IGEC2: A 17-month search for gravitational wave bursts in 2005–2007

We present here the results of a 515 day search for short bursts of gravitational waves by the IGEC2 observatory. This network included 4 cryogenic resonant-bar detectors: AURIGA, EXPLORER, and NAUTILUS in Europe, and ALLEGRO in America. These results cover the time period from November 6th 2005 until April 15th 2007, partly overlapping the first long term observations by the LIGO interferometric detectors. The observatory operated with high duty cycle, namely, 57% for fourfold coincident observations, and 94% for threefold observations. The sensitivity was the best ever obtained by a bar network: we could detect, with an efficiency $>50\%$, impulsive events with a burst strain amplitude $h_{rss}\lesssim 1\times {10}^{-19}{\mathrm{Hz}}^{-1/2}$. The network data analysis was based on time coincidence searches over at least three detectors, used a blind search technique, and was tuned to achieve a false alarm rate of $1/\mathrm{\text{century}}$. When the blinding was removed, no gravitational wave candidate was found.

Figure 1

Typical strain noise spectral densities (single-sided) of IGEC2 detectors. All detectors are sensitive in a region around 900 Hz, and the minimum level of noise is comparable in all spectra. The wider bandwidth of AURIGA includes the bandwidths of the other detectors. The noise curve of EXPLORER (blue line) refers to the operation after the 2006 retuning.

Figure 2

Distribution of the fourfold observation time in the hours of the day: sidereal (solid line, red) and solar (dotted line, blue).

Figure 3

Amplitude distribution of the exchanged candidate events above the minimal thresholds (AURIGA $\mathrm{SNR}>4.5$, ALLEGRO $H>4\times {10}^{-21}{\mathrm{Hz}}^{-1}$, EXPLORER and NAUTILUS $\mathrm{SNR}>4.0$).

Figure 4

Offsets in the event time differences between EXPLORER and NAUTILUS for damped sinusoids with $\tau =1$, 3, 10, and 30 ms. This systematic error varies between $-15$ and $+10\mathrm{ms}$.

Figure 5

Cumulative number of coincidence found in EXPLORER-NAUTILUS vs the absolute value of the event time difference, obtained by subtracting the theoretical cumulative of accidentals from the cumulative of events found in proximity of the injection times. The common excitations are software injections of damped sine waves with $f=914\mathrm{Hz}$, $\tau =1\mathrm{ms}$. About 95% of the events with $4.2\le \mathrm{SNR}\le 4.6$ are retrieved with a coincidence window of $\pm 25\mathrm{ms}$.

Figure 6

Responses of detectors to a damped sinusoid ($h_{rss}=1\times {10}^{-18}{\mathrm{Hz}}^{-1/2}$, $\tau =30\mathrm{ms}$). The plot shows the amplitude response of Explorer and of Nautilus vs the wave center frequency. Markers indicate values measured via software injections, while the lines show the behavior predicted by an analytical model.

Figure 7

Responses of detectors to damped sinusoids with various duration ($\tau =1$, 3, 10, 30 ms). The plot shows the ratio of the responses (Amplitude EXPLORER/Amplitude NAUTILUS) vs the wave center frequency. Markers and lines as in Fig. 6.

Figure 8

Contour lines of the accidental rate (events/century) estimates in the AU-EX-NA (data stretch B) case.