Monday, 15 January 2007
Evaluating optimal techniques for compressing NOGAPS data sets for quick response applications
Exhibit Hall C (Henry B. Gonzalez Convention Center)
The NOGAPS (US Navy Operational Global Atmospheric Prediction System) data sets have been used extensively for data quality control, data assimilation, and model initialization. NOGAPS can provide valuable forcing fields for quick response applications such as severe weather prediction and hazardous aircraft/ship warnings. In these types of applications, timely and expeditious transmission of large NOGAPS data is of utmost importance. Lossless compression algorithms can reduce the volume of data by 2 to 4 times but to achieve an order of magnitude better compression performance, lossy compression must be employed. Over the years, compression techniques (JPEG 2000 part I and II) have made significant progress. These compression algorithms can achieve very high compression ratios if certain losses in terms of resolution are acceptable for specific applications. In the atmospheric sciences, significant events could be lost during the compression and decompression processes. The “CompressMD” analysis tool developed by the Computer Engineering Department at the University of Texas at El Paso is used to evaluate various compression techniques appropriate for compressing the NOGAPS data set. A realistic case study is conducted in which a representative compressed/decompressed NOGAPS data set is used to initialize a mesoscale model. Tolerable RMS errors and maximum allowable errors for each meteorological parameter of interest in the data set had been pre-selected. Evaluation of the mesoscale model's performance using the compressed NOGAPS data set is discussed and special features of the analysis tool are highlighted.