Glenn Grant, NSIDC, CIRES, University of Colorado Boulder
Crunching Ice: Compacting Massive Antarctic Surface Temperature Datasets
Abstract: Satellite observations have revolutionized the Earth Sciences, but data and imagery are accumulating at an accelerating rate. Efficient tools for data discovery and quality checking are not keeping pace, especially for studies that do continental-scale analyses at high spatial and temporal resolutions. The Condensate Database Project was designed as an alternative method for data exploration and quality checking. The premise of the project is that much of the data in any satellite dataset is redundant or unneeded and can be eliminated, leaving only the most interesting indicators of change. Using our methods, massive datasets are compacted into more manageable sizes for easy data exploration. As a proof of concept, we have processed 17 years of twice-daily MODIS land surface temperatures (LSTs), covering all of Antarctica at a 1 km resolution. The "condensed" dataset provides interesting insights into the LSTs of Antarctica and the quality of the underlying data.