Visualizing of Clusters of Earth Grid Cells Based on Global Atmosphere Data and Cluster Changes Over Time

• Activity Number: 255
• Type: Contributed
• Date/Time: Monday, August 1, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistical Graphics
• Author(s): Daniel Carr and John Ashley*+
• Companies: George Mason University and NVidia
• Keywords: Evolving clusters ; color palettes ; land and ocean

Abstract

Studying global multivariate multi-altitude atmosphere data derived from satellite monitoring poses visualization challenges. Based on radiances obtained from the Atmospheric Infrared Spectrometer on the Aqua satellite, NASA computes geophysical parameters for satellite foot prints. One level 3 NASA product consists of monthly entropy constrained vector quantized summaries of footprints in 5 degrees earth grid cells. The number of summary vectors varies from cell to cell. The Wasserstein metric provide a way to obtain the expected distance between grid cells over space and time. Clustering based on very large distance matrices provide an overview based on clusters that grow or shrink and change locations over time. Encoding many clusters using color in a readily understandable way is a visualization challenge. The proposed solution uses sea surface temperature of clusters primarily over the ocean as a basis for picking colors from an ocean color palette. Similarly, for land clusters the NDVI greenness index proves a basis for picking colors from a land palette. for primarily land clusters. The graphics also address change blindness is showing changes across sessions and years.