CASA Imaging Issues Affecting Some ALMA Products
Over the course of the last several months an issue in CASA mosaic imaging was recognized. The Observatory elected to understand and characterize the issue before informing the user community of potential impacts on the resulting images. The CASA team has now identified a couple of issues using both the 7-m and 12-m arrays over the course of the past few cycles and CASA releases. The full extent of the data affected is still being investigated.
The two issues identified by the team with mosaic imaging are:
1. Mosaicking observations with the 7-m array in Cycles 1-4 (range of CASA versions) were all affected. The Observatory will reimage the maps in the archive for those projects and we will inform users of the timeline of that campaign. The impact of this error is the fluxes are overestimated by ~10% throughout the map. This specific issue was resolved in Cycle 5 (CASA version 5.1.1) and beyond.
2. A different and more complex issue arose in the Cycle 5 software (CASA version 5.1.1 to 5.3, all versions), which affects all mosaics in Cycle 5. While there are a variety of factors that contribute to the range of impacts, the most important parameter is the aspect ratio of the mosaic pointing distribution, i.e., the width/length ratio of the pointing centers (not the image size). For projects with aspect ratios <1.5, the impact on fluxes throughout the bulk of the map is small, with large deviations in flux (>5%) largely confined to the edges of the map where the mosaic .pb file reaches values of <0.5. If your science does not depend sensitively on fluxes measured at the edges of the mosaic and your aspect ratio is <1.5 the impact is negligible. If your aspect ratio is ≥1.5, fluxes in the bulk of the map can be affected. This issue has been resolved in the Cycle 6 Pipeline (CASA version 5.4.0 or beyond). The Observatory will determine the aspect ratios over the coming weeks and contact PIs regarding corrective action timelines.
The Observatory regrets any inconvenience experienced by our ALMA users due to these issues. It is recommended that mosaic images are regenerated using the latest CASA release, CASA 5.4 (available at https://casa.nrao.edu/). In addition, the Observatory will contact PIs further when aspect ratios are determined and follow up actions (e.g. QA3) are ready. The actions that the Observatory is undertaking include:
1) While we determine aspect ratios, all MOUSs with mosaic observations will have their proprietary periods paused.
2) Once aspect ratios are determined, all project MOUSs still in proprietary period subject to issue 1, generically, or issue 2, for aspect ratios greater or equal to 1.5, will have their proprietary time paused retroactively to July 2018. Projects subject to issue 2 with smaller aspect ratios will have proprietary period paused while it is determined if the PI needs to recover fluxes in the outer regions of the map with high accuracy. The proprietary time will be extended by the amount from July 2018 until the new data products are available to the PI.
Additional supporting documentation is posted on NAASC Memo 117.
If you have any questions about this issue and the steps taken by the Observatory, please contact your local ARC for Support through the ALMA Helpdesk at https://help.almascience.org.
Update on 2020 September 24:
The product images in the downloadable datasets are free from this issue. The datasets which include images that have not been re-imaged yet by the Observatory remain to be in the QA3 status, and users cannot download them from the ALMA Science Archive.
Please note that CASA imaging issues affect only images and do not affect the visibility calibration part. Please see details in the Knowledgebase article;.