3.1 Environmental impact on equipment

The observatory site has provided a very challenging environment for the equipment. In particular, humidity levels often rise to above 90% during the night, mainly between June and September. Even though the system automatically shuts down at this humidity level, the primary mirror and camera (which is thermo-electrically cooled) are partially exposed. Condensation combined with dust, salt, and pollen has accelerated the degradation of the mirror coating. To reduce this, the mirror has to be cleaned regularly on the time scale of 4–6 weeks. Figure 2 shows the corrosion on the primary mirror coating and the effect of condensation on the main charge-coupled device (CCD) camera circuit board. UWA technical staff are testing several filter options for the CCD that will inhibit dust intake and condensates on the circuit board.

Figure 2. Top: The primary mirror, showing corrosion of the coating (grey patterns) from the high levels of humidity and condensation, combined with dust and pollen. Bottom: The main circuit board of the IKON-L CCD camera, which is exposed to the atmosphere, showing corrosion on all metallic components and connectors. This eventually led to a camera failure after about 14 months of use.

3.2 RAPIDO: Plans for a co-located wide field rapid response telescope

The French leaders of the TAROT/Zadko projects have designed and constructed a prototype extremely rapid response telescope mount (slew speed 45° per s)Footnote ² . This mount is planned to control two off-the-shelf wide-field [combined 8–10 deg² field of view (FoV)] Takahashi (2 × 20 cm) telescopes. The first imaging using a FLI Kaf16003 camera obtained a limiting magnitude of about 18 (clear filter). RAPIDO and Zadko will share the same robotic control system, which will enable RAPIDO to automatically send alerts to Zadko for deep follow-up of GW and fast radio burst (FRB) candidates. The instrument is upgradable to accommodate a 0.4-m telescope and will be co-located with the Zadko Telescope, subject to funding.

4.1 Gamma-ray bursts

One class of extra-galactic transient that triggers the Zadko Telescope is GRBs. They are the brightest EM explosions in the Universe (see the review by Meszaros, Reference Meszaros2006). Usually detected first in gamma and X-rays by NASA’s Swift satellite (Gehrels et al., Reference Gehrels2004, albeit other instruments can detect GRBs), they have distinct but poorly understood optical signals, including prompt emission, and a delayed afterglow.

The popular progenitor model for a GRB is either a massive stellar collapse, or a compact object merger triggering an explosion causing a burst of collimated gamma-rays powered by accretion onto the newly formed compact object. There are two classes of GRBs categorised by their durations and spectra: namely short-hard GRBs and long-soft GRBs. The former are supposed to be associated with the merging of compact binaries (Eichler et al., Reference Eichler1989), whereas the latter are expected to occur after the cataclysmic collapse of a massive star (Woosley, Reference Woosley1993).

The popular description for GRB emission is the standard model, also termed the fireball model (Rees & Mészáros, Reference Rees and Mészáros1992; Mészáros & Rees, Reference Mészáros and Rees1997; Panaitescu, Mészáros, & Rees, Reference Panaitescu, Mészáros and Rees1998). In this framework, GRBs originate from the dissipation of kinetic energy of an ultra-relativistic outflow (with an initial Lorentz factor of 100–1 000), decelerated by interaction with interstellar matter to produce a fading X-ray and optical afterglow. After the energy decreases to some threshold, the radiation beam is wider than the jetted outflow, so the afterglow becomes observable from angles greater than the high-energy jet.

The fireball model has considerable deficiencies and challenges. For example, the energy budget deduced from the observed emissions and the measured distance implies an on-jet-beamed geometry to keep the energy budget physically reasonable (Rhoads, Reference Rhoads1999). However, Swift has put this geometry in question (Meszaros, Reference Meszaros2006), allowing for exotic theoretical models (Gehrels et al., Reference Gehrels2008) that go-beyond standard relativity. Here, only coordinated multi-spectral observations starting a few seconds after the trigger can test theories.

4.2 Zadko telescope GRB afterglow observations

The implementation of a robotic telescope for fully automated, calibrated imaging of GRBs is complex. The telescope must respond automatically to alerts sent by satellites through the Gamma-ray burst Co-ordinated (GCN) network (Barthelmy et al., Reference Barthelmy1994), if all the environmental conditions are fulfilled. These conditions include the time of day (observations are at night), the position of the GRB on the sky, and the priority of the observation. Because of these limits, not all alerts can be followed-up.

