Investigating the nature of INTEGRAL Gamma-ray Bursts and sub-threshold triggers with Swift follow up

TL;DR

This study investigates weak INTEGRAL triggers for gamma-ray bursts, confirming some as real GRBs through Swift follow-up, and compares their properties with Swift BAT GRBs, revealing similar low-fluence detection capabilities.

Contribution

It demonstrates that real GRBs can be identified among INTEGRAL's weak triggers and compares their properties with Swift BAT GRBs, highlighting similar low-fluence detection levels.

Findings

Six weak triggers classified as GRBs, including one discovered by the campaign.

INTEGRAL and Swift detect similar low-fluence GRBs, with Swift being more sensitive to short, low-fluence events.

Correlation between gamma-ray and X-ray properties supports previous findings.

Abstract

We explore the potential of INTEGRAL to improve our understanding of the low fluence regime for explosive transients, such as GRBs. We probe the nature of the so-called "WEAK" INTEGRAL triggers, when the gamma-ray instruments record intensity spikes that are below the usual STRONG significance thresholds. In a targeted Swift follow-up campaign, we observed 15 WEAK triggers. We find six of these can be classified as GRBs. This includes GRB150305A, a GRB discovered from our campaign alone. We also identified a source coincident with one trigger, IGRW151019, as a candidate AGN. We show that real events such as GRBs exist within the IBAS WEAK trigger population. A comparison of the fluence distributions of the full INTEGRAL IBAS and Swift BAT GRB samples showed that the two are similar. We also find correlations between the prompt gamma-ray and X-ray properties of the two samples, supporting previous investigations. We find that both satellites reach similar, low fluence levels regularly, although Swift is more sensitive to short, low fluence GRBs.

Tables

Table 1. Table 1: Table containing the properties of the 15 WEAK triggers. The four triggers at the bottom are the four candidate GRBs with previous XRT observations and were not part of our selected ToOs. The Trigger No., significance ( σ 𝜎 \sigma ), RA, Dec and localisation error (90 per cent confidence) were all taken from IBAS.

ToO
Name	INTEGRAL
	IBAS detection Significance ($\sigma$)	RA
(Deg)
	Dec	(J2000)
(Deg)
	Localisation
Error
	(arcmin)
IGRW 160610	7488/0	6.7	359.90	61.57	3.8
IGRW 151019	7277/0	7.0	292.82	31.14	3.5
IGRW 150903	7231/0	6.7	239.17	-33.81	3.6
IGRW 150610	7005/0	7.1	178.32	16.03	4.8
IGRW 150305	6905/0	7.6	269.79	-42.62	3.4
IGRW 140219	6467/0	6.7	204.10	-45.06	3.6
IGRW 130904	6931/0	6.7	256.88	-32.01	3.6
IGRW 110718	6323/0	6.8	256.78	40.05	3.6
IGRW 110608	6297/0	6.8	315.28	32.041	3.6
IGRW 110428	6169/0	7.2	320.27	-33.96	3.5
IGRW 110112	6127/0	7.4	10.56	64.41	2.6
IGRW 150831	7228/0	7.3	220.98	-25.65	3.4
IGRW 121212	6720/0	7.9	177.90	78.00	3.3
IGRW 100909	6060/0	7.7	73.95	54.65	2.0
IGRW 091111	-	7.2	137.81	-45.91	2.9

Table 2. Table 2: Swift XRT ToO observations and candidate counterpart source detections. XRT position (90 per cent confidence), RA and Dec were taken from UKSSDC. T START subscript 𝑇 START T_{\rm START} refers to the time elapsed between the GRB/trigger occurring and the time when the Swift observation began and T STOP subscript 𝑇 STOP T_{\rm STOP} refers to the time elapsed between the GRB occurring and the time when the final Swift observation finished. T EXP subscript 𝑇 EXP T_{\rm EXP} is the total XRT exposure time.

