In the period between the austral summers of 1982/83 to 1997/98, iceberg observations were recorded by most research vessels cruising Antarctic waters, under an international programme for systematic collection of Antarctic iceberg data initiated by The Norwegian Polar Institute (NPI) in 1981 with the endorsement of the Scientific Committee on Antarctic Research (SCAR). The resulting database contains records of 323 520 iceberg positions from 26 634 individual observations. Of these, 262 007 icebergs have been classified by size into five different length categories: 10-50, 50-200, 200-500, 500-1000 and ˃1000 m. The database also includes 8061 additional iceberg observations collected by the Australian National Antarctic Research Expeditions ships from 1984 (AAD), operating in the 50-150° E longitude sector. Some of these have been classified by slightly different size categories. For this reason, the complete database has been packaged in three separate csv files:

- one file containing the entire NPI database, including AAD observations 1978-1984

- one file containing the 8061 additional AAD observations 1984-2010

- one file containing 970 AAD observations 1978-1984, also included in the NPI data file

The dataset is described in detail in the attached data report, with an overview of the contents of the database and their potential use, the data quality and quality control process, and some concluding remarks on iceberg occurrence and distribution around Antarctica; drift patterns, dissolution rates, calving rates and their contribution to the mass balance of Antarctica.

During the three decades of observations, most of the ocean around Antarctica has been observed, but there are large differences in data density, primarily because nearly all ship tracks follow repeated routes to the various research stations, and these are not evenly located around the continent.


Figure 1: The 34 695 individual observations in the SCAR database shown as blue dots for the NPI data collection scheme with 5 classes and red for the AAD data with 7 classes (described in subsequent sections). The red dots overprint and hide many blue dots.