Second, there is no manual intervention at the time of the automated triggered imaging to confirm if the sky is clear and the photometry is calibrated. About a third of the sources followed-up are not useful, either because of cloud cover ora technical problem corrupted the photometry. For the Zadko Telescope GRB afterglow follow-up campaign, we list the successfully imaged bursts in Table 2.

Table 2. Zadko telescope follow-up of GRBs. The localizations are obtained from the Swift X-ray Telescope (XRT) position, with an uncertainty of 1.5 arcsec, or the Swift Burst and Transient detector (BAT; denoted by a γ), with an uncertainty of 1°–4°. XRT observations were not available for triggers marked with a γ. All times are given relative to the trigger time, and we report in this table only the upper limits (see Table 3 for the measurements). Zadko employs the same imaging strategy as TAROT: The first image is a 60 s exposure with tracking off (trailed images), followed by 30, 60, 120, and 180 sec exposures in the tracking mode. For late follow-up (hours post trigger), the 180 sec exposures are stacked. The label ‘ld’ indicates that the data were lost during a computer update.

One can clearly see in the distribution of observation times ‘holes’ corresponding to major refurbishments of the instrument, such as the installation of a new robotic observatory in 2012 (see above). Despite this, and other technical issues, the Zadko telescope detected 11 GRBs in optical out of 40 successful follow-ups during its 6 yr campaign. This is a typical rate for a robotic instrument (see for instance Klotz et al., Reference Klotz, Boër, Atteia and Gendre2009, for a comparison with the TAROT instruments).

4.3 Automated response time, detections, and photometry

The response time is a key parameter for all robotic telescopes. It will constrain the main scientific drivers in a given field. Comparing the shortest possible response time with the distribution of GRB durations (Kouveliotou et al., Reference Kouveliotou, Meegan, Fishman and Bhat1993), it is clear that the Zadko telescope cannot be used to study the early prompt phase of normal long GRBs, with a mean duration of the order of 20 s, but can image the dynamic early afterglow phase during the transition from prompt to afterglow. We highlight that the response time has improved to a minimum possible response time of > 30 s, during the last 2 yr because of the new observatory (see above).

The 90% confidence level limiting magnitude reached by the Zadko telescope is consistent with theoretical expectations shown as the thick blue lineFootnote ³ in Figure 3. It clearly indicates that the telescope cannot probe the entire GRB afterglow flux/luminosity distribution. One day after the burst, only half of the afterglows can be detected. After 3 to 4 d, Zadko cannot detect any afterglows. As a consequence, this constrains the observation strategy against long follow-up delay times (1 d or more) of the afterglow and highlights that the optimal follow-up strategy is a low latency response (comparable to the reaction time). At 1 d post-alert, it is more practical to search for the host galaxy using deep (stacked) exposures.

Figure 3. Distribution of GRB afterglow observations and detections by the Zadko telescope. GRB afterglows are represented by filled circles, whereas upper limits are represented as empty triangles. The red shaded area represents the light curves compiled in Kann et al. (Reference Kann2010). The solid blue line is the theoretical observation limits under normal weather conditions with the previous observatory (terminated in 2011). The dashed blue line (at 30 s) is the shortest response time possible for the new observatory (see the electronic version in colour).

In Figure 3, the point representing GRB 090205 is an outlier located below the detection limit. However, it was observed during the commissioning phase with a different camera and settings than the one used after robotization: The detection limit presented in the figure does not apply to this point.

The Zadko telescope can be used to study the so-called dark GRBs (Jakobsson et al., Reference Jakobsson2004). This subset of GRB afterglows reveals a lack of emission in optical when compared to the X-ray, and their nature still remains puzzling. Because Zadko can obtain upper limits during the first minutes of the burst (up to about 20, see Figure 3), the data can be used to classify a burst as dark (the faintness observed in optical can also be linked to a global faintness observed at all wavelengths), and provide a secondary trigger for a deep follow-up of the GRB location and host galaxy.