ToO
Name	Swift
	XRT
Position
Error
	RA
(Deg)
	Dec
(Deg)
	$T_{\mathrm{START}}$
	$T_{\mathrm{STOP}}$
	$T_{\mathrm{EXP}}$
	(s)
IGRW 151019	20558	2.5	292.7836	31.1319	9600	2810017	14938
GRB 150831A	653838	1.6	221.0243	-25.6351	82	38575	11828
IGRW 150305A	33663	3.5	269.7606	-42.6638	17838	735268	908
GRB 121212A	541371	1.4	177.7923	78.0371	60	145420	22940
GRB 100909A	20147	3.3	73.9488	54.6579	11693	25787	7720
GRB 091111	20120	7.7	137.8233	-45.9253	100360	197466	10386

Table 3. Table 3: Table containing the X-ray spectral and afterglow light curve properties of WEAK GRBs with > 10 absent 10 >10 binned data points. N H subscript 𝑁 H N_{\rm H} (Gal) is the fixed Galactic absorption column density and N H subscript 𝑁 H N_{\rm H} (Int) is the excess absorption. Spectral analysis was performed using xspec and fitting an absorbed power law where Γ Γ \Gamma is the photon index. The X-ray decay slopes were calculated using non-linear least squares fitting with various broken power law models. For each case a simple non-broken power law provided the best fit. All errors given at 90 per cent confidence level apart from the X-ray decay slopes - they are given at 1 ​ σ 1 𝜎 1\sigma .

ToO Name	$N_H$(Gal) (${10}^{20}$ cm-2)	$N_H$(Int) (${10}^{20}$ cm-2)	$\Gamma$	C-Stat (dof)	$\alpha$
GRB 121212A	4.48	${21}_{-0.4}^{+0.5}$	${2.24}_{-0.13}^{+0.14}$	341 (369)	$-{0.71}_{-0.03}^{+0.03}$
GRB 150831A (WT)	11.4	$0_{-0}^{+80.0}$	${1.15}_{-0.1}^{+0.18}$	322 (404)	$-{2.67}_{-0.22}^{+0.22}$
GRB 150831A (PC)	11.4	$0_{-0}^{+18.0}$	${1.53}_{-0.29}^{+0.28}$	99 (93)	$-{2.67}_{-0.22}^{+0.22}$

Table 4. Table 4: UVOT multi-band magnitudes (AB) and 3 ​ σ 3 𝜎 3\sigma upper limits of the 6 XRT detected sources. Filters ordered with decreasing wavelength. * * {}^{\textbf{*}} The GRB 091111 XRT position was within 10 arcsec of a very bright, saturated source and further analysis of the images suggest that the detections in the v, b and u filters are probably not real.

Name	white	v	b	u	w1	m2	w2	Source?	$A_V$ (Mag)
IGRW 151019	-	-	-	-	$>21.75$	22.16($\pm 0.15$)	-	Yes	0.75
GRB 150831A	$>21.54$	$>19.82$	$>19.90$	$>21.34$	$>21.85$	$>22.39$	$>23.10$	No	0.30
GRB 150305A	-	-	-	-	$>22.68$	$>22.74$	-	No	0.45
GRB 121212A	23.89($\pm 0.38$)	$>20.22$	$>20.75$	$>22.07$	$>23.00$	$>22.73$	$>22.95$	No	0.18
GRB 100909A	$>22.52$	$>20.07$	$>21.26$	$>21.78$	$>21.73$	$>22.19$	$>21.99$	No	1.37
GRB 091111	-	19.48($\pm 0.24$)${}^{\text{*}}$	18.92($\pm 0.28$)${}^{\text{*}}$	21.76($\pm 0.23$)${}^{\text{*}}$	$>21.94$	$>22.35$	$>22.32$	No	4.79

