Source-linked AI summary
CASA, the Common Astronomy Software Applications for Radio Astronomy
THE CASA TEAM, Ben Bean, Sanjay Bhatnagar, Sandra Castro, Jennifer Donovan Meyer, Bjorn Emonts, Enrique Garcia, Robert Garwood, Kumar Golap, Justo Gonzalez Villalba, Pamela Harris, Yohei Hayashi, Josh Hoskins, Mingyu Hsieh, Preshanth Jagannathan, Wataru Kawasaki, Aard Keimpema, Mark Kettenis, Jorge Lopez, Joshua Marvil, Joseph Masters, Andrew McNichols, David Mehringer, Renaud Miel, George Moellenbrock, Federico Montesino, Takeshi Nakazato, Juergen Ott, Dirk Petry, Martin Pokorny, Ryan Raba, Urvashi Rau, Darrell Schiebel, Neal Schweighart, Srikrishna Sekhar, Kazuhiko Shimada, Des Small, Jan-Willem Steeb, Kanako Sugimoto, Ville Suoranta, Takahiro Tsutsumi, Ilse M. van Bemmel, Marjolein Verkouter, Akeem Wells, Wei Xiong, Arpad Szomoru, Morgan Griffith, Brian Glendenning, Jeff Kern
TL;DR
Radio astronomy requires computational software for processing diverse telescope data, motivating a broad and maintainable CASA framework. The paper surveys CASA’s architecture, data model, processing procedures, documentation, development, and next-generation infrastructure. CASA supports calibration, imaging, and analysis across single-dish and interferometric data, while its evolving infrastructure targets increasing data rates and scalability.
Problem
Radio astronomy needs software that supports calibration, imaging, and analysis across diverse telescope data and increasingly demanding processing environments.
Method
The paper provides a high-level overview of CASA’s structure, data model, APIs, processing stages, documentation, development process, and next-generation design.
Results
CASA provides open-source functionality for calibrating, imaging, and analyzing single-dish and interferometric data, including pipeline support for ALMA, VLA, and VLASS.
Takeaways & Limitations
CASA serves as a leading radio-data processing package while developing modernized infrastructure for the increasing complexity and scale of current and next-generation telescope data.
Takeaways & Limitations
Recently observed data may require an up-to-date CASA data repository because runtime data are updated more frequently than CASA releases.
Abstract
from arXiv · showhide
CASA, the Common Astronomy Software Applications, is the primary data processing software for the Atacama Large Millimeter/submillimeter Array (ALMA) and the Karl G. Jansky Very Large Array (VLA), and is frequently used also for other radio telescopes. The CASA software can handle data from single-dish, aperture-synthesis, and Very Long Baseline Interferometery (VLBI) telescopes. One of its core functionalities is to support the calibration and imaging pipelines for ALMA, VLA, VLA Sky Survey (VLASS), and the Nobeyama 45m telescope. This paper presents a high-level overview of the basic structure of the CASA software, as well as procedures for calibrating and imaging astronomical radio data in CASA. CASA is being developed by an international consortium of scientists and software engineers based at the National Radio Astronomical Observatory (NRAO), the European Southern Observatory (ESO), the National Astronomical Observatory of Japan (NAOJ), and the Joint Institute for VLBI European Research Infrastructure Consortium (JIV-ERIC), under the guidance of NRAO.
1. INTRODUCTION
CASA is an open-source software package for calibrating, imaging, and analyzing radio-telescope data. This paper introduces its structure, processing stages, pipeline support, development, and planned next-generation design.
- CASA enables calibration, imaging, and analysis of data produced by major radio telescopes.
- The package processes both single-dish and radio interferometric data through applications implemented in C++ and accessed via Python.
- The paper offers a high-level overview intended to familiarize readers with CASA’s principles and philosophy rather than provide complete functionality or radio-processing coverage.
- It surveys CASA’s software structure, data model, API, documentation, data-processing stages, development process, and tentative next-generation design.
- CASA supports calibration and imaging pipelines for ALMA, VLA, and VLASS.
2. THE CASA SOFTWARE
CASA combines a layered C++ and Python architecture with MeasurementSet-based data handling, user-facing tasks and tools, documentation, parallel processing, and externally maintained runtime data.
- 2.1. CASA structure and Python: CASA is a monolithic application in which Python tasks and tools interface with a C++ application layer and Casacore infrastructure.
- 2.2. The MeasurementSet and Data Model: The MeasurementSet stores interferometric or single-dish data in a Main table and metadata subtables.
- 2.2. The MeasurementSet and Data Model: Calibration preserves original data in DATA or FLOAT_DATA while storing calibrated values in CORRECTED_DATA; MODEL_DATA can store source-model visibilities.
- 2.3.2. CASA tasks, tools, and GUIs: CASA tasks provide higher-level, parameterized functions for common processing workflows, while tools expose separate lower-level operations and support scripting and pipelines.