Among the sample, several afterglows were imaged, but (for most of them) only for a short duration (typically for about 1 to 2 h). In addition, most sources were visible with a delay of several hours. Only four events were followed-up a few tens of seconds after the trigger and lasting for more than 3 h: GRB 090313, GRB 101024A, GRB 130427A, and GRB 140311A. We list all the observations and detections in Tables 2 and 3, respectively. It is beyond the scope of this paper to present a detailed analysis of each burst observed by the Zadko telescope, but more detailed studies have been undertaken (e.g. Gendre et al., Reference Gendre2011, for GRB 101024A).

Table 3. GRB afterglows imaged by the Zadko telescope. Filters are labelled ‘C’ for clear and ‘R’ for R band. For clear filter images, the magnitude is expressed in equivalent R band for comparison with other instruments. The time given relative to the trigger time, and the imaging strategy is the same as described in Table 2.

4.4 Further GRB science opportunities

The response time and limiting magnitude set the potential for breakthrough science in relation to testing the standard fireball model. Potential new science and discoveries from Zadko telescope automated triggered imaging include:

• • The spectral evolution of the prompt phase. The Zadko telescope will not compete with faster robotic instruments for observing the start of the prompt phase, as can be noted in Figure 3; however, its response time and sensitivity could probe the origin of the early optical emissions prompt phase of the brightest and longest duration GRBs.

• • The plateau phase. One of the main results from the Swift satellite is the presence of a plateau phase seen in X-ray between the prompt and the afterglow phases. The nature of this plateau is debated, and the comparison with optical data is critical to test the current models. Boër, Gendre, & Stratta (Reference Boër, Gendre and Stratta2015) have shown that the distribution of the end of the prompt phase in X-ray peaks at about 100 s. Accordingly, Willingale et al. (Reference Willingale2007) have shown that this plateau phase ends about 10⁴ s after the trigger. Thus, there is a significant fraction of bursts with the plateau phase starting within the window of Zadko telescope triggered imaging times.

• • Population III stars. The first stars are metal-poor and proposed relics of the early Universe. They should form at very large redshift (z ~ 10–20, Bromm & Loeb, Reference Bromm and Loeb2006), but could rarely be present at z ~ 2–4 (Tornatore, Ferrara, & Schneider, Reference Tornatore, Ferrara and Schneider2007), and should produce GRBs (Suwa & Ioka, Reference Suwa and Ioka2011; Nakauchi et al., Reference Nakauchi, Suwa, Sakamoto, Kashiyama and Nakamura2012). Moreover, isolated blobs of original matter, without metals, could be present near the Milky Way, and could form, if disturbed, new population III stars (Morales-Luis et al., Reference Morales-Luis, Sanchez Almeida, Aguerri and Munoz-Tunon2011). The properties of these GRBs are expected to be similar to ultra-long GRBs (Gendre et al., Reference Gendre2013; MacPherson, Coward, & Zadnik, Reference MacPherson, Coward and Zadnik2013), with a prompt phase lasting more than 600 s (indicated by the 10 min line in Figure 3).

5.1 L-type and Barbarian asteroids

Asteroids belonging to the L/K taxonomic classes could hold the key to understand the early phases of formation of the Solar System, when the solid phase in the proto-planetary nebula, which was largely dominated by highly refractory material, which solidify at the highest temperatures. Recent numerical simulations (Johansen and Klahr, Reference Johansen and Klahr2011) suggest that, in those conditions, typical of the first ~ 1 Myr after the disk formation, local concentrations of solid grains could have been rapidly assembled in large planetesimals of ~ 100 km. Recently, near-IR observations (Sunshine et al., Reference Sunshine, Connolly, McCoy, Bus and La Croix2008) have suggested that some asteroids could also include in their composition a high fraction of that primordial material, hence be extremely ancient. A large network led by P. Tanga recently started an extensive activity of physical characterization of those asteroids, the so-called Barbarians, from the name of (234) Barbara, the first of this type discovered (Cellino et al., Reference Cellino, Belskaya, Bendjoya, Di Martino, Gil-Hutton, Muinonen and Tedesco2006). The anomaly leading to the identification of the Barbarians was the peculiar variation of their linear polarization with the phase angle (the angle between the directions to the Sun and to the Observer, as seen from the asteroid).

