Data-driven techniques are vital to improve the efficiency of waste management systems in Canada and around the world. Canadians generated 935.6 kg/cap of waste in 2016 and spent $3.3 billion to manage this waste. This research aims to solve problems related to waste region optimization and landfill siting. Specifically, this research develops topologies for regional waste management systems and analyzes specific features of preexisting regionalized and non-regionalized systems to increase system efficiencies. Methods including Geographic Information Systems (GIS), remote sensing (RS), spatial statistics, and data science are incorporated to inform data driven policies in waste management.

First, a novel recursive algorithm is proposed to optimize the shape of waste management regions using ArcGIS. The outcome was a set of regions where the spread of populated places, roads, and landfills were more equally spread throughout each region. The recursive algorithm was then altered, applying the Central Feature spatial statistics instead of the geometric centroid as the ‘seed’ input in the algorithm. In some cases, this alteration was able to further improve the ability of the algorithm to equally spread infrastructure within regions. Along with provinces, this method was also tested on cities, and the results were positive for larger metropolitan cities, however, limited improvements were observed in smaller urban centers. After proposing a new recursion algorithm and an alteration, the shape characteristics of the regions were examined. Two new metrics were proposed, accounting for the percent spread of landfills, populated places, and roads, as well as the area of each proposed region in a tessellation. The isoperimetric quotient and its measures of central tendency are used to further evaluate proposed tessellations. The number of sub-regions in a tessellation and the area of the resulting polygons are important factors in spatial optimization of waste management regions. Finally, other spatial statistics are investigated in smaller regions of Saskatchewan. Centroidal Voronoi Tessellation is generally the most efficient method. However, when the ratio of populated places to road length was low, the median center spatial statistic was better able to optimize populated places. A trade off exists between the number of sub-regions, computational efficiency, and optimized percent standard deviation. Size of sub-regions remains important.

Three novel analysis methods are also presented. First, a method for determining optimal landfill expansion locations is proposed. The method was able to determine the best areas for landfill expansion. Independent of expert opinion, the method relies solely on vector and remote sensing data;​ but is flexible enough to apply weighting methods if so desired. Remote sensing and vector data can capture distinctly different aspects of the study area, and vector data can be used as a proxy when cloud cover is present. Next, Standard Deviational Ellipses were used to assess costs related to waste collection in Nova Scotia. The distribution of waste facilities and road network significantly impact collection costs and a framework for geospatially dependent policies in Nova Scotia is presented. Finally, landfill design criteria are assessed and compared using data science techniques. Word distributions and other characteristics are related to climatic and other features in each province. Tables and figures are important in Canadian landfill design standards and guidelines