Filename
stringlengths 22
64
| Paragraph
stringlengths 8
5.57k
|
|---|---|
Processed_The_lunar_Askaryan_technique_with_the_Square_Kilom.txt | The lunar Askaryan technique was proposed by Dagkesamanskii & Zheleznykh (1989) as a method to study the highest-energy cosmic rays and neutrinos. By observing the Moon from a satellite or ground-based telescope and searching for Askaryan emission (Askaryan 1962) from particle interactions, the entire visible lunar surface (of order 20 million km2) can be utilised as a particle detector. This method has been used successfully to place limits on exotic (‘top-down’) models of ultra-high-energy (UHE; E & 1020 eV) neutrino production (Buitink et al. 2010; Gorham et al. 2004). However, due to the large Earth-Moon distance (3.844·105 km on average), the signals on Earth are very weak, which has precluded current radio telescopes from being sensitive to either neutrino fluxes from conventional ‘bottom-up’ cosmic ray acceleration models, or to the cosmic ray flux itself. |
Processed_The_lunar_Askaryan_technique_with_the_Square_Kilom.txt | This situation will change with the advent of the Square Kilometre Array1 (SKA). The largest radio telescope ever conceived, the SKA will consist of multiple components, with the low-frequency telescope SKA-low covering the range 50–350 MHz. This instrument, to be built in Western Aus- tralia (Dewdney et al. 2013) between 2018–2023 (phase 1) and 2023–2030 (phase 2), is particularly appealing as an UHE particle detector, due to the increased angular width of the Askaryan emis- sion at low frequencies (Scholten et al. 2006). When complete, SKA-low is expected to detect an unprecedented number of UHE cosmic rays, with sufficient angular resolution to study their origin. In order to enable lunar Askaryan observations with SKA-low, groups that have performed observations with existing radio telescopes — LUNASKA (Bray et al. 2015; James et al. 2010), NuMoon (Buitink et al. 2010), RAMHAND (Beresnyak et al. 2005), LaLUNA (Spencer et al. 2010), and RESUN (Jaeger et al. 2010) — have jointly2 formed the SKA High Energy Cosmic Particles (HECP) Focus Group3. This group will apply the necessary methods and technology, de- veloped as part of the aforementioned forerunner projects, to observations with SKA-low, and work with SKA engineers to implement this observation mode. In parallel, simulations have been per- formed in order to estimate the sensitivity of SKA-low during both its initial phase 1 deployment, and final phase 2 configuration. |
Processed_The_lunar_Askaryan_technique_with_the_Square_Kilom.txt | The results of this effort are reported here. Sections 2 and 3 respectively give an overview of SKA-low and the simulation chain used to simulate its sensitivity to lunar Askaryan pulses. In Section 4, estimates of the ability of SKA-low phases 1 and 2 to detect and study UHE cosmic rays (CR) are given, including preliminary energy and direction resolutions. Section 5 discusses the prospects to study UHE neutrinos and constrain the remaining top-down models of UHE particle production. Finally, Section 6 reviews the science case of the HECP (lunar) group in light of these results and in comparison to other experiments, and discusses the necessary technical requirements in order to achieve the simulated sensitivity. |
Processed_The_lunar_Askaryan_technique_with_the_Square_Kilom.txt | Table 1: Parameters of SKA-low used for sensitivity calculations: sensitivity Aeff/Tsys, minimum fmin and maximum fmax observation frequencies, characteristic lunar beam coverage, and detection threshold s thresh. |
Processed_The_lunar_Askaryan_technique_with_the_Square_Kilom.txt | During standard astronomical observations, the station beams will be sent to a central corre- lator facility for image processing. For both pulsar timing and lunar observations, these station beams will be added in-phase to form ‘array beams’, with up to 16 dual polarisation array beams being produced for the phase 1 instrument. Since the angular size of each array beam will be deter- mined by the baseline over which it is formed, only the core stations will be used for lunar detection mode, so that the 16 phase 1 beams will cover approximately 50% of the lunar limb. These beams will be analysed in real-time by purpose-built hardware, which will perform the necessary process- ing to search for broad-bandwidth pulses (Bray et al. 2015). During phase 2, a similar scheme is envisaged, but with sufficiently many array beams to cover the entire Moon. |