Name	Fluence [$20-200$ keV] (${10}^{-7}$ erg cm-2)	X-ray Flux at 11 hours [$0.3-10$ keV] (${10}^{-12}$ erg cm-2 s-1)	T90 (s)
GRB 030227${}^{\text{*}}$	${6.10}_{-5.90}^{+3.50}$	-	15
GRB 030320${}^{\text{*}}$	${54.2}_{-11.7}^{+13.3}$	-	48
GRB 030501${}^{\text{*}}$	${17.2}_{-3.10}^{+1.60}$	-	25
GRB 030529${}^{\text{\#}}$	0.52	-	16
GRB 031203${}^{\text{*}}$	${10.6}_{-3.00}^{+2.70}$	-	19
GRB 040106${}^{\text{*}}$	${95.0}_{-30.0}^{+23.0}$	-	47
GRB 040223${}^{\text{*}}$	${27.2}_{-1.90}^{+0.80}$	-	258
GRB 040323${}^{\text{*}}$	${20.6}_{-2.90}^{+2.30}$	-	14
GRB 040403${}^{\text{*}}$	${4.00}_{-3.70}^{+1.60}$	-	15
GRB 040422${}^{\text{*}}$	${4.90}_{-3.60}^{+1.00}$	-	10
GRB 040624${}^{\text{\#}}$	4.81	-	27
GRB 040730${}^{\text{*}}$	${6.30}_{-3.30}^{+4.40}$	-	42
GRB 040812${}^{\text{\#}}$	1.40	-	8
GRB 040827${}^{\text{*}}$	${11.1}_{-4.00}^{+2.80}$	-	32
GRB 040903${}^{\text{\#}}$	0.96	-	7
GRB 041015${}^{\text{\#}}$	5.12	-	30
GRB 041218${}^{\text{*}}$	${58.2}_{-3.70}^{+3.50}$	-	38
GRB 041219A${}^{\text{*}}$	${867}_{-129}^{+0.50}$	-	239
GRB 050129${}^{\text{\#}}$	4.10	-	30
GRB 050223	${10.8}_{-2.10}^{+2.70}$	0.19	30
GRB 050502A${}^{\text{*}}$	${13.9}_{-4.00}^{+1.10}$	-	$>11$
GRB 050504${}^{\text{*}}$	${10.0}_{-4.50}^{+4.10}$	-	44
GRB 050520${}^{\text{*}}$	${16.6}_{-5.00}^{+4.90}$	0.20	52
GRB 050522${}^{\text{\#}}$	0.69	-	11
GRB 050525A${}^{\text{*}}$	${154}_{-8.40}^{+5.70}$	1.5	9
GRB 050626${}^{\text{*}}$	${6.30}_{-1.00}^{+0.40}$	-	52
GRB 050714A	${5.58}_{-1.84}^{+2.75}$	-	34
GRB 050918${}^{\text{*}}$	${30.2}_{-9.0}^{+10.5}$	-	280
GRB 050922A${}^{\text{\#}}$	0.59	-	10
GRB 051105B${}^{\text{*}}$	${2.80}_{-2.00}^{+1.50}$	-	14
GRB 051211B${}^{\text{*}}$	${16.1}_{-3.30}^{+4.60}$	0.92	47
GRB 060114${}^{\text{*}}$	${16.0}_{-3.30}^{+4.60}$	-	80
GRB 060130${}^{\text{\#}}$	2.25	-	19
GRB 060204A${}^{\text{*}}$	${4.80}_{-3.30}^{+2.40}$	-	52
GRB 060428C${}^{\text{*}}$	${18.6}_{-3.90}^{+2.20}$	-	10
GRB 060901${}^{\text{*}}$	${62.2}_{-5.90}^{+3.50}$	1.2	16
GRB 060930${}^{\text{\#}}$	2.63	-	9
GRB 060912B${}^{\text{*}}$	${12.0}_{-5.10}^{+5.80}$	-	140
GRB 061025${}^{\text{*}}$	${10.1}_{-4.80}^{+1.30}$	0.14	11

