Visual Analytics for Comparison of Ocean Model Output with Reference Data: Detecting and Analyzing Geophysical...

Area of Science:

• Oceanography
• Computer Science
• Data Visualization

Background:

• Ocean model quality assessment relies on comparing model output with previous versions or observations.
• Current methods use statistical aggregation and visual comparison, which can be limited in scope and prone to overlooking details.
• Analyzing large datasets of geographically referenced temporal profiles presents significant challenges for modelers.

Purpose of the Study:

• To develop a Visual Analytics approach to broaden the scope and reduce subjectivity in ocean model data comparison.
• To facilitate the detection and analysis of differences and similarities in geophysical processes between datasets.
• To improve the quality assessment of ocean models by providing a more comprehensive comparison framework.

Main Methods:

• Multiple clusterings of temporal profiles based on various features and algorithms to capture diverse temporal behaviors.
• Clustering ensembles to consolidate multiple clusterings into a single overview of geospatial distribution of temporal patterns.
• An interactive visual interface for comparing consolidated clusterings and identifying geophysical processes.

Main Results:

• The proposed Visual Analytics approach enables a broader scope of analysis compared to traditional methods.
• It reduces subjectivity by consolidating information from multiple clusterings.
• The approach facilitates the detection of specific geophysical processes and aids in identifying areas for ocean model improvement.

Conclusions:

• The Visual Analytics approach effectively enhances the comparison of ocean model datasets.
• It provides a more comprehensive and less subjective method for analyzing geophysical processes.
• This technique was successfully applied in collaboration with ocean modelers to refine the Ocean Model for Circulation and Tides (OMCT).