GEOG 506 2013 Advanced GeoInformation Science Final Projects
Congratulations to all the students and my co-instructor, Raja Sengupta, for another great Advanced GIScience course! Below are the final project abstracts. Be sure to click through to read the entire list:
Dynamic Multi-Scale Standardization in Watersheds
Scale and scale type variation between disciplines often makes it difficult for researchers performing watershed analysis to compare their results with other studies. The scales used in watershed analysis are not well defined and are rarely reviewed, leading to many questions as to how spatial and temporal scale can be standardized, if it is even possible at the present time. The result of analysis of scale has resulted in the development of a criteria and framework, known as “Dynamic Multi-Scale Standardization” for the classification of scales into common standards. This framework provides cross-discipline understanding to watershed scale and is further aided by the creation of an algorithm to group scale. The techniques shown provide a non-arbitrary scaling that is both justifiable and replicable, which provides legitimacy and credibility to scale.
Experiencing Gender in Cyberspace: Examining the Possibilities and Limitations of the “Cyber-flâneuse”
This project explored the way that gender is experienced with respect to the mediation of technology. I took on the role of “cyber-flâneuse”-- a situated female observer --with the aim of producing an emotional map of four cyberspace locales. Taking virtual strolls through Twitter, Facebook, Youtube and Tumblr, I sought to understand whether these web platforms were sites of violence, and how the hostile acts contained therein impacted on my experience. Charting my emotional reactions, I collected notes and excerpts from my stroll that were used to produce a composite map of my emotional experience. ArcGIS representational tools were used on a DEM dataset to employ spatial metaphor in the service of producing emotional cartography. I found that cyberspace, while often understood as being less real than physical space, facilitated repeated reassertion of my own physicality. I conclude that while many of the barriers of flânerie are broken down by the online domain, there remain features of the space that make it unsafe or undesirable for women, and which generally prompts self-preservation or avoidance behaviours. The production of emotive geographies arises as a challenge to the bourgeouis notion of objectivity that permeates both flânerie and GIS.
Flickr as VGI: Geotagged photo sharing as a mode of volunteered geographic information analysis
The advent of volunteered geographic information (VGI) has spawned numerous Web applications encouraging user interaction with geographic data. One form of data that is not often considered geographic, per se, is photography. However, Flickr, a popular photo sharing website has implemented a map to accompany the geotagging and geolocation of users’ photographs. More interestingly, those photos are distributed spatially and across a wide range of users. This study examines Flickr photos as a form of VGI, focusing not only on explicit geographic information, but also on how geotagged photography can help us understand how users understand and interact with their environments.
Uncertainty in Canadian Postal Code Geography: A comparative analysis assessing multiple postal code datasets
Uncertainty is inherent in geographic information, and postal codes are no exception. The lack of standard boundaries in postal code areas produces discrepancies in postal code information across datasets. This analysis reviews geographic information uncertainty and applies concepts of positional accuracy and ambiguity to the geography in several postal code datasets of varying data quality standards. Comparing third party datasets to benchmark standards concludes the level of relative positional accuracy and allows recommendations of whether the dataset is fit for use in postal code applications. Analysing ambiguity in postal code area boundaries allows recommendations of whether the geographical discrepancies across multiple datasets poses implications for research and postal code applications.
A GIScience Based Approach to Investigating Motivational and Privacy Issues of foursquare Users
With recent telecommunication advancements, Location-Based Services have become increasingly ubiquitous, used for things like information services, navigation and geosocial-networking. Due to the large amount of personal and locational information available through these services, issues of privacy within the realm of GIScience have come to light. Furthermore, investigations into why people use certain LBS have been proposed. This study investigates these two broad issues by analyzing the foursquare profiles of top and average users. The findings suggest that the average foursquare user is not highly concerned with privacy implications, choosing to display various kinds of personal and locational information. I also found that many people use foursquare mainly for the gaming aspect, such as becoming a mayor of a certain location or earning badges. Additionally, it was determined that some users participate in foursquare to meet new people, project a self-image, leave recommendations about a location and/or promote a specific business or institution.
Modeling Changes in Commercial Land Use Using Cellular Automation: Montreal 1996-2015
Combining principles of modeling land use, time and GIScience can lead to a spatial decision support system which has implications for zoning and conservation policies. Temporal GIS is not yet entirely integrated into commercial GIS software, providing room for innovation and creativity while limiting the usability by GIS practitioners. Data models, temporal topology, and database structure are key components that must be assessed before any spatio-temporal analysis can take place. Cellular automation, a spatio-temporal modeling method based on weighted neighborhoods and spatial complexity, is an ideal tool for geographers because it embraces the influence of space while combining principles of process-based modeling. It is able to process more data points, faster, and output higher resolutions than analysis in traditional GIS software, while having the potential to be easily integrated into a GIS package.
The goal of this study was to temporally interpolate between two snapshots of land use in Montreal using cellular automation, calibrating the model in the process. The secondary objective was to use this calibrated model to predict the expansion of commercial land use by the year 2015. The added complexity of multiple land use interactions was treated as a modified neural network, where each interaction had its own unique value. The results show that, while the automation was able to interpolate somewhat accurately in terms of magnitude and location of conversions, the rate of conversions was likely unrealistic, and the model failed to account for stochastic events, missing clumps of converted cells as a result. The potential for cellular automation to evolve as a part of GIScience will depend on the assessment of the influence of scale and aggregation, database structure, data representation, and use of data models and topology. This will require the collaboration of concepts from computer science and geography in an effort to make cellular automation a tool that will meet the needs of geographers and modelers alike.