A total of 13 Barbarians have been found so far (Gil-Hutton et al., Reference Gil-Hutton, Mesa, Cellino, Bendjoya, Pealoza and Lovos2008; Masiero & Cellino, Reference Masiero and Cellino2009; Cellino et al., Reference Cellino, Bagnulo, Tanga, Novaković and Delbo2014). Seven of them are in the Watsonia dynamical family, caused from the disruption of a single parent body (Cellino et al., Reference Cellino, Bagnulo, Tanga, Novaković and Delbo2014). In terms of taxonomy, based on their similar spectro-photometric properties, Barbarians are classified as members of the unusual classes: L, Ld, and K.

In absence of solid theoretical models capable of explaining the observed polarization, Cellino et al. (Reference Cellino, Belskaya, Bendjoya, Di Martino, Gil-Hutton, Muinonen and Tedesco2006) suggested, as a speculative explanation, that anomalous polarization could also be due to large-scale concavities responsible for causing an anomalous distribution of incident sunlight and scattering angles. In this respect, (234) Barbara has a very irregular shape and reveals large concavities, as shown by Delbo et al. (Reference Delbo, Ligori, Matter, Cellino and Berthier2009), and later confirmed by a more extensive analysis using different techniques (Tanga et al., Reference Tanga2015).

For the small sample of known rotation periods of the largest Barbarians (seven objects), all of them are slow rotators (period 12–24 h or more) with respect to the average period of similar-sized asteroids (5–6 h). Slow rotation periods and the possible presence of concavities are indicators of past collisional history. However, the verification of this hypothesis requires a large amount of data on rotations and shapes. One tool for determining rotation periods and indications of shape irregularities is to obtain light curves spanning a near complete period.

5.1.1 Zadko photometry of asteroids

As the rotation period of the so-called slow rotators generally exceeds the duration of night on Earth, telescopes distributed across many longitudes are required. Ideally, a light curve should be obtained over ~ 1 week; otherwise the changing illumination/observation geometry will make its interpretation more complex. For these reasons, Zadko is exceptionally well placed relative to other instruments that have also been employed (in France, Poland, USA, and Chile). We note that this type of observation requires blocks of dedicated time, several hours of imaging spread over a night. To achieve this task automatically on the Zadko telescope, we created a high priority schedule that could only be interrupted by a GRB alert.

During 1 yr (2014 to 2015), the Zadko telescope contributed to the period determination for 11 asteroids. These objects (and most of the known asteroids) do not have measured rotation periods, or at best preliminary ones with unconfirmed values (with accuracies of the order of 0.1 d or worse) obtained from partial light curves not covering full rotations. Figure 4 illustrates a typical example, in which Zadko observations covers a portion of a light curve observable only from Australia, thus capturing the maximum of brightness during a single rotation. The European longitudes of C2PU close to Nice, France ensured the coverage of the minimum over the same rotation of the asteroid.

Figure 4. Light curve of (387) Aquitania. The observations by the Zadko telescope provided light curve measurements around the brightness peak, unobserved at other longitudes. With a rotation period close to 24 h, a single site can only acquire partial light curves.

By obtaining the near complete light curves, we were able to measure rotation periods with uncertainties better than 1 × 10⁻³–1 × 10⁻⁴ d with a measured dispersion < 0.01 magnitudes for targets of relatively high brightness (V ~ 11–13). This accuracy exceeds the photometric accuracy required for deriving most asteroid rotation periods (having typical amplitude variation of 0.2–1 mag). Zadko photometry contributed to the shape determination of (234) Barbara (Tanga et al., Reference Tanga2015), and updated results are due to appear in a paper in preparation.

7.1 Zadko supernova follow-up

Because the field of view of the Zadko Telescope cannot scan a large part of the sky, an innovative SN search strategy has been developed. The goal is to detect SNe during the early stage of the explosion, and to follow-up until they reach their maximum brightness ( ~ 20 d after the explosion) with a high sampling. For some fraction of CCSNe, we expect to detect the tail of the thermal shock breakout emission in the optical/near-UV domain ( ~ 1 d after the explosion) as already observed for a few of CCSNe, SN1987A (Hamuy & Suntzeff, Reference Hamuy and Suntzeff1990), SN1993J (Richmond et al., Reference Richmond, Treffers, Filippenko, Paik, Leibundgut, Schulman and Cox1994), SN1998bw (Galama et al., Reference Galama1998), SN1999ex (Stritzinger et al., Reference Stritzinger2002), and SN2008D (Modjaz et al., Reference Modjaz2009). To achieve this, we designed a catalogue of nearby galaxies of interest which historically are prolific in producing SNe, or for which detection SN is easier, i.e. face-on galaxies. After studying the properties of 1 000 galaxies that have produced SNe since 1885Footnote ⁵ , we conclude that the most prolific galaxies include:

