AAAR 32nd Annual Conference
September 30 - October 4, 2013
Oregon Convention Center
Portland, Oregon, USA
Abstract View
An Interactive Visual Analytics Framework for Multidimensional Data in a Geo-Spatial Context
ALLA ZELENYUK, Dan Imre, Zhiyuan Zhang, Jenny Hyunjung Lee, Klaus Mueller, Kevin McDonnell, Pacific Northwest National Laboratory
Abstract Number: 523 Working Group: Instrumentation and Methods
Abstract Single particle mass spectrometers (SPMS) by their very nature produce vast amounts of detailed data, the mining and analysis of which calls for unconventional methods that must draw on statistical methods, while preserving the wealth and depth of information. Success is critically dependent on a powerful data analysis and visualization software packages.
We have developed software packages called ClusterSculptor and SpectraMiner, with the former using statistical tools combined with the operator’s scientific knowledge to steer the data classification process, while the latter is a unique data visualization and mining program that makes it possible to mine datasets of millions of particles using intuitive, visually driven tools.
Recently we showed that with our SPMS, SPLAT II and mini-SPLAT, in addition to single particle size and composition, we measure particle density, number concentrations, mass, asphericity, asymmetry, dynamic shape factor, morphology, phase, fractal dimension, rates of evaporation, and interaction of particles with water vapor (hygroscopic growth factor, activity as cloud condensation nuclei and ice nuclei).
To make use of this wealth of multidimensional information we implemented the parallel coordinates’ interface that allows us to include the additional particle attributes measured by SPMS and other aerosol characterization instruments. The classification and mining process was extended to include, in addition to mass spectral peak intensities, all other observables measured by us or any other instrument, making it possible to quantify their relationships.
During airborne field deployment these data often have a geospatial reference and so it is of interest to show them within their geospatial context. Here we will present a new data visualization and analysis approach that uses Google Earth and is tightly linked to parallel coordinates display. Several other visual representations, such as pie and bar charts are integrated into the Google Earth display and can be interactively manipulated.