Recalibration of the Internal Compass
Previous research has found that the average cognitive map global reference frame is oriented north. The speed at which, and degree to which, people comprehend spatial information is reduced at orientations away from north. Geographic information systems (GIS) primarily convey spatial information visually. Cartographic standards dictate that all two dimensional spatial representations are to be oriented with north at the top. This conforms to the orientation of the cognitive map global reference frame. Spatial information is not necessarily best suited to be portrayed with north oriented at the top. The goal of the study is to determine whether people are able to reorient their global reference frames and better understand variations the orientation of spatial information. Participants were put through a training process in which they were presented with two= dimensional spatial visualizations at varying orientations, in an uncontrolled environment. Pointing accuracy and latency were recorded to determine their comprehension of the spatial information. Results found that there was little to no improvement in both pointing accuracy and latency. Based on the results, GIS should continue to conform to contemporary cartographic standards.
Using simulated data to evaluate space-time detection and classification of outbreaks of waterborne gastrointestinal disease in Montreal: an application of spatial statistics
Spatial statistics is an interdisciplinary field, heavily influenced by theoretical statistics, biostatistics, geographic information science, and applied disciplines ranging from economics to medicine. This research briefly reviews the field from spatial autocorrelation to hierarchical modeling, and discusses the application of a range of spatial statistical methods to the detection of infectious disease outbreaks. Through the analysis of 100 simulated datasets, I find that syndromic surveillance data via emergency department visits can be used to detect outbreaks of gastrointestinal disease across the island of Montreal. I also explore the hypothesis that outbreaks exhibit unique space-time patterns based on the outbreak source (i.e., location of water plant failure), which could potentially allow for disease pattern classification. Interest in better understanding spatial heterogeneity in disease risk and modeling disease spread over space and time has motivated much of the research in spatial statistics over the last two decades.
Quality Assessment of Geocoding Services and Reference Datasets for Rural Street Addresses in Southern New Brunswick
Geocoding is used by people all over the world for a wide variety of applications, but most geocoders are not good at locating rural addresses and popular online geocoding services generally sacrifice transparency for convenience. Adding to the work of previous geocoding quality assessments, this study compares two different reference datasets in ArcGIS and one online geocoding service to evaluate their performance with respect to match rate and positional accuracy, as well as to identify the source of error among various geocoding components. A list of 11,977 rural New Brunswick addresses was geocoded under each configuration, and the 3 result sets were mapped and analyzed. Results show low match rates among the two reference datasets geocoded in ArcGIS stemming from erroneous place-name attributes, while the online geocoding service exhibits a very high match rate but much lower positional accuracy. The study finds significant problems with place name attributes in the reference datasets, and evidence of reference dataset and processing algorithm issues influencing the online geocoding results.
Integration of Remote Sensing and GIS in Hydraulic Fracturing in Alberta, Canada
A geographic information system (GIS) is a processing tool to input, store, manipulate, analyze, and display the spatial resource data that supports a decision making process. Remote Sensing deals with the immediate access of primary data at a more continuous scale, collected over extensive areas at rapid temporal frequencies. This research considered the historical problems and developments in integrating these two approaches. Current integration practices, as well as their limitations, were also identified. Then, the research explored the potential developments and concerns going forward. The purpose of this analysis was composed of two parts. First, it sought to build a comprehensive map of hydraulic fracturing (fracking) activities, as well as the areas affected by these activities, in the Canadian province of Alberta. The second part of this analysis identified park areas within the study region, using a combination of remotely sensed satellite imagery and GIS manipulated data. The results were compared to data obtained through DMTI of confirmed park locations to evaluate accuracy. Overall, the analysis was fairly accurate in identifying potential park areas. However, the analysis was limited by the quality of data, software availabilities, and time constraints. Overall, the results can be taken as evidence of the advancements in the field of the integration of Remote Sensing and GIS.
Spatial Data Mining Species-Specific Preferences from GPS Tracks
Dipto Sarkar (IIIT-D exchange)
Understanding animal behaviour plays a central role in knowing its preferences. As animals move through their environment, they generate movement patterns, which are a combined result of habitat characteristics and species-specific preferences. GPS technology has enabled capturing of the movement patterns in great detail both in terms of spatial and temporal dimensions. These movement patterns contain a large volume of latent information about the species specific preferences and how they are manifested in the habitat in which they dwell. This project aims at data mining the preferences exhibited by the species by mining the movement data obtained through GPS transmitters. Spatial Data Mining approaches will be used to extract the spatio-temporal patterns and other habitat preferences hidden within the large GPS datasets, which will be further refined to derive rules that define the preferences of the species as a whole. In this research the first part of the entire process have been achieved and is discussed in details. It also provides some direction as to how next part will evolve to determine the preferences of the species as a whole.
Augmented Reality for News Media
Three successive moments have led to the increased influence of augmented reality (AR) within the urban environment - smartphones, mobile web, and the growth of geographic content. The ability to superimpose digital information on increasingly popularity of social media has led to a new trend in use and functionality for AR systems. This research is an exploration of a location based AR application (“app”) to browse for online video media (specifically Montreal Gazette news media). The research objective is to develop an AR app that will overlay video streams uploaded by the Montreal Gazette on their YouTube channel for Android phones to access the ways in which interpretations can differ from augmented video rather than traditional text and the ramifications for the field of GIScience.