Name	Fluence [$20-200$ keV] (${10}^{-7}$ erg cm-2)	X-ray Flux at 11 hours [$0.3-10$ keV] (${10}^{-12}$ erg cm-2 s-1)	T90 (s)
GRB 061122${}^{\text{*}}$	${155}_{-5.30}^{+3.40}$	2.2	12
GRB 070309	${4.93}_{-1.98}^{+3.12}$	-	22
GRB 070311${}^{\text{*}}$	${23.6}_{-5.30}^{+1.70}$	1.22	32
GRB 070615	2.01	-	15
GRB 070707	${3.58}_{-1.94}^{+4.04}$	-	0.7
GRB 070925${}^{\text{*}}$	${36.1}_{-3.40}^{+1.70}$	-	19
GRB 071003	${94.6}_{-2.96}^{+4.22}$	3.5	38
GRB 071109${}^{\text{*}}$	${3.60}_{-3.50}^{+4.00}$	-	30
GRB 080120	${13.2}_{-7.67}^{+17.0}$	0.13	15
GRB 080603A	${12.3}_{-5.90}^{+1.70}$	1.5	150
GRB 080613A${}^{\text{*}}$	${12.3}_{-5.90}^{+1.70}$	-	30
GRB 080723B${}^{\text{*}}$	${396}_{-6.70}^{+6.70}$	12.6	95
GRB 080922${}^{\text{*}}$	${17.3}_{-6.50}^{+6.90}$	-	60
GRB 081003B${}^{\text{*}}$	${26.2}_{-24.5}^{+2.00}$	-	20
GRB 081016${}^{\text{*}}$	${22.0}_{-4.50}^{+1.40}$	-	30
GRB 081204${}^{\text{*}}$	${5.10}_{-4.80}^{+5.10}$	-	12
GRB 090107B${}^{\text{*}}$	${12.4}_{-4.60}^{+1.30}$	0.73	15
GRB 090625B${}^{\text{*}}$	${12.4}_{-2.00}^{+1.20}$	0.38	8
GRB 090702	${1.93}_{-0.81}^{+1.44}$	-	6
GRB 090704${}^{\text{*}}$	${54.0}_{-8.00}^{+4.90}$	-	70
GRB 090814B${}^{\text{*}}$	${15.1}_{-2.40}^{+2.30}$	1.4	42
GRB 090817${}^{\text{*}}$	${18.7}_{-9.80}^{+10.9}$	2.4	30
GRB 091111	${20.0}_{-0.82}^{+5.90}$	-	100
GRB 091202	${7.03}_{-2.35}^{+3.02}$	-	25
GRB 091230	${17.9}_{-9.57}^{+20.5}$	-	70
GRB 100103A${}^{\text{*}}$	${52.5}_{-4.00}^{+2.10}$	2.1	30
GRB 100518A${}^{\text{*}}$	${5.20}_{-3.80}^{+4.40}$	0.87	25
GRB 100713A	${5.65}_{-1.80}^{+2.65}$	0.20	20
GRB 100909A	${21.5}_{-4.70}^{+7.00}$	0.26	60
GRB 101112A${}^{\text{*}}$	${21.1}_{-7.40}^{+4.40}$	0.50	6
GRB 110206A	${17.2}_{-6.10}^{+11.6}$	2.0	15
GRB 110708A${}^{\text{*}}$	${24.8}_{-4.60}^{+1.90}$	-	50
GRB 110903A${}^{\text{*}}$	${148}_{-17.5}^{+11.9}$	3.8	430
GRB 120202A	${8.00}_{-7.70}^{+2.10}$	-	70
GRB 120419A	${3.88}_{-2.49}^{+6.18}$	-	15
GRB 120711A	${440}_{-5.00}^{+50.0}$	40	135
GRB 121102A	${24.1}_{-8.10}^{+12.4}$	0.56	25
GRB 121212A	1.50	0.46	10

