULATE. However, we found that the simulated trips have lower dispersion across the city than normally observed.</p> / Master of Arts (MA)