1. 1. Nearby galaxies ( ⩽ 40 Mpc) with large angular radius ( ⩾ 4 arcmin)

2. 2. Face-on galaxies

3. 3. Starburst or HII galaxies that have an enhanced stellar formation

4. 4. Spiral galaxies or merging galaxies

According to the above criteria, we would expect to detect 7–10 SNe per year from our catalogue of 112 galaxies. However, taking into account the variable weather conditions, the moon phases and expected technical difficulties, we should observe 2–3 SN during their early phase with good observational conditions. In order to optimise the number of observed galaxies per night, we specify an exposure time for each galaxy to detect a SN to a limiting absolute magnitude of $R_{\text{abs}}\sim$ − 13. This limiting magnitude is deep enough to detect faint sources like young SNe without consuming too much observation time.

The program was implemented in June 2014 and in 2014-2015, 8 SNe were discovered in galaxies belonging to our catalogue: SN2014A (NGC5054), SN2014B (NGC4939), SN2014C (NGC7331), SN2014J (NGC3034), SN2014L (NGC4254), SN2014df (NGC1448), ASAS-SN2014ha (NGC1566), and SN2014dt (NGC4303). Among them, six were classified as CCSNe and two as Type Ia SNe. Unfortunately, the first five SN were discovered 4 months before the program commenced. Due to observational constraints, the Zadko Telescope detected the three other supernovae (SN2014df, ASAS-SN2014ha and SN2014dt) days after their discovery (see Figure 6 for light curves of SN2014df and SN2014ha). Example images can be found at http://www.rochesterastronomy.org/sn2014, http://www.astronomerstelegram.org/?read=6460 and http://www.rochesterastronomy.org/sn2014/.

Figure 6. Light curves of SN2014df (left) and SN2014ha (right) with R-band (red), V-band (green), and unfiltered observations (black). The Zadko photometry, not corrected for Galactic dust extinction and converted to R and V filters, are shown as filled circles. Discovery of SN2014df and SN2014ha (squares), were made by Berto Monard ( ± 0.14 mag), and the All Sky Automated Survey for SuperNovae (V ≈ 14.6 and 14.9, no errors provided), respectively.

Since the beginning of the SN program, the number of SN discovered per galaxy is in good agreement with the expected rate. So far, the project has not detected an early SN, which we attribute to a poor duty cycle from weather conditions, technical upgrades and observational constraints. The project will continue to search for young and nearby SNe with a future update of the targeted galaxy catalogue.

8.1 The aLIGO/AdV follow-up program

For the first aLIGO observation run (O1 that ran from September 2015 to January 2016), Zadko was one of a number of EM facilities selected for follow-up GW sources. The LIGO/Virgo collaboration had over 63 worldwide EM partners with signed MoUs for conducting follow-ups of GW triggers at that time; around 70 MoUs will be in place for the second observation run in late 2016 (O2). Of the partners, 73% are in the UV/Optical/IR, 11% in the radio and the remainder are in high energy X-ray and Gamma-ray. Australian involvement includes MWA (Tingay et al., Reference Tingay2013); ASKAP (VAST) (Murphy et al., Reference Murphy2013); AAT (Tinney et al., Reference Tinney, Ryder, Ellis, Churilov, Dawson, Smith, Waller and Whittard2004); SkyMapper (Keller et al., Reference Keller2007), GOTOFootnote ⁶ and the ‘Deeper Wider Faster’ project that targets simultaneous, fast cadenced observations through shadowing of optical and radio fields; interesting candidates are followed up using a number of partner EM facilities, including Zadko, covering low to high energy (Howell et al., Reference Howell2015). In the high energy gamma-ray, there will be Australian involvement through the Cherenkov Telescope Array (CTA; Acharya et al., Reference Acharya2013), The High Energy Stereoscopic System (H.E.S.S; Lennarz et al., Reference Lennarz2013), as well as coordinated neutrino observations using IceCube (IceCube Collaboration et al., 2006).