Table 7. Table 5: The prompt emission fluence, T 90 and X-ray afterglow flux of all 92 IBAS GRBs within our sample. All T 90 values were taken from IBAS 1 . Some GRBs were not observed by Swift or had very poorly sampled Swift XRT afterglows and therefore an X-ray flux at 11 hours could not be obtained. Some fluence values contain no errors as the fitted model would not converge and would not provide an error on the normalisation. This meant no error could be found on the flux values and therefore the fluence values but the values should still be representative of the actual fluence. Fluence values for GRB 150831, GRB 151120A, GRB 160221A, GRB 160629A are approximations from the GCN Circulars (Mereghetti et al., 2015a , b , 2016 ; Gotz et al., 2016 ) as the spectral data were not yet public. * * {}^{\textbf{*}} Fluence values taken from Bošnjak et al. ( 2014 ) . # # {}^{\textbf{\#}} Fluence values taken from Vianello et al. ( 2009 ) .

Name	Fluence [$20-200$ keV] (${10}^{-7}$ erg cm-2)	X-ray Flux at 11 hours [$0.3-10$ keV] (${10}^{-12}$ erg cm-2 s-1)	T90 (s)
GRB 130513A	${17.0}_{-6.50}^{+10.3}$	-	50
GRB 130514B	${10.2}_{-6.20}^{+14.4}$	1.6	10
GRB 130903A	${17.1}_{-5.40}^{+8.10}$	-	70
GRB 131122A	${24.8}_{-8.20}^{+12.3}$	-	80
GRB 140206A	${16.0}_{-3.00}^{+3.00}$	9.2	$>60$
GRB 140320B	${12.7}_{-5.94}^{+11.8}$	0.55	100
GRB 140320C	3.52	-	30
GRB 140815A	${5.00}_{-2.59}^{+5.10}$	-	8
GRB 141004A	${6.92}_{-3.40}^{+6.88}$	0.09	4
GRB 150219A	${57.1}_{-11.2}^{+14.9}$	0.62	60
GRB 150305A	${12.1}_{-6.45}^{+14.2}$	-	100
GRB 150831A	$\approx 3$	0.33	2
GRB 151120A	$\approx 20$	0.66	50
GRB 160221A	$\approx 5$	-	10
GRB 160629A	$\approx 60$	-	100

Full text

††pagerange: Investigating the nature of INTEGRAL Gamma-ray Bursts and sub-threshold triggers with Swift follow-up–5

1 Introduction

GRBs are among the most luminous events in the universe, releasing energies $>10^{51}$ erg typically in time periods of seconds (Gehrels & Mészáros, 2012). During these events a huge amount of gravitational energy is released from a central engine which leads to the formation of jets where particles are accelerated to ultra-relativistic speeds (Woosley & Heger, 2006). Internal shocks within the jet produce the high energy prompt gamma-ray emission we first observe (Gehrels & Mészáros, 2012; Piran, 2003). The jet then shocks with the surrounding medium producing broad-band afterglow emission (Mészáros & Rees, 1997; Wijers et al., 1997). Classically, GRBs are split into two sub-groups based on their T90 - the duration over which 90 per cent of the gamma-ray flux is received (Kouveliotou et al., 1993). The two groups are short GRBs where T${}_{90}<2$ s and long GRBs where T${}_{90}>2$ s, linked with two different progenitor models.

GRBs span a large range of isotropic equivalent luminosities: $10^{45}\leq{\rm L_{ISO}}\leq 10^{54}$ erg s*-1*. Investigations into their luminosity function and formation rates coupled with observations of several local GRBs (Sazonov et al., 2004; Soderberg et al., 2004) have suggested that there should be a large number of low-luminosity GRBs (Daigne & Mochkovitch, 2007; Liang et al., 2007; Pescalli et al., 2016). There are further suggestions that these could exist as a separate local population (Norris, 2002; Norris et al., 2005; Chapman et al., 2007; Liang et al., 2007). We use two currently active GRB detecting missions, Swift (Gehrels et al., 2004) and The INTErnational Gamma-Ray Astrophysics Laboratory (Winkler et al. 2003; INTEGRAL), to look at potentially faint GRBs.

