This dissertation investigates how visual analytics tools and techniques can address ambiguity in complex risk assessment, prediction, and monitoring, focusing on the domain of avalanche forecasting. Drawing on a broad set of methods and theory from complex cognitive systems engineering and visualization research, this dissertation delves into the cognitive work demanded by this domain and explores visual analytics solutions to enhance sensemaking.
In a study using a variety of methods including interviews, observational research, and situated-recall, this research identifies and characterizes the issues of ambiguity in avalanche forecasting as they pertain to individual and collaborative sensemaking around data. It presents the results of a participatory design study that develops visualization tools to tackle these challenges and an evaluation study investigating the analytic affordances and sensemaking support provided by newly designed and existing tools used by forecasters. In addition, a preliminary study using participatory design and diary study methods investigates how knowledge construction and synthesis can be supported to better address challenges of shared sensemaking in asynchronous sequential collaboration.
Findings from this dissertation reveal the shortcomings of conventional visualization guidelines in being able to tackle ambiguity in this complex domain. Instead of employing efficient and effective perceptual encodings and summary overviews, it highlights the significance of flatter visual hierarchies, visual difficulty, and rapid access to details for better support of sensemaking around ambiguity. In addition, it reveals new challenges and opportunities for improved knowledge synthesis support in visual analytics tools. The theoretical framing and methodological approach used in this dissertation is novel for the domain of visual analytics.
Click here to access a copy of Stan’s thesis.