Processed_The_lunar_Askaryan_technique_with_the_Square_Kilom.txt | The program used to simulate the sensitivity of the SKA was developed by James & Protheroe (2009)4. It is a Monte Carlo routine that can simulate both neutrino and cosmic ray interactions in — and in the former case, propagation through — the Moon, and produces frequency-dependent radio emission from the resulting cascades according to parameterisations based on full simulations (Alvarez-Muñiz et al. 2006; Alvarez-Muñiz & Zas 1998). The emission is propagated through a roughened lunar surface, and the signal spectrum as seen by a telescope on Earth is calculated, assuming no loss of coherence due to e.g. small-scale lunar surface roughness, or the ionosphere. Here, the former effect is ignored, since it is expected to be negligible for low-frequency cosmic ray observations, while the latter effect is assumed to be corrected for during signal processing. This program has been verified through comparison with analytic calculations of Gayley et al. (2009). |
Processed_The_lunar_Askaryan_technique_with_the_Square_Kilom.txt | Figure 1: Left: effective apertures of SKA-low phases 1 and 2 to cosmic rays as a function of their energy. Right: expected cosmic ray detection rates, after convolving with the measured flux (Abraham et al. 2010). |
Processed_The_lunar_Askaryan_technique_with_the_Square_Kilom.txt | The effective aperture of SKA-low to cosmic rays is given in Fig. 1 (left). The phase 1 in- strument achieves an instantaneous aperture of 104 km2 sr, comparable to that of the Pierre Auger Observatory with a typical 60◦ zenith-angle cut (Abraham et al. 2010), at approximately 1020 eV, while phase 2 does so at ∼ 3 · 1019 eV. The relatively small reduction in threshold between phases 1 and 2 is due to the scaling of Askaryan emission: radiated power increases with the square of primary energy, so that a four-fold increase in sensitivity is required to halve the detection thresh- old. The rapid increase in Aeff above the detection threshold however means that at the very highest observed cosmic ray energies near 1020 eV, SKA-low phase 2 will have an effective aperture of over 105 km2 sr. |
Processed_The_lunar_Askaryan_technique_with_the_Square_Kilom.txt | Figure 2: Left: effective area of SKA-low phase 2 to 1020 eV cosmic rays as a function of their arrival direction relative to the Moon. Right: the resulting effective area after placing cuts on the signal strength, origin, and polarisation (see text). The corresponding resolution for such cuts is ∼ 5◦. |
Processed_The_lunar_Askaryan_technique_with_the_Square_Kilom.txt | would be expected to see only a handful of events (∼ 3) per year, with the expected number being subject to uncertainties in the spectrum above 1020 eV. However, SKA phase 2 will be sensitive to the cosmic ray flux above 2 · 1019 eV, with an event rate of ∼ 355 yr−1, and more importantly, ∼ 171 yr−1 above 56 EeV, where anisotropies in UHECR arrival directions have been observed (Aab et al. 2015; Abraham et al. 2007). |
Processed_The_lunar_Askaryan_technique_with_the_Square_Kilom.txt | Explicit reconstruction methods to determine the primary particle energy and arrival direc- tion have not yet been developed for the lunar technique. Such methods would involve fitting the observed signal spectrum e ( f ) over a broad bandwidth for a fully coherent component at low fre- quencies (e ( f ) (cid:181) f ECR) to give the primary energy, measuring the (linear) polarisation to give the projection of the arrival direction in the plane of the sky, and fitting e ( f ) for a spectral downturn at high frequencies to detect any decoherence due to observations away from the Cherenkov angle, in order to determine the angle of emission q out of the plane of the sky. |
Processed_The_lunar_Askaryan_technique_with_the_Square_Kilom.txt | A simple estimate of the minimum accuracy of this technique can be made by placing trivial cuts on the observable signal properties, and looking at the energy- and directional-dependence of the resulting reduced Aeff for these cuts. The expected detection threshold of 10s gives an approximate error of 10% on the measured signal strength, and 0.1 rad on the polarisation angle, which are modelled by only accepting signals with a magnitude in the 10–12s range, and with a polarisation aligned within ±5.7◦ radially from the lunar limb. The expected angular resolution is taken as being 0.5′, corresponding to only the inner part of the array, and only events originating from within 0.5′ of a point on the lunar limb are accepted. |
