Application Spotlights

Visualization in Meteorology and Climate Sciences: Recent Research and Open Challenges

Sunday, October 20: 2:20-3:50PM in Room 15

Organizers: Marc Rautenhaus (Regional Computing Center, Visualization, Universität Hamburg, Hamburg, Germany), Michael Böttinger (German Climate Computing Center (DKRZ), Hamburg, Germany)

Meteorology and climate sciences are recurrent application domains in research on visualization and display design, and of great societal significance. Likewise, from the meteorological point of view, visualization is an important and ubiquitous tool in the daily work of weather forecasters and atmospheric and climate researchers. Characteristics of recent research in the application domain include the increasing use of ensemble simulations to represent uncertainty, increasing resolutions of numerical models, combinations of heterogeneous data types, and raising interest in machine learning methods. These themes are also represented in recent visualization research that include uncertainty visualization and interactive visual analytics workflows. Thus, both scientific areas –visualization and meteorology– can gain much benefit from collaboration and already do. The objective of our spotlight session is to further enhance collaboration aspects by discussing recent research and open challenges. We propose an overview talk, based on our recent survey on visualization in meteorology (Rautenhaus et al., TVCG 2018), followed by a number of invited presentations that spotlight selected challenges from both visualization and domain-specific points of view, including uncertainty, interactive analytics and large data. Speakers will be recruited from both communities. The session will be concluded with a discussion on how to improve collaboration.

Visual Analysis of Air and Maritime Trajectory Data

Sunday, October 20: 4:10-5:40PM in Room 15

Organizers: Andy Wilson (Sandia National Laboratories, Albuquerque, New Mexico, USA)

Air and sea transport are critical parts of worldwide business and commerce. Data about airplane and ship movement are plentiful. With trillions of dollars in play, even small improvements and insights can make a big difference in practice. Visual analysis helps us identify patterns of behavior, inform decisions about routing and scheduling, and infer external influences upon the vehicles in motion.

Analysis of air and maritime traffic is influenced by a few major factors that distinguish it from other types of trajectory data:

  1. Aircraft and ships have greater freedom of motion than automobiles and trains, which are limited to roads and railroad tracks.
  2. Behavior is determined by a combination of convention, external management, and responses to local, exigent circumstances like weather.
  3. Data coverage and quality vary widely depending on geography and sensor type.

In our session, we will discuss the synergy between algorithms and visualization that arises while analyzing air and sea trajectory data. We will survey the current state of the art in addressing these challenges including work from computational geometry, machine learning and visual analytics, and demonstrate applications that address real-world needs.

Visualization Software Development for Researchers and End Users - from general issues to specific challenges for medical applications

Monday, October 21: 2:20-3:50PM in Room 15

Organizers: Hank Childs (University of Oregon, USA), Takayuki Itoh (Ochanomizu University, Japan), Michael Krone (University of Tübingen, Germany), Guido Reina (University of Stuttgart, Germany), Christina Gillmann (University of Leipzig, Leipzig, Sachsen, Germany), Thomas Wischgoll (Wright State University, Dayton, Ohio, USA), Gerik Scheuermann (Institute of Computer Science, Leipzig University, Leipzig, Germany)