INTEGRAL carries two gamma-ray instruments, IBIS (Ubertini et al., 2003) and SPI (Vedrenne et al., 2003). Alerts for GRBs and other transient sources are communicated with low latency by the INTEGRAL Burst Alert System, IBAS111 http://ibas.iasf-milano.inaf.it/ (Mereghetti et al., 2003) discussed in more detail in section 2. Since the launch in 2002 INTEGRAL has detected over 900 soft gamma-ray sources222 http://www.isdc.unige.ch/integral/ (Bird et al., 2016) and has localised 114 GRBs (numbers correct as of 2016 July 1). INTEGRAL has made some important discoveries regarding GRBs, reviewed in Götz (2012) including investigations utilising IBIS and SPIs capability to perform spectral analysis on the INTEGRAL sample of GRBs (Vianello et al., 2009; Bošnjak et al., 2014). Furthermore, Foley et al. (2008) suggested that INTEGRAL may be capable of detecting the local, low-luminosity GRB populations.

In the fully coded field of view (FOV), i.e. the central $9^{\circ}\times 9^{\circ}$, the INTEGRAL IBIS instrument is more sensitive than the Burst Alert Telescope (BAT) (Barthelmy, 2004) on board Swift, despite its smaller effective area (2600 cm2 compared to 5200 cm2). This is due to the fact that, at the energies we are interested in ($15-200$ keV), the background is dominated by the Cosmic X-ray diffuse emission, which is proportional to the FOV (a factor of about ten smaller for IBIS than for BAT). Therefore INTEGRAL should be able to reach lower peak flux limits, especially for GRBs with hard spectra where peak energies $>50$ keV (Bošnjak et al., 2014). However, since INTEGRAL spends a large fraction of its observing time observing at low Galactic latitudes, its sensitivity is reduced by the additional background caused by bright Galactic sources and hard X-ray Galactic diffuse emission. It is only since the INTEGRAL sub-threshold trigger campaign began (see section 2) that lower sensitivities have been routinely accessible through WEAK alerts.

Swift has two additional instruments, the X-Ray Telescope (XRT) (Burrows et al., 2005) and the Ultraviolet/Optical Telescope (UVOT) (Roming et al., 2005), and has the ability to slew towards a BAT-detected burst or pre-selected target. Therefore it can complement INTEGRAL with rapid multi-wavelength, follow-up measurements. Using observations from both satellites we expect to uncover both the temporal behaviour and energetics of both the WEAK alerts and INTEGRAL GRB sample and characterise their properties.

We start by discussing the INTEGRAL Burst Alert System (IBAS) in more detail and describe our chosen WEAK triggers in section 2. Our Swift follow-up analysis is discussed in section 3. These are then analysed in conjunction with the total IBAS GRB sample in section 4 with some comparisons to the Swift GRB population. We conclude with our summary in section 5.

2 INTEGRAL IBAS Alerts

INTEGRAL was designed as a general purpose gamma-ray observatory, not specifically optimized for the study of GRBs. However, its good imaging capabilities over a field of view of $\approx 30\times 30^{\circ}$ ($9\times 9^{\circ}$ fully coded and $19\times 19^{\circ}$ half coded) and the continuous telemetry downlink (due to its high elliptical orbit with a period of 3 days) made it possible to search and localize GRBs on the ground in near real time. This is done with the INTEGRAL Burst Alert System, IBAS (Mereghetti et al., 2003), software running at the INTEGRAL Science Data Centre, ISDC (Courvoisier et al., 2003) since the launch of INTEGRAL in October 2002.