- 2.3.2. CASA tasks, tools, and GUIs: CASA includes GUIs for inspecting visibility, calibration, image, and metadata products, including plotms and the CASA viewer.
- 2.3.3. External data repository: Recently observed data may require updating the CASA data repository because runtime data are updated more frequently than CASA releases.
- 2.4. Parallelization: CASA parallelizes tasks across multiple cores and computing nodes, while imaging partitions continuum data by time and spectral cubes by frequency.
3. DATA PROCESSING
CASA supports an end-to-end radio data-processing workflow spanning import, examination and flagging, calibration, manipulation, imaging, visualization, analysis, and simulation. Its calibration framework models and corrects multiple instrumental and environmental effects through Measurement Equation-based visibility corrections.
- CASA’s typical interferometric workflow covers data import/export, examination and flagging, calibration, manipulation, imaging, visualization, analysis, and simulations.The workflow is presented as a sequence of processing stages for radio interferometric data.
- ALMA and VLA data are imported from ASDM into CASA MeasurementSet format using importasdm before processing.CASA also directly imports some telescope data and UVFITS or FITS-IDI radio astronomy data.
- Flagging excludes compromised observations from calibration and scientific analysis without deleting the original data from the MeasurementSet.Flags can target antennas, baselines, spectral windows, shadowing limits, or automatically detected anomalies, and flag versions can be backed up and restored.
- CASA calibration corrects instrumental and environmental propagation effects and transforms measurements into absolute standard units using complex correction factors.Solutions are typically derived from well-characterized calibrators and can sometimes be improved through self-calibration on a sufficiently bright target.
- The Measurement Equation represents observed visibilities as true visibilities modified by accumulated baseline and antenna-based corruption operators.The model includes effects from the atmosphere, antennas, electronics, correlator, and downstream signal processing.
- Calibration terms generally follow the signal’s corruption order and cannot be arbitrarily reordered, while selected terms may be omitted according to scientific goals and dynamic-range requirements.The recognized terms include baseline-based errors, antenna-based factors, ionospheric effects, tropospheric effects, and parallactic-angle rotation.
3.4. Data processing IV: Data manipulation
CASA provides tasks for manipulating visibility data and supports the transformation of sky brightness and instrumental response into a dirty image. These operations prepare data for subsequent imaging and reconstruction.
- CASA tasks concatenate, average, split, weight, and regrid visibility data, and also support smoothing, continuum subtraction, model subtraction, position correction, and phase-center shifting.Examples include concat, split, statwt, cvel, hanningsmooth, uvcontsub, uvsub, fixplanets, and phaseshift.
- The figure’s components are labeled Point Spread Function, Sky Brightness Distribution, and Dirty Image.
- The dirty image results from convolving the sky brightness distribution with the instrumental point-spread function.Figure 3 illustrates this relationship using simulated VLA data.
3.5. Data processing V: Imaging & Deconvolution
CASA imaging reconstructs sky brightness from incompletely sampled calibrated visibilities through weighting, Fourier-domain processing, iterative deconvolution, and configurable imaging modes. Weighting and gridding choices trade resolution, sensitivity, noise, and computational cost.
- Interferometric imaging: Interferometric imaging solves for sky-brightness parameters from calibrated visibilities using a measurement operator combining spatial-frequency sampling and a two-dimensional Fourier transform.The operator is generally represented as [Sdd][F], with [Sdd] including direction-dependent instrumental effects.
- Interferometric imaging: Because finite antenna arrays leave unsampled regions of the (u,v)-plane, the imaging system is non-invertible and must be solved iteratively with model and instrumental constraints.
- Interferometric imaging: CASA forms a dirty image from calibrated visibilities, then iteratively cleans it through χ2 minimization to estimate the true sky model and construct the final image.The dirty image contains the sampled sky information convolved with the instrument PSF and affected by noise.
- Weighting: Natural weighting minimizes noise but often reduces resolution, whereas uniform weighting increases resolution with higher noise; Briggs weighting provides a resolution–sensitivity trade-off.A Gaussian uvtaper further downweights longer baselines, decreasing resolution and increasing surface-brightness sensitivity.
- Imaging pipeline: Imaging includes convolutional resampling onto a regular (u,v)-grid, fast Fourier transformation, and normalization before further reconstruction.
- Imaging modes: CASA supports continuum, spectral-cube, polarization, mosaic, and multi-term wideband imaging, with advanced gridders for direction-dependent and widefield or wideband effects.More sophisticated gridder modes can improve suitability for certain data sets but are more computationally intensive.
- Deconvolution and image reconstruction: The minor cycle separates sky emission from the PSF and builds the sky model using delta-function, maximum-entropy, or multiscale deconvolution algorithms.The multiscale algorithm assigns model components at different spatial scales and is useful for images containing extended emission.
3.6. Data processing VI: Analysis & Visualization
CASA supplies graphical tools for inspecting radio-astronomical data, metadata, visibility measurements, and image products. CARTA extends image visualization and analysis for ALMA, VLA, and SKA pathfinder data.
- CASA GUIs support inspection of raw telescope data and metadata, including CASA tables and antenna positions, and visualization of visibility data.Examples include browsetable, plotants, and plotms.
- CARTA is an external image-visualization and analysis tool designed for ALMA, the VLA, and pathfinder telescopes for the Square Kilometre Array.
3.7. Data processing VII: Simulations
CASA supports simulation workflows for interferometric and total-power observations, helping users assess telescope capabilities and benchmark software performance. Its simulation tools also support calibration-effect modeling and streamlined ALMA simulations.
- CASA simulations help users understand telescope capabilities and expected outputs while providing benchmarks for software performance, optimization, and reproducibility.
- simobserve creates simulated MeasurementSets for interferometric or total-power observations with a specified telescope.
- simanalyze images and analyzes simulated MeasurementSet data, including diagnostic images and plots.
- simalma combines simobserve and simanalyze to simulate ALMA observations across 12-m, 7-m ACA, and total-power measurements.
- CASA simulator tools extend beyond ALMA use cases and can calculate and apply calibration tables representing atmospheric and instrumental effects.
3.8. Very Long Baseline Interferometry
CASA processes VLBI data from arrays including the VLBA and EVN using dedicated tasks for calibration and imaging. External development has extended CASA into a generic VLBI pipeline used for Event Horizon Telescope data.
- CASA processes VLBI data from the VLBA and EVN using the VLBI-specific fringefit and accor tasks.
- fringefit determines phase, delay, delay-rate, and optionally dispersive-delay solutions as functions of time and spectral window.
- The rPICARD pipeline, built on CASA, formed a critical component in processing Event Horizon Telescope data that produced the first black-hole image.
3.9. Algorithm Research & Development
CASA algorithms are continually improved as scientific use cases and operational needs evolve. Planned work includes advances in deconvolution, wideband imaging, automated flagging, GPU acceleration, and user interfaces.
- CASA continually improves its features and algorithms in response to evolving use cases and needs.
- Planned algorithm developments include wideband Adaptive Scale Pixel deconvolution and Full Mueller imaging for wide bandwidths.
- Further planned developments include automated flagging based on binned (u, v)-data and GPU implementation of widefield A-Projection.
- The CASA team is developing GUIs with improved Python integration and usability in Notebook environments.
4. ALMA AND VLA PIPELINES
CASA underpins calibration and imaging pipelines for ALMA, VLA, and VLASS, supporting both automated and customized processing. It also supports Nobeyama 45-m single-dish processing and custom external pipelines.
- ALMA calibration and imaging pipelines are built on CASA and support automated and customized processing of ALMA data.
- NRAO maintains different CASA pipeline versions for processing VLA and VLA Sky Survey data.
- CASA supports processing single-dish data from the Nobeyama 45-m telescope observed in On-The-Fly mode.
- CASA’s flexibility permits custom-built tasks and external pipelines, including VLBI pipelines.
5. DEVELOPMENT PROCESS
CASA development balances stakeholder priorities with maintenance and evolution through a half-yearly planning process. Features proceed from requirements through verification, validation, and targeted characterization.
- Stakeholder priorities are balanced with software maintenance and evolution to prioritize CASA work on a half-yearly timescale.
- Each feature begins with defined requirements and specifications before development and ends with internal verification and external validation.
- Verification combines functional, unit, stakeholder, performance, and benchmark tests, while validation evaluates applicability to stakeholder use-cases.
- Numerical and algorithmic features may receive additional characterization to assess implementation effectiveness for intended use-cases.
6. NEXT-GENERATION CASA
Next-generation radio telescopes will produce sharply increasing data rates and volumes, requiring processing software that is efficient and scalable. CASA has begun designing a next-generation system on new infrastructure.
- Next-generation facilities such as ngVLA and the ALMA wideband sensitivity upgrade are expected to generate massive increases in data rates and volumes.
- Future data-processing software needs to be efficient and easily scalable to large computing environments.
- A design phase has started for next-generation CASA, built on top of CASA’s next-generation infrastructure.
7. CONCLUSIONS
CASA is presented as versatile, pipeline-oriented software for processing data from major and other radio telescopes. Its stakeholder-based development and modernization initiative target increasingly complex, high-volume datasets.
- CASA supports calibration, imaging, and analysis of data from ALMA, the VLA, and other radio telescopes.
- CASA aims to maintain full functionality for aperture-synthesis arrays and single-dish radio telescopes.
- Comprehensive documentation and stakeholder-based development support CASA’s role as a leading software system for radio-data processing.
- CASA is developed by an international consortium based at NRAO, ESO, NAOJ, and JIV-ERIC under NRAO guidance.
- The software ecosystem includes CASA, Casacore, CARTA, AIPS, AIPS++, Python, IPython, C++, and Fortran.