Visualization has evolved into a mature field of science; it has become widely accepted as a standard approach in diverse fields, ranging from physics over medicine to business intelligence. Over the years, many methods and tools have been developed and published, however, most of these are prototypes and never reach a state that can be reliably used by domain scientists. The prototypes are often neither sustainable nor extensible for subsequent research projects. Additionally, current problems and data sets have grown so large and complex that novel methods require an exceedingly large amount of engineering to approach.
This barrier to entry is a major hindrance for new Ph.D. students, which have no time budget for software engineering but have to produce research results. The vis community has no mechanism for amortizing development time. Thus, we either need to find a way to enable developing sophisticated and novel visualizations with minimal overhead or we need to incentivize and reward developing mature visualization applications and frameworks.
We recently organized a Shonan seminar (#145) addressing this topic. As a group, we identified nine topics for visualization software in general that require significant attention, with a special focus on research challenges. As one part of the discussion, we want to give a detailed perspective on medical visualizations, which results in more severe constraints and challenges: Software designed for clinical daily routine can require years until its deployment due to extensive test phases and legal processes.
Generally, requirements for successful visualizations differ from the requirements for successfully publishing a paper in the VIS community, i.e. scientific novelty does not speak for its applicability in daily routine. For example, the number of novel VIS contributions in the medical area does not directly translate into a larger number of VIS techniques being used in the clinical setting.
The proposed spotlight will be organized similar to a panel discussion. We will give one primer talk summarizing the Shonan seminar and one detailing the specific issues of medical visualization ourselves. Subsequently, four invited panelists will give additional perspectives on the problems: Chris Johnson, David Laidlaw, Paul Navratil and Anna Vilanova. There will be ample time for discussion since we want to give the vis community as a whole the opportunity to extend and refine the scope of the problem and to join the efforts of the seminar attendees to tackle these challenges. One specific outcome of this session will be a rough concept for a follow-up Dagstuhl meeting on this topic.

Visualization Paradigms in the Renewable Energy Space

Monday, October 21: 4:10-5:40PM in Room 15

Organizers: Kenny Gruchalla (Computational Science Center, National Renewable Energy Lab, Golden, Colorado, USA), Mr. Haiku Sky (National Renewable Energy Laboratory, Golden, Colorado, USA), Nicholas Gilroy (National Renewable Energy Laboratory, Golden, Colorado, USA), Melissa Queen (University of Washington, Seattle, WA, USA), Xiaoying Pu (University of Michigan, Ann Arbor, MI, USA)

Visualization is commonly used in the renewable energy space, not only by analysts, but also for communication to policymakers, city planners, and the general public. The scope and style of these visualizations vary based on the data and end goals. At the National Renewable Energy Laboratory, 3D immersive visualization is a prominent method for exploring spatially complex data, such as the morphology of photovoltaic materials. However, this paradigm is not appropriate for many techno-economic studies, such as investigating where to put new wind farms or the cost of adding new batteries to the power grid. In these cases, visualizations tend to be more map-based and deployed as web or desktop applications and on large scale 2D displays. While the types and data and targeted audience within this space is broad, there are some overarching themes including the relationship between cost, energy capacity, and economic and regulatory policy. These themes make both the science and visualization more complicated, forcing simulations to integrate energy engineering with policy and economics and to visualize multitudes of scenarios simultaneously. In this spotlight, we will give a primer to the varied types of energy technologies within this space as well as common visualization strategies for these types. We will then dive into 4-5 lightning talks on visualization systems that are pushing state of the art and changing how renewable energy is created, deployed and utilized.

Visualization Enabled Scientific Discovery

Tuesday, October 22: 4:10-5:40PM in Room 2+3

Organizers: Anders Ynnerman (Department of Science and Technology , Linköping University, Norrköping, Sweden), Ingrid Hotz (Department for Media and Information Technology, Linköping University, Norrköping, Sweden)

This spotlight session will analyse and discuss the instrumental role of visualization in academic research projects. In the Swedish eScience Research Center (SeRC), visualization is one of the core methodologies supporting a number of strategic application areas (eScience communities). We will highlight significant results from SeRC application areas to provide examples of visualization enabled discoveries, and also discuss the mechanisms behind successful academic collaborations based on visualization research and development.

Does AI Mean Data Visualization is Dead? A discussion with IBMers working at the intersection of AI and data visualization about the opportunities and challenges of building next generation business intelligence products.

Wednesday, October 23: 11:00AM-12:20PM in Room 1

Organizers: Jamie Waese (Cloud and Capplication-spotlights.mdognitive Software, IBM, Toronto, Ontario, Canada), Anne Stevens (Cloud and Cognitive Software, IBM, Toronto, Ontario, Canada), Afrooz Samaei (Cloud and Cognitive Software, IBM, Toronto, Ontario, Canada), Stephen O’Connell (Cloud and Cognitive Software, IBM, Toronto, Ontario, Canada), Frank van Ham (Cloud and Cognitive Software, IBM, Weert, Netherlands)

Data visualization leverages visual and cognitive processes to make connections, recognize outliers and forecast trends. It is a uniquely human endeavour that is central to how we process information, make decisions and communicate. However, modern statistical algorithms (sometimes called ““AI””) provide powerful ways to analyze ever larger datasets. If AI is better at finding patterns and outliers than we are, do we still need charts and graphs to gain insights from data?

IBM Cognos Analytics is a reporting tool for business intelligence. As we expand our AI capabilities to improve data engineering, forecasting, natural language query, annotation and exploration processes, we must reconcile the implications of what AI means for Business Intelligence tools more broadly.

Can AI predictions and annotations have as much impact as a well-crafted chart? Can they be as informative and engaging? Is AI any less biased than a human analyst? Will dashboards become obsolete? How can AI support time series forecasting, segment clustering, dimension reduction and other prediction and classification tasks for visualization purposes? How do humans respond to all of these suggestions and what’s their future role in this process?

In this Application Spotlight we will discuss these questions and some of the new AI features we are working on. We look forward to a lively discussion of this subject and seek feedback from the visualization community for guiding our product roadmap into the future.

Knowledge-assisted Visual Analytics meets Guidance and Onboarding

Thursday, October 24: 4:10-5:40PM in Room 2+3

Organizers: Christina Stoiber (St. Pölten University of Applied Sciences, St. Pölten, Austria), Markus Wagner (St. Pölten University of Applied Sciences, St. Pölten, Austria), Davide Ceneda (TU Wien, Vienna, Austria), Theresia Gschwandtner (TU Wien, Vienna, Austria), Margit Pohl (TU Wien, Vienna, Austria), Silvia Miksch (TU Wien, Vienna, Austria), Marc Streit (Johannes Kepler University, Linz, Austria), Dominic Girardi (datavisyn GmbH, Linz, Austria), Wolfgang Aigner (St. Pölten University of Applied Sciences, St. Pölten, Austria)

Using explicit knowledge for both, onboarding and guidance can be highly beneficial towards more usable and customized VA products. However, these approaches have not yet been systematically combined. In this proposal, we want to bring together these three emerging topics, elaborating on their commonalities, differences, and discussing the necessity of new methods to effectively integrate them into VA. Research in this area is particularly application-driven and empowering users with onboarding and guidance is not only an open scientific-technical challenge but also crucial for software solution providers in concrete application domains.

Feature-based Visual Interactive Systems to Optimize Decision Making

Friday, October 25: 9:00-10:40AM in Room 2+3

Organizers: James Ahrens (Los Alamos National Laboratory, Los Alamos, New Mexico, USA), Soumya Dutta (Data Science at Scale, CCS-7, Los Alamos National Lab, Los Alamos, New Mexico, USA)

Real-world decision makers, such as scientists and engineers are confronted with the needs to make critical decisions on a daily basis. Typically, this requires the understanding of the high-dimensional input parameter space and associated outcomes. To facilitate this understanding, one promising approach involves the definition and extractions of features of interest and the use of these features to drive the interactive exploration of the parameter space. Our application spotlight will focus on examples of these systems and synthesize common problems and themes.
We envision incorporating talks from leading visualization and application experts in this area including Dan Keefe (University of Minnesota) on the design of medical instruments, Valerio Pascucci/Jackie Chen (University of Utah/Sandia National Laboratory) on the design of combustion reactions and David Rogers/Richard Sandberg (Los Alamos National Laboratory) on the design of shock physics experiments. We will solicit lightning talks to gather other examples from the community. A discussion session will summarize common themes, methodological advances and next steps and turned into a journal paper for publication. In addition, a set of short papers describing each system will be archived. Potential contributions to basic research include high dimensional space representation, traversal methods and interactivity techniques.