Faraday depth spectra for Scenario 2 (Faraday complex) with 20% (top), 30% (middle) and 40% (bottom) of channel data flagged and SNR chan = 5 at a reference frequency of 1.4 GHz. Left: Faraday depth spectrum from flagged data. Centre: Faraday depth spectra reconstructed from the GP estimate as described in § 4.4. Right: Faraday depth residuals formed by subtracting the noiseless model from the GP estimate. The ±5 í µí¼Ž í µí¼™ limits are shown as a light grey shaded area.

Contexts in source publication

Context 1

... this case, recovering the Faraday depth of the thin component is not as strong an indicator of performance as in Scenario 1. Instead we compare the posterior mean of the optimized GP to the input model without noise and the residuals for this comparison are shown in the lower left hand panel of Figure 4. Figure 5 shows the Faraday depth spectrum reconstructed from the optimized GP for Scenario 2 in the presence of different levels of RFI flagging. These should be compared to the noiseless model spectrum shown in Figure 1. ...

Context 2

... can be seen in Figure 5, using an optimized GP to reconstruct the Faraday depth spectrum in the case where Faraday thick structure is present can reveal these structures without the need for assuming any specific parametric model of Faraday depth a priori. Although it is not possible to recover the parameterised geometry of the thick structure from the hyper-parameters of the GP in this case, the shape is far more clearly delineated in the GP-reconstructed Faraday depth spectrum than in the spectrum calculated from the original data. ...