Visualizing simulation ensembles of extreme weather events. (May 2022)

• Record Type: Journal Article
• Authors: de Souza, Carolina Veiga Ferreira; Luz Barcellos, Priscila da Cunha; Crissaff, Lhaylla; Cataldi, Marcio; Miranda, Fabio; Lage, Marcos
• Is Part Of: Computers & graphics. Volume 104(2022)
• Page Start: 162
• Page End: 172
• Publication Date: 2022-05
• Subjects: Visual analytics -- Weather visualization -- Ensemble visualization; Computer graphics -- Periodicals; 006.6
• DOI: 10.1016/j.cag.2022.01.007
• Languages: English
• ISSNs: 0097-8493

Abstract

Abstract: In the last 20 years, extreme weather-related events like floods, landslides, droughts, and wildfires have caused the death of 1.23 million people and a loss of 2.97 trillion dollars. Studies show that low and lower-middle income countries are the most impacted ones given the lack of investment in disaster risk management. To reduce the impact of these events, weather researchers have been developing numerical weather models that inform public agencies about the impending extreme events in advance. Despite being powerful tools, these models can suffer from several sources of uncertainty, ranging from the approximation of micro-scale physical processes to the location-dependent calibration of parameters, which is especially critical in developing countries. To minimize uncertainty effects, researchers generate several different weather scenarios to compose an ensemble of simulations that typically are inspected using manual, laborious, and error-prone approaches. In this paper, we propose an interactive visual analytics system, called X-Weather, developed in close collaboration with weather researchers from Brazil. Our system contributes a set of statistics and probability-based visualizations that allows the assessment of extreme weather events by effortlessly navigating through and comparing ensemble members. We demonstrate the effectiveness of the system through two case studies analyzing tragic events that happened in the mountain region of Rio de Janeiro in Brazil. Graphical abstract: Highlights: Our visualizations help identify extreme weather scenarios in large simulation ensembles. Our web-based system enables the investigation of individual weather ensemble members. Two case studies highlight our system's utility by analyzing extreme weather events.