Data Challenges

The SKAO challenges dataset will consist of real data from currently operating radio facilities, and of simulated SKA data. Data at different stages along the data reduction pipeline have been broadly categorised into four main Data Layers (DL)

• DL1: raw visibility data. These will be typically a few hours observations, consistent with a single SKA scheduling block. Data will be typically uncalibrated and the main focus of the challenges will be calibration strategy and implementation, efficiency and scalability.
• DL2: calibrated products. These products will mimic what the SDP will typically provide: calibrated data, from a list of standard products, corresponding to a few hours observations, consistent with a single SKA scheduling block. The focus of challenge exercises will be to carry out the kind of processing that SRCs will ultimately do, to provide the PI/KSP teams with science-ready data products.
• DL3: science ready products. These products will mimic the typical SRC output. It will be a standard product with long integration time, consistent with having combined several SKA scheduling blocks, over a wide area and/or a deep integration. The objective of data challenges will be to extract science from the data, with a focus on algorithm development.
• DL4: scientific results. This is a proposal-specific product, that’s ultimately the goal of the whole observation and analysis.

SKAO challenges data can be made available at any of the four stages. Objectives for each data challenge will be specified, however usage of the data will be unrestricted.

The SKAO challenges datasets available are:

• Science Data Challenge #1 STATUS: challenge closed , dataset and scoring code available
• Science Data Challenge #2 STATUS: challenge closed, dataset and scoring code available
• Science Data Challenge #3 STATUS: SDC3 EoR Inference challenge ongoing. Challenge ends 30/4/25