Virgo Logbook

Apart from the usual far-field projections, it is also possible to make a magnetic projection based on the coherence with the magnetometers . You could say that in this case, you use the background environmental field as the noise injection. The only problem is that if a noise does not contribute a lot to a signal, it takes a lot of averages to get a meaningful value of the coherence. But since we have plenty of data, there is no excuse not to use this brute force method: I took 7 hours of data from October 4th, when the horizon was around 9.5 Mpc. This is enough to calculate a coherence using more than 5000 averages.

Fig 1 shows the obtained coherences between B1_ACp and the 3 magnetometers in the central building. Looking at the calculated coherence at high frequency, which should be zero, you can say that any bin with a coherence above 1e-3 is probably real, which means that you can retrieve signals which have an amplitude with a factor of 30 below the sensitivity. Since the noise measured by the 3 magnetometers is not indipendent, the maximum of the 3 coherences is used instead of the incoherent sum.

Fig 2 shows a noise projection, which is calculated by multiplying the sensitivity with the square root of the coherence. At low frequencies, this result is about a factor 4 or 5 more optimistic than the one obtained by the far-field projections, which show the contribution of the central building close to Virgo design. At bump around the 150 Hz line, however, is higher than predicted before.

Fig 3 shows how the sensitivity would improve if we could remove the contribution of all the magnetic noise observed by the magnetometers. This is calculated by multiplying the 'dirty sensitivity' by sqrt(1 - coherence).

One problem with these really long coherences might be the influence of glitches. I tried to do the same calculation, but now vetoing data sections using BRMSMon_MAB_CE_10_40 and _16_64. This difference seems neglectible, but this has to be understood better.