Exploration of Climate Data Using Interactive Visualization

Abstract

In atmospheric and climate research, the increasing amount of data available from climate models and observations provides new challenges for data analysis. The authors present interactive visual exploration as an innovative approach to handle large datasets. Visual exploration does not require any previous knowledge about the data, as is usually the case with classical statistics. It facilitates iterative and interactive browsing of the parameter space to quickly understand the data characteristics, to identify deficiencies, to easily focus on interesting features, and to come up with new hypotheses about the data. These properties extend the common statistical treatment of data, and provide a fundamentally different approach. The authors demonstrate the potential of this technology by exploring atmospheric climate data from different sources including reanalysis datasets, climate models, and radio occultation satellite data. Results are compared to those from classical statistics, revealing the complementary advantages of visual exploration. Combining both the analytical precision of classical statistics and the holistic power of interactive visual exploration, the usual workflow of studying climate data can be enhanced.