87th AMS Annual Meeting

Monday, 15 January 2007
Java libraries for reading GRIB and BUFR or you can't always get what you want but if you try sometimes you get what you need
Exhibit Hall C (Henry B. Gonzalez Convention Center)
Robert P. Kambic, Unidata/UCAR, Boulder, CO; and J. Caron, E. Davis, and Y. Ho
These are read-not-write libraries written in 100% Java and open source. The libraries main requirement is to access the desired data without reading the whole dataset. They satisfy the requirement by exploiting the internal structure of the data format and by adding auxiliary information about the location of the datasets in the file.

The libraries solve the problem of decoding large GRIB and BUFR datasets that can expand 10 to 50 times in size and the raw data file size could be 500 megabytes. In the extreme case, the resulting decoded dataset would be 25 gigabytes if the libraries were not used.

Another benefit of using the libraries, a table of contents of the dataset parameters are known before the file is opened for data extraction.

The libraries provide quick data access with minimal overhead to the data extraction process. Sounds to good to be true, stop by my poster session to get the inside scoop on fast GRIB and BUFR data access.

Supplementary URL: