GEOG 506, Advanced Geographic Information Science Projects for 2015


Welcome to another year of GEOG 506 projects, ranging from Smart Cities to geospatial ontologies for agriculture. This year we had eleven. Here are the abstracts, with names, (GIScience field) and Project Title

Beedell (Drones): Drones in GIScience

The advancement of drone technology and decreasing costs in recent years have led to a rapid rise in drone use in official research and citizen science. This paper exhibits how small drones can be used as a platform to create spatially referenced data from attachable sensors. While traditional advantages and disadvantages of remote sensing by satellite or airplane have been extensively studied, there is relatively little literature on the new use of drones as a tool in GIScience. Prior literature focuses on drones as a tool for particular GIS applications, such as traffic monitoring or agricultural monitoring but fails to address the utility of drones as a platform for other sensors and uses. Given that quadcopter drones have a highly limited payload and battery-life, few attempts have been made towards flights with other sensor attachments. Through several experimental drone flights, this paper demonstrates that a quadcopter drone can be used in combination with a smartphone-based anemometer to take measurements of wind velocities at various altitudes to measure wind shear. This method is expected to expand the possibilities of wind sensing, particularly in terms of rapidly measuring and sharing the data for applications in wind resource assessments or wind sports. As drone technology moves away from the military realm, more research is required to determine the limits and uses of this new technology, particularly in the area of how drones can be used in citizen science.

Bloom (Spatial Cognition): Meta-analysis of Haptic Mobile Pedestrian Navigation Systems: A GIScience Perspective

Mobile pedestrian navigation systems have traditionally been dominated by visual interfaces. However, the advancement of technology in recent years has enabled the development of multimodal location based services. This burgeoning area of research has produced technologies using a variety of approaches and conceptual frameworks. Most experimental studies about haptic-GIS for wayfinding are published in engineering journals, and focus on sensory-feedback navigational systems within a tool-making context. However, there is a lack of literature that approaches these technologies from a geographic information science (GIScience) framework. GIScience may be defined by ​ GIScience is defined by, “the development and use of theories, methods, technology, and data for understanding geographic processes, relationships, and patterns” (Goodchild, 2010). ​Therefore, the project performs a meta-analysis to synthesize these studies in the context of a GIScience perspective. Results sort literature based on user-scale, spatial knowledge of the environment (i.e landmarks, routes, and survey knowledge), and cognitive and perceptual spaces. Synthesis of haptic-GIS literature hopes to inform future considerations for applying haptic-feedback to navigational interfaces and uncover any limitations in existing literature.

Coffman (Critical GIS): An Emotion Map of Stress at McGill University

This project explores emotion mapping, grounded in the framework of critical GIS. Critical GIS critiques the nature of GIS, and the producers, users and societal implications of these technologies. The literature of critical GIS focuses on a variety of topics, such as feminist GIS and qualitative GIS. Incorporating qualitative methods and data in GIS is difficult and complex; however, it can unveil new knowledge and interpretations. Emotion mapping has become one such form of qualitative critical GIS. This project draws on other work in emotion mapping in order to explore spaces of stress on McGill’s campus, by observing students in a variety of spaces, and measuring my own personal stress. In doing this project, I hope to demonstrate how qualitative data like emotions can be incorporated in GIS, and what these types of methodologies and data can bring to enhance our knowledge.

Conzon (Volunteered Geographic Information): Challenges In Georeferencing User-Generated Content From Harvested Tweets Regarding London’s Leytonstone Stabbing