The main latency bottleneck during O1 was the automatic and manual verifications that included checks on data quality and conditions. Therefore, to allow alerts to be sent out significantly faster, automation is a particular aim for future science runs in the advanced GW detector era. The statistical significance of a GW trigger is given through its false alarm rate (FAR), which is the rate that false positives appear above a given signal-to-noise ratio threshold. Typical FARs for triggers sent out during the O1 were of order 1/month.

To rapidly communicate the information on GW events, the VOEventFootnote ⁷ standard was adopted (Williams et al., Reference Williams, Barthelmy, Denny, Graham, Swinbank, Peck, Seaman and Comeron2012). The content of a VOEvent alert sent out by aLIGO for a compact binary coalescence event included estimates of the FAR (in Hz) and the sky position of the source provided by a way of a probability sky map (which can be multi-modal and non-Gaussian). Future runs could include additional information such as estimates of the luminosity distance of the source and an indication of the component masses of the system.

8.2 The first EM follow-up of a GW transient

The first confirmed GW transient, GW150914 (a binary black hole merger) was discovered by the aLIGO detectors on 2015 September 14 (Abbott et al., Reference Abbott2016c). Preliminary estimates of the time, significance, and sky location of the event candidate were shared with 63 teams of observers covering radio, optical, near-infrared, X-ray, and gamma-ray wavelengths with ground- and space-based facilities. Following detection through the low-latency burst detection pipelines a Coherent WaveBurst (cWB) skymap, based on minimal assumptions, was available within 17 min; this was followed by more comprehensive LALInference Burst (LIB) skymap within 14 h. Due to additional data integrity and verification checks, these two maps were not circulated till 2 d post-detection. Of the 63 teams, 25 follow-ups were reported via a private Gamma-ray Coordinates Network Circulars. Details of the skymap production and the follow-up observations (by 25 teams) for this first campaign are outlined in Abbott et al. (Reference Abbott2016b); this included a disputed claim of a possible short GRB coincident (within 60 ms of GW150914) with the GW event (Connaughton et al., Reference Connaughton2016). Despite over 60 participating facilities with MoUs for EM follow-up, around a third managed to acquire data for this first event. This can be attributed to many factors, including weather (optical), instrumental problems, and rapid follow-up not possible because of longitude separation. For the first follow-up campaign, the only Australian facilities or collaborations to obtain follow-up data were SkyMapper and TZAC (TAROT—Chile), of which Zadko (the Australian node) was not operational at the time of the trigger.

8.3 Gravitational wave follow-up challenges and strategies

Looking ahead to the second aLIGO/Virgo observation run (02; late 2016), the most likely GW sources to yield optical counterparts are the mergers of coalescing binary systems of neutron star. As discussed in Section 4.1, these events have been strongly suggested as the progenitors of short hard γ-ray bursts (Tanvir et al., Reference Tanvir, Levan, Fruchter, Hjorth, Hounsell, Wiersema and Tunnicliffe2013; Berger, Fong, & Chornock, Reference Berger, Fong and Chornock2013). The evidence stems from the fact that short-hard GRBs are often observed in older stellar populations with offsets of tens of kpc from their galactic centres; this is consistent with binary compact object formation channels. Additionally, follow-up observations in the infrared of GRB 130603B provided evidence of a ‘kilonova’; a faint transient predicted to form after the merger of two neutron stars, and powered by the radioactive decay of the ejected neutron rich matter (Li & Paczyński, Reference Li and Paczyński1998; Rosswog, Reference Rosswog2005; Metzger et al., Reference Metzger2010).

As highlighted during O1, a particular challenge for optical follow-ups of GW sources during O2 and beyond will be error boxes of 100 s deg²; compare this with the typical error boxes for GRB follow-ups of order arc mins. As mentioned earlier, these error regions are provided to optical facilities as probability sky maps. The morphology of the skymaps is dependent on the location of the source in the sky relative to the antenna pattern function of the GW detector network. The shapes of the probability map can take the form of a single elongated arc that could cover several hundred square degrees, or two or more degenerate arcs that could include small isolated regions. When Virgo increases the network to three-detectors (in 2017) localisations should improve; one can expect confidence regions of 10 s of degrees in less than 17% of events (Singer et al., Reference Singer2014).

To make the most of opportunities, instrument dependent strategies must be employed to cover the most probable regions of the error region in sufficient time to capture a fading EM source. These strategies will have to be adapted depending on the particular morphology and for partially accessible error regions, the component available to the southern hemisphere. The most likely EM counterparts, short hard GRBs have fainter afterglows than the longer duration bursts. However, for on-axis sources within the aLIGO/AdV horizon one would expect these events to be bright enough for identification; however, event rates and beaming considerations suggest that in most cases the bursts would be off-axis requiring deep searches.

Different follow-up techniques to screen candidate counterparts are still in the testing stage; these include simple tiling strategies employing galaxy catalogues to more sophisticated clustering algorithms based on machine learning. Source confusion will also be an important consideration. Sources such as Solar System objects can be excluded by employing a cadence of order 10 s to mins between images of the same field. Supernovae can be excluded by their relatively long durations and flare stars can be monitored for repeated variability. The most important strategy for identifying the type of candidate transients will be coordinating deep follow-up with other facilities (both radio and optical). We are currently in discussions with multiple collaborations with access to globally distributed meter class facilities for the above.

Initial follow-ups by Zadko will employ tiling the highest probability sky regions and secondly target only the brightest galaxies utilizing the current automated catalogue matching pipeline that creates a list of unknown candidates for an alert imageFootnote ⁸ . Image subtraction can also be used for galaxies that we have reference images, i.e. from our current supernovae search.

8.4 Coordinated GW follow-up imaging with Zadko

With the intense interest in GW follow-up, a number of facilities are being constructed as dedicated instruments with FoVs large enough to cover the GW error regions in reasonable times. One such facility that could be operational during the later stages of aLIGO/AdV is RAMSES; if successful a similar instrument could operate alongside Zadko. The RAMSES prototype will consist of a set of 16 × 40 cm telescopes equipped with rapid cameras capable of readout times from 100 ms to several minutes; it will be able to reach magnitudes of 15 in 1 s and mag 20 within around 10 min. Each telescope has a 2.5 deg² FoV, resulting in a combined 100 deg² FoV; this in combination with a slew time of around 5 s would enable instantaneous coverage of a large proportion of a typical aLIGO/AdV error region. RAMSES has received funding and will be based in National Aures Observatory in AlgeriaFootnote ⁹ with a prototype built in France.

Comparable instruments employed in GW follow-ups include BlackGEM (40 deg² FoV; mag 22 in 5 min; Bellm, Reference Bellm, Wozniak, Graham, Mahabal and Seaman R2014) and the Zwicky Transient Facility (ZTF; 47 deg² FoV; mag 20.5–21 in 30 s; Ghosh & Nelemans, Reference Ghosh and Nelemans2015); other than being more sensitive, these telescopes operate as stand-alone facilities; therefore, a telescope like RAMSES could have a niche role alongside a more sensitive instrument. The wide FoV and rapid slew of RAMSES allows for rapid scanning of the GW error region for bright transients that can then be targeted for instantaneous deep follow-up by the Zadko telescope.

We plan to build and install RAPIDO (very rapid slewing mount co-located with Zadko) using two wide field telescopes combined with a CCD imager that will achieve m < 17 with FoV (8–10)deg². The single telescope system, comparable to RAMSES but using off the shelf optics, is expandable and can accommodate either 0.4 m or 2 × 0.2 m telescopes on each mount. The system will provide low latency wide-field searches for transients and will automatically send alerts to Zadko for deep follow-up (m < 21) of any transient candidate. We note that when Virgo joins aLIGO, for the brightest GW sources the localisation error regions will reduce to some tens of degrees, providing opportunities for low-latency imaging of the entire field by RAPIDO and deep follow-up by Zadko.

9.1 Zadko Telescope automated follow-up of Parkes

The Zadko Telescope and in the future RAPIDO telescopes, based on their reaction times and sensitivities for GRB imaging, have the potential to fill a niche role in extremely rapid follow up of radio transients events (see Figure 7). Previously, the most rapid follow up of FRB140514 was 7 h later, with ATCA radio telescope. The most rapid optical follow up was performed by GROND (GRB Optical/Near-Infrared Detector) but this was 16 h post-alert. SWOPE Footnote ¹⁰ observations, also more than 16 h after the event, provided a limiting magnitude of only R = 16. None of these follow-ups could offer a candidate for the host galaxy.

Figure 7. Zadko on source delay time (horizontal left line) for FRB follow-up, compared to all other optical follow-ups (symbols). The low latency FRB response was obtained during a Parkes shadowing experiment in December 2015, which set the record for the shortest time delay to image the FRB alert. The Zadko limiting magnitude of 19.5 (60 s exposure in r) is improved by about 0.8 mag in the stacked image. A complete analysis of this data will be presented in a subsequent work.

As of April 2015, the Zadko Telescope joined a large collaboration that is searching for FRBs in real time. The project is part of the alert network for the SUPERBFootnote ¹¹ and UMOSTFootnote ¹² projects. As part of this effort the Zadko control software has been updated to respond robotically to FRB alerts. Results from these observations have the potential to probe a region of parameter space not covered.

In December 2015, the team commenced a pilot project for several weeks that enabled the Zadko Telescope to shadow (follow) the same sky location as Parkes. Because Parke’s beam is about four times the area of Zadko’s FoV, we employed a tiling strategy. This first Zadko pilot experiment broke the record for the fastest response (about 40-min post-burst), and earliest optical data following an FRB alert (see Figure 7). The first imaging of the source occurred before the human vetted alert was sent out from Parke’s (analysis of this low latency data is on-going and will be reported in a subsequent work).

Based on event rate estimates from the 2015 observing sessions, 10 FRBs should be discovered in Parkes data in 2017. This provides an opportunity to search for a very early optical counterpart, to identify a possible short GRB afterglow, localise a host galaxy to confirm the distance, and to study the host environment.

10.1 Science education enrichment

The Zadko Telescope has been used for several novel science education programs. Local PhD student projects include the evaluation of authentic learning by participation in research projects (Heary et al., Reference Heary, Coward, Blair and Venville2012; Coward et al., Reference Coward2011). These projects include local school students identifying and searching for near Earth asteroids using the Zadko Telescope. This was extended as part of the Yachad visiting scholar award (led by Coward in November 2012) who initiated a pilot scheme that provided local school students in Israel access to the UWA Zadko Telescope for participating in the above project. In addition, there has been numerous undergraduate student training programs, both hosted locally and remotely from France that have used the Zadko Telescope for their projects.

10.2 Space science training: space debris

The potential exponential growth of artificial debris in Earth orbit together with its consequent hazard to satellite operations (Kennewell & Brockman, Reference Kennewell and Brockman2007) has been known for some time (Kessler and Cours-Palais, 1978). Most observations and studies have concentrated on the low Earth orbital (LEO) regime. There are several military and civilian observing patrol networks that contribute to the maintenance of catalogues of the larger debris sizes ( > 10 cm). Occasional scientific campaigns are undertaken to explore smaller sized debris, as it has been realised that the growth of even micron sized debris has the potential to interfere with astronomical observations in the medium future (Biggs & Kennewell, Reference Biggs and Kennewell2012).

Collisional velocities in geosynchronous Earth orbit (GEO) are lower than those in LEO, and this has led, along with the greater difficulty of observation, to a lesser interest in GEO debris. However, the discovery a few years ago of a previously unknown GEO debris population (Schildknecht et al., Reference Schildknecht2004; Schildknecht, Musci, & Flohrer, Reference Schildknecht, Musci and Flohrer2007) has led to the realisation that the hazard in this region has been considerably underestimated. This high altitude region is totally beyond the range of radars used for LEO patrol, and metre-class optical instruments are required for useful analyses.

The Zadko telescope has previously been involved in imaging Australian space debris (Laas-Bourez, Kennewell, & Coward, Reference Laas-Bourez, Kennewell and Coward2011), and we currently have an open PhD project to characterise the GEO debris population in this longitudinal region of the geosynchronous orbit. (One of us has been involved in a study with Kitt Peak National Observatory that has shown that the large space object population at GEO shows a significant longitudinal variation.)

The substantial public and schools interest in artificial space debris (space debris is a component of the WA school science curriculum) offers a great opportunity for public outreach in this field as well as for training of teachers and students in the acquisition of hands on skills in planning and implementing telescope scheduling and debris database access and analysis. These activities are expected to be coordinated with a co-located node of the Falcon Telescope Network dedicated space surveillance facility (Chun et al. 2014).