Science Data Challenge 3

Our 'Foregrounds' challenge asks participants to remove obscuring sources of emission which prevent analysis of the underlying hydrogen-21cm signal from the Epoch of Reionisation. This foreground emission stems from both Galactic and extragalactic sources, both of which have previously observed, and unobserved components.

Given the lack of a model for the finer structure of Galactic emission at SKA-LOW frequencies, the removal of Galactic emission from the dataset represents a significant challenge. By similar reasoning, source confusion from previously unknown extragalactic sources, especially at the coarser resolution at metre-wavelengths, complicates the matter further.

From our synthetic datasets, participants are asked to extract the cylindrically-averaged power spectrum of the EoR signal, clean from foregrounds contamination. 

To assess resulting submissions, our scoring ('figure-of-merit') algorithms will take resulting cylindrical power spectra, and return a score. Ancillary analytical data products can be assessed, however, the cylindrical power spectra results will be the only ones that affect the 'leaderboard'. Given that this challenge runs over 6 months, we expect this to be an iterative process and encourage competing teams to make full use of the computing facilities assigned to them.