Supporting Event-based Geospatial Anomaly Detection with Geovisual Analytics

Abstract

Collecting multiple geospatial datasets that describe the same real-world events can be useful in monitoring and enforcement situations (e.g., independently tracking where a fishing vessel travelled and where it reported to have fished). While finding the obvious anomalies between such datasets may be a simple task, discovering more subtle inconsistencies can be challenging when the datasets describe many events that cover large geographic and temporal ranges. This paper presents a geovisual analytics approach to this problem domain, automatically extracting potential event anomalies from the data, visualizing these on a map, and providing interactive filtering tools to allow expert analysts to discover and analyze patterns that are of interest. A case study is presented, illustrating the value of the approach for discovering anomalies between commercial fishing vessel movement data and their reported fishing locations. Field trial evaluations confirm the benefits of this geovisual analytics approach for supporting real-world data analyst needs.

A geovisual analytics approach for analyzing event-based geospatial anomalies within movement data

“…VEMEA automates the menial aspects of anomaly detection, allowing the analyst to consider, examine, and explore among a much larger number of potential anomalies than with the existing methods. Previously, we have outlined VEMEA’s core features and presented evidence of its value; 7 this article provides a more detailed explanation of the system, two case studies showing how the system can be used to support the confirmation of suspected or known movement-event anomalies (RQ2) and the isolation of unknown movement-event anomalies (RQ3) and a comprehensive analysis of expert data analysts’ field trial evaluations of VEMEA, which focus on the usefulness (RQ4) and ease of use (RQ5) of the approach.…”