Abstract:
Over a two year period, 1976 - 1977, 253 itineraries of 86.3 km. were randomly conducted under all weather conditions along a well defined urban - rural traverse, encompassing all major land usage, upwind and downwind of Winnipeg's core area. The automobile mounted, dew point hygrometer provided dew point and temperature data which were then standardized to the initial traverse commencement time. Climatological parameters, such as mixing ratio, saturation deficits, and absolute humidities were then calculated. Ward's error sum of squares hierarchic fusion algorithm was utilized to achieve cluster traverse organization. Linearly interpolated surface to 500 mb. radiosonde data aided in detecting air mass groupings. To faciliate urban - rural climatological comparisons with the established literature, the primary air masses chosen for investigated were the bench mark air masses, continental Arctic and continental maritime Arctic. The general postulation throughout the thesis was that no hypothesis would be formulated regarding the traverse data sample distriubtions, the spatial distributions of the climatological parameters under investigation, or the temporal variability of these climatic parameters without statistical supporting evidence. As background information regarding the bench air mass data was absent, non - parametric statistical techniques were used. The equivalent parametric statistics were also utilized on the same data inorder to measure outcome variability. The procedure generated unique conclusions which often were contrary to those found in the literature field dealing with urban - rural humidity and temperature islands. In the case of Winnipeg, it was statistically proven that a mixing ratio island does exist under cA - cmA air mass conditions and that the spatial zones are significant even if the climatological differences were less than 1.0g/kg. The spatial forms do not necessarily parallel that of temperature. Furthermore, under specific combinations of pressure, wind speed, wind direction and cloud cover, the mixing ratio humidity island can exist separately. Statistically it was found that alteration of the analysis from the parametric to the non - parametric drastically modified the resulting conclusions, epecially when air mass surface layer complexity increased. Previously undetected and insignificant mixing ratio and temperature regimes become exposed with non - parametric statistics.