Processed_The_lunar_Askaryan_technique_with_the_Square_Kilom.txt | Figure 3: Left: effective apertures of SKA-low phases 1 and 2 to UHE neutrinos as a function of their energy. Right: Projected 90%-confidence limits on the UHE neutrino flux from 1,000 hours of observations. Predictions are shown for neutrino fluxes from the so-called “top-down models" involving the production of UHE neutrinos in the Early Universe from kinks (Lunardini & Sabancilar (2012), dash-dotted) and cusps (Berezinsky et al. (2011), dot-dash-dotted) in cosmic strings, and also for the neutrino flux produced in in- teractions of UHE cosmic-rays with the Cosmic Microwave Background radiation - “cosmogenic neutrinos" (Allard et al. (2006), shaded). Limits set by other experiments - the Pierre Auger Observatory (Aab et al. 2015), RICE (Kravchenko et al. 2012) and ANITA (Gorham et al. 2010, 2012) - are also shown. |
Processed_The_lunar_Askaryan_technique_with_the_Square_Kilom.txt | The instantaneous effective apertures Aeff to UHE neutrinos of SKA-low phases 1 and 2 are shown in Fig. 3. The apertures are much smaller than that to cosmic rays, both because only 20% of the primary neutrino energy is characteristically given to hadronic cascades, and because the majority of neutrinos interact too deeply in the Moon to be detected (field absorption length at 100 MHz is of the order of 200 m (Olhoeft & Strangway 1975). Therefore, an appreciable sensitivity is not reached until well above 1020 eV, so that the SKA will not be sensitive to neutrinos from the GZK process. However, the SKA will be able to probe “top-down” models of cosmic ray production, as shown in Fig. 3 (right) for a nominal 1000 hr observation. While the projected sensitivity of ARA (Allison et al. 2012) and ARIANNA (Barwick et al. 2014) are not shown, their target sensitivity is to the GZK flux around 1018 eV, and is not expected to be competitive with SKA-low phase 2 at energies much above 1020 eV. |
Processed_The_lunar_Askaryan_technique_with_the_Square_Kilom.txt | effective aperture of approximately 105 km2 sr to cosmic rays at 1020 eV. This gives it the poten- tial to detect an unprecedented number of UHECR. Even folding in the lunar visibility of 29% (assumed elevation limit of 30◦) from the site of the SKA-low core in the Murchison Radio As- tronomy Observatory in Western Australia, the expected annual detection rate of ∼ 50 cosmic rays with energy above 56 EeV gives a method to increase detection statistics beyond that of the current Pierre Auger Observatory (Aab et al. 2015) and Telescope Array samples (Abbasi et al. 2014). The only proposed experiments which could rival this detection rate are JEM-EUSO, to be placed on-board the international space station (Takahashi et al. 2009), and the LORD mission (Ryabov, Gusev & Chechin 2013). |
Processed_The_lunar_Askaryan_technique_with_the_Square_Kilom.txt | The angular and energy resolutions of observations with SKA phase 2 will not be competitive with those of current ground-based experiments, although they are expected to improve beyond the simplistic estimates of 5◦ and 50% respectively presented here. The angular resolution however is already comparable to, or better than, the angular deflection expected from galactic magnetic fields (Armengaud, Sigl & Miniati 2005; Dolag et al. 2005), and would certainly be able to test the observed excess in the ∼ 15◦ surrounding Centaurus A (Aab et al. 2015). In general, the SKA will be sensitive to objects lying within ∼ 45◦ of the lunar orbit, i.e. more than half the sky, in a band centred on the Moon’s orbit. For SKA-low, this band also includes the Galactic Centre and M87. |
Processed_The_lunar_Askaryan_technique_with_the_Square_Kilom.txt | In order to compete with dedicated experiments such as JEM-EUSO or LORD (or, indeed, the 10-year exposure of Pierre Auger), it will be vital to observe the Moon with SKA phase 2 whenever it is visible. This will require implementing a fully commensal observation mode, so that lunar observations will not compete with the many other science goals of the SKA. Such a mode will be feasible for SKA phase 2, due to the planned ability to form multiple beams per station (see Sec. 2). For SKA phase 1 however, there will be more competition for the available beams, so the target is to perform engineering studies for phase 2, and observations targeting models of the UