Correlated magnetic noise in global networks of gravitational-wave detectors: Observations and implications

One of the most ambitious goals of gravitational-wave astronomy is to observe the stochastic gravitational-wave background. Correlated noise in two or more detectors can introduce a systematic error, which limits the sensitivity of stochastic searches. We report on measurements of correlated magnetic noise from Schumann resonances at the widely separated LIGO and Virgo detectors. We investigate the effect of this noise on a global network of gravitational-wave detectors and derive a constraint on the allowable coupling of environmental magnetic fields to test mass motion in gravitational-wave detectors. We find that while correlated noise from global electromagnetic fields could be safely ignored for initial LIGO stochastic searches, it could severely impact Advanced LIGO, Advanced Virgo, KAGRA, as well as third-generation detectors.

Figure 1

Magnetometer coherence spectra for LHO-LLO during the LIGO S5 science run (top, $t_{\mathrm{obs}}=330\mathrm{dy}$) and for LHO-Virgo during S6-VSR2/3 (bottom, $t_{\mathrm{obs}}=100\mathrm{dy}$). Schumann resonance peaks are indicated with black circles while electronic noise lines are indicated with green diamonds. The red dashed line indicates the average value expected for uncorrelated noise. Some LHO-LLO peaks are obscured by 60 Hz electronic noise. The frequency resolution is 0.1 Hz.

Figure 2

Magnetic cross-amplitude spectra $\sqrt{|M_{12}(f)|}$ during the S5 and S6-VSR2/3 LIGO-Virgo science runs. Note that Fig. 1, 2 are related through Eq. (8). The frequency resolution is 0.1 Hz. The S5 HL A and S6 HL traces utilize the same magnetometer pair, whereas the S5 HL B trace comes from a magnetometer pair only available during S5.

Figure 3

Strain amplitude spectra for correlated and uncorrelated noise. Black is the uncorrelated noise $\sqrt{N_{12}^{\mathrm{u}}(f)}$ for the H1L1 detector pair operating at Advanced LIGO design sensitivity and assuming 1 yr of integration. Purple indicates the uncorrelated noise $\sqrt{N_{12}^{\mathrm{u}}(f)}$ achieved during initial LIGO using $\approx 300$ days of coincident data. Red is $\sqrt{N_{12}^{\mathrm{m}}(f)}$ (the estimated correlated noise due to EM fields during initial LIGO). Electronic noise lines have been notched. The spectra have been scaled to assume a frequency resolution of 0.25 Hz, which is typical for stochastic searches [26].

Figure 4

The design-sensitivity Advanced Virgo and aLIGO critical coupling function $T(f)$ above which correlated noise is comparable to uncorrelated noise. The critical coupling function is calculated by considering the sensitivity obtained from a given pair of detectors, each assumed to have the same coupling function. Electronic noise lines have been notched. We also include the critical coupling function (solid black) for initial LIGO. The measured initial LIGO coupling function (dashed black) falls below it.