No GRB triggering algorithm is implemented on board the satellite. The data reach the ISDC typically within 20 seconds after they have been collected and are immediately fed into the IBAS software which exploits several burst detection programs in parallel. When a burst (or any other new transient source) is detected inside the field of view of the IBIS instrument, its coordinates are automatically distributed via the Internet by means of Alert Packets based on the User Datagram Protocol (UDP). Their coordinates derived by IBAS have a mean with $1\sigma$ uncertainty of 2.1($\pm 0.5$) arcmin.

IBAS also searches for GRBs detected in the Anti-Coincidence Shield (ACS) of the SPI instrument, which provides a good sensitivity over nearly the whole sky, but without localisation and spectral information (von Kienlin et al., 2003). The ACS lightcurves are used for GRB localizations by triangulation with other satellites of the IPN network (Cline et al., 1999). In this investigation we will not discuss SPI ACS results.

The search for GRBs in the IBIS data uses two different kinds of programs: rate monitor and image monitor programs. Rate monitors look for excesses in the light curve of the whole detection plane, while image monitors search for excesses in the deconvolved images. Both use data from ISGRI (Lebrun et al., 2003), the lower energy detector of IBIS, which provides photon by photon data in the energy range 15 keV - 1 MeV. Several instances of the rate and image monitors run in parallel using different settings for integration time scales and energy range. When one (or typically more) of the monitor programs triggers, an imaging analysis is performed on an optimally selected time interval in order to confirm the source presence and derive its significance.

Two significance threshold levels, labelled STRONG and WEAK, have been implemented in IBAS for what concerns the distribution of Alert Packets. The positions of new sources with significance above the STRONG threshold are immediately distributed with Alert Packets. These positions automatically derived by the IBAS software can be later refined by interactive analysis. Until 2011, Alert Packets for sources with significance above the WEAK threshold and below the STRONG were distributed in real time only to members of the IBAS Team, who, after interactive analysis could in some cases confirm the presence of a GRB and distribute its coordinates. However, in the majority of the cases it was not possible, based on the INTEGRAL data alone, to confirm the real astrophysical nature of these low significance events. Since 2011 January 26, all the Alert Packets corresponding to detections above the WEAK threshold have been automatically distributed in real time to the external users who wish to receive them.

Among the 114 confirmed GRBs detected by IBAS, 17 have been detected as sub-threshold WEAK alerts and 54 were observed with Swift, either through an independent autonomous BAT trigger and subsequent follow-up, or via ToO follow-up that was uploaded at a later time, and have available XRT data.

2.1 Selection of WEAK alerts and follow-up

There have been 402 INTEGRAL WEAK triggers, below $8\sigma$ significance, before 2016 July 1; six of which were promoted to STRONG triggers and were later confirmed as GRBs. Out of the other 396 we analysed 15 WEAK triggers. They consisted of:

• •

11 triggers that did not have prompt Swift slews and were target of opportunity observations from our campaign. We named them IGRWYYMMDD prior to source-type identification, broadly following the GRB naming convention, see table 2.1. These are termed as "our chosen ToOs".

• •

Two other WEAK INTEGRAL triggers with ToOs requested elsewhere and had XRT data, but were not related to our 11 chosen triggers, were analysed. These are termed as "candidate GRBs".

• •

Two WEAK triggers that also triggered BAT and had XRT data were also analysed. These are also termed as "candidate GRBs".

Candidate triggers for our Swift ToO follow-up were selected with the requirement that at least one of the following criteria were met. Firstly, triggers were chosen to be close to the $8\sigma$ STRONG threshold (our lowest was $6.7\sigma$). This was to increase the chance of the trigger representing a real GRB. Trigger positions were also checked for high Galactic extinction and close proximity to nearby catalogued X-ray sources. Finally, triggers were generally only followed up if the trigger time coincided with the working hours of the on-call member of the Swift team. The criteria described above were not stringently adhered to for all triggers. We cannot claim that these triggers form a uniform or complete sample and biases towards high significance and lower Galactic column density are present. This was a pilot campaign aimed at determine whether real transient events exist among the WEAK trigger population and we stress that we do not make conclusions for the entire WEAK trigger population.