Compression of the global Advanced Very High Resolution Radiometer 1-km data set

The cost to acquire, process, and use these data necessitates the lossless compression of data to store and distribute multispectral images. The size of the data sets also makes it difficult for users to access the data; users need to have quick access to geographic subwindows of the data and need to view the data at multiple resolutions. The amount of data in the 10-band global Advanced Very High Resolution Radiometer (AVHRR) 1-km image, which is produced every 10 days, is approximately 10.5 gigabytes. A method of lossless compression was developed that provides multiresolution decompression within geographic subwindows of multispectral global AVHRR 1-km images. The compression algorithm segments the image into blocks and compresses each block in a hierarchical format. Users can access the data by specifying a geographic subwindow or the whole image and a resolution (1, 2, 4, 8, or 16 km). The time required to decompress the subwindow is proportional to the resolution selected. An image area of water, for example, and the unused parts of the framed data in a map projection are masked. A quad tree algorithm is used to compress these masked areas. Because the masked areas are the same in all 10 bands, the compressed mask is stored only once, saving space. Preliminary results using this compression method indicate that these 10-day composite data sets are compressed to about 15 percent of the original size-a compression ratio of more than 6:1