Currently, with the copious amounts of user-generated content (UGC) that are instantaneously and unceasingly used and shared, social media websites, like Twitter, have contributed to big data (Jung 2015, 52). More specifically, Twitter produces 560 million tweets per day, and these tweets have a “geographic footprint” that can socially and spatially enrich geospatial datasets (Stefanidis et al. 2011, 319; Sui and Goodchild 2011, 1742). Furthermore, tweets can provide both “volume” and “data depth” that used to be a trade-off when producing traditional geospatial datasets (Sui and Goodchild 2011, 1742). This widespread strategy to inductively collect spatially referenced UGC to produce new (or complete pre-existing) geospatial databases is a recent phenomenon called volunteered geographic information (VGI) (Goodchild 2007, 212). Nevertheless, these produced datasets are extremely spatially heterogeneous because data is collected from a varied and dispersed population who are unaware that their social media posts are mined (Li et al. 2013, 62); even more, data collection methods have focused simply on geotagged tweets rather than accounting for the various ways tweets are spatially referenced. With this in mind, this GIScience research project will explore how to account for these limitations and the varying ways to identify spatial information when creating a VGI database of mined tweets. The questions that were explored included: (1) What are the challenges in georeferencing user-generated content from harvested tweets? (2) What is the spatial distribution of the geocodable Leytonstone tweets within England? Therefore, this research project harvested tweets regarding the December 5, 2015, stabbing at Leytonstone subway station in east London; consolidated a geospatial database and highlighted the limitations and errors that persist within these heterogeneous user-generated databases; and then visualized and statistically analyzed the spatial distribution of the geocodable Leytonstone tweets within England. From the 15,127 tweets assessed, 0.2% were explicitly geotagged, 43.3% were implicitly georeferenced through the user’s profile location, and 0.1% were implicitly georeferenced through tweets’ texts, which means there was a total of 6,602 tweets that were geocodable (43.6% of all collected tweets). This reaffirms that it is important to look “beyond the geotag” when analyzing spatially referenced tweets (Crampton et al. 2013, 133). In regards to the second research question, the statistical analysis of the spatial distribution indicated a distance decay.

Crook (Spatial Data Uncertainty): Spatial Data Uncertainty in Land Cover Classifications: A Case Study of Northern China's Grasslands

Uncertainty is an inevitable reality of spatial data, and can belong to one of three types: error, vagueness and ambiguity. In China, policies to mitigate overgrazing on grasslands use classifications of grassland type that are subject to uncertainty. A GIS and remote sensing data were used to determine whether error, vagueness or ambiguity were the most appropriate ways of understanding uncertainty in the classification of grasslands in Inner Mongolia and Xinjiang. Methods included the creation of confusion matrices as well as cluster analysis. The advantages and disadvantages each of the three approaches were then compared. The error approach was found to be overly simplistic in its representation of reality, but easier to apply. The vagueness approach was found to more representative of the reality of grasslands, but less practical to apply to research. Ambiguity was not found to be a major issue. The choice of vagueness versus error was deemed to be partly dependent on the desired resolution of the data. The findings reflected tension between GIScience perspectives of data uncertainty and the pragmatic aspects of applying spatial data to policy.

Davies (Geospatial Ontologies): Geospatial Ontology of Agricultural Regions: An Exercise in Knowledge Engineering

This research project attempts to construct a geospatial ontology of global agricultural regions. A geospatial ontology is a formal conceptualization of the spatial relationships of geographic phenomenon. The purpose of a geospatial ontology is to increase the understanding of the semantics in a domain for computers and to increased interoperability between data sets.

This research project explores the challenges of creating a ‘universal’ conceptualization of a social geographic phenomenon as it pertains to GIScience. This ontology of agricultural regions aims to answer questions like what is a farm? How is a farm differentiated from other landscapes on the Earth’s surface? What is the typology of farming systems? The ontology in this study takes a top-down approach using semantics identified from Derwent Whittlesey’s agricultural region classification scheme. Ultimately this ontology may be useful for governing bodies in environmental monitoring or environmental decision-making on a national scale.

Herrmann (Open Data): Exploring the Opportunities and Challenges to Using Open Datasets for Geospatial Analysis: A Case Study of Gentrification in Chicago

With increased government transparency and efficiency as its stated ends, the open “data deluge” is upon us. Increasingly, efforts have been made to formalize open data implementation by establishing some parameters of how and where open data should be made accessible. Underreported and overlooked, however, is the performability of open datasets for the purpose of analysis, and specifically, how well available open data conforms to the needs and constraints of geospatial analysis. This paper will focus on a variety of benefits and issues surrounding the use of open datasets for geospatial analysis, including: use of different geometries (using point data versus data provided as or aggregated to polygons, such as census tracts); the scaling and aggregation of open data; the use of geostatistics to interpret data, and the use of geographic masks to anonymize potentially sensitive data. To illustrate the utility strengths and limitations of open datasets, this paper will conduct a case study using open datasets related to public and private investment and gentrification in the city of Chicago. The aim of this project is to: (a.) review literature in which open data is used to study various geographic, social phenomena; (b.) synthesise and implement procedures identified in the literature review for the aims of the case study; and (c.) to discuss the limitations of open data used in the case study and recommend best practices for open data providers and researchers hoping to use open datasets relating to geographic phenomena.

Huddart (Smart Cities): Tracking #Manifencours: Smart City Surveillance of GeoSocial Media

The “smart city” concept is a prescriptive package of diverse policies for the future of urban governance, generally proposing more effective and widespread use of information and communications technologies (ICT), particularly for the gathering and analysis of spatial data. As the discipline concerned with the use of spatial data handing, GIScience is integral to the rollout of technologies that constitute the smart city. Advanced sensing techniques, which may include human sensors as well as embedded or remote sensors, have been proposed by GIScience researchers as a promising technology for smart-city urban monitoring. Humans, communicating through Twitter and other publically viewable social media, provide a promisingly effective and inexpensive way to sense relevant spatial data, with examples of social-media-based systems to detect and analyze earthquakes, epidemics, and other phenomena relevant to urban governance. This project attempts to model how a “smart city” might use Twitter data to detect and monitor protest events in the context of ongoing anti-austerity protests in Quebec, Canada, by parsing tweets marked with the hashtag #manifencours, analyzing the spatial data contained within the text of the tweet, and visualizing the spatially-referent tweets on a public webmap. It is argued that governments may use geospatial data harvested from twitter to both detect and locate protest events and analyze the path of protests in urban environments.

Johnson (Geocomplexity): Role of Geocomplexity and Spatial Distribution and Density of Agents in Determining Structure of Friendship Networks

Geocomplexity as an area of inquiry within geographic information science (GIScience) has received little attention from the academic community despite its great potential to advance the field. Complex systems, such as spatial social networks, exhibit sensitivity to initial conditions, emergent behaviour, and scale dependence. Simulation of interactions between individuals through agent-based modelling (ABM) has shown to be an effective way of quantifying the fundamental processes of these complex systems, as they allow for a high level of manipulation of the underlying variables and assumptions.

The primary goal of this study was to see the relative importance that spatial distribution and density of agents within space may play in determining the structure of friendship networks. The formation and maintenance of friendships over time and space were modelled in NetLogo, a simple ABM application. In this model, friendship links between agents are based on (1) their similarity (homophily), and (2) the social preferences of the individual, and (3) the social reach of the individual. A holistic approach centered on geocomplexity theory was taken to achieve this, building off of general systems theory approaches and traditional geospatial network analysis. Quantitative measures were used to characterize the network as population density was increased to test for scale dependence or emergent behaviour of the network. Though this simulation is an abstraction of reality and thus highly simplistic, empirical observations of spatial friendship networks appear to agree with model output thus far.

Smailes (Spatial Scale): Analysing the Length of the Mekong River

Spatial scale is a fundamental concept in GIScience, where it is well documented that results from analyses of processes are scale dependent and cannot be transferred broadly across scales. This paper studies the magnitude of the effects of changing scale properties of input data in the context of river channel analysis, where the observational scale of a study has a direct effect on the end results. Observational scale in this case refers to the resolution, or size, of the pixels used in the raster grid analysis. This project critically assess assumptions about scale by comparing resultant river channel lengths based on an input range of 3 arc-second, 15 arc-second, and 30 arc-second resolution digital elevation models (DEMs), as well multiple line smoothing algorithms. When compared to a control length of the Mekong River derived from high-resolution satellite imagery, the closest final length came from the 15 arc-second DEM that had a Bezier Interpolation line-smoothing algorithm applied. The change in observational scale had a noticeable effect, with an average range of 310 km between the shortest and longest length per DEM for each smoothing application.

Stadler (Sharing Economy): Exploring the Socioeconomic Implications of the Sharing Economy in New York City and Montreal

This study is an assessment of the sharing economy in New York and Montreal. Airbnb and the two prevalent bikeshares of these cities, Citi bike and Bixi bike, are the focus of a study that seeks to explain the distribution of shared resources in metropolitan areas. In exploring the relationship between the sharing economy and average household income, this study found many other determining factors for the distribution of sharing in metropolitan areas. Shared transportation is largely dependent on existing infrastructure, while housingshares are dependent on community structure and proximity to established tourist areas and city centers.