A Visual Analytics Approach to Multiscale Exploration of Environmental Time Series

Area of Science:

• Environmental Sciences
• Data Science
• Computer Science

Background:

• Detecting temporal patterns in numerical time series is vital for environmental science.
• Identifying these patterns often depends on the chosen time scale and starting point.

Purpose of the Study:

• To develop a visual analytics approach for detecting interesting patterns in numerical time series.
• To address the limitations of predefined time scales and starting positions in pattern detection.

Main Methods:

• An algorithm to compute statistical values across all possible time scales and interval starting positions.
• A matrix visualization for identifying potentially interesting patterns.
• Interactive exploration tools for detailed analysis of detected patterns.

Main Results:

• The approach successfully identifies significant temporal patterns in environmental data.
• Demonstrated utility in two distinct scientific scenarios.
• Facilitated the discovery of new insights into environmental system dynamics.

Conclusions:

• The visual analytics approach offers a flexible and powerful method for time series pattern discovery.
• It enables scientists to explore environmental data without prior assumptions about temporal scales.
• The method enhances understanding of complex environmental system dynamics.