LOFAR power spectrum recovery estimates

The plots on this page show estimates for errors on the 21cm power spectrum measured by LOFAR. The datacubes use the foreground simulations of Jelić et al. (2008). We conservatively assume that 300 hours of integration with one synthesized beam would give noise of 78mK for a 1MHz bandwidth at 150MHz; here we use 0.5MHz channels so the noise on each individual map is √2 times higher than this. The cosmological signal comes from the f250C simulation. All the power spectra shown are for maps convolved with the instrumental response. For these datacubes we model the uv coverage as being the same at all redshifts. Foreground subtraction uses Wp smoothing in the image cubes with λ=0.5. The original datacubes have 256×256×170 points, but these are binned up to 64×64×170 before foreground subtraction and power spectrum estimation. All the power spectra are measured in slices of the datacube encompassing 16 frequency bins, since this gives a depth comparable to the size of our current cosmological signal simulation and prevents there having been too much evolution. The power spectra are 'dimensionless' 3D power spectra (not really dimensionless, since they're measured in Kelvin rather than using the dimensionless quantity δ used in cosmology, but they do give the variance per log wavenumber as usual).

On this page we examine the trade-off between increasing the integration time on a single window and increasing the number of windows. This has required separating the error into a contribution from the signal power and from the noise power; it's not just a variance of the variance estimate as we used here. I still don't think the error estimates are completely correct; I'm working on a better way.

1 beam

Low redshift Mid redshift High redshift
900 hrs, 1 window TIFF TIFF TIFF
300 hrs, 3 windows TIFF TIFF TIFF
180 hrs, 5 windows TIFF TIFF TIFF



6 beams

Low redshift Mid redshift High redshift
900 hrs, 1 window TIFF TIFF TIFF
300 hrs, 3 windows TIFF TIFF TIFF
180 hrs, 5 windows TIFF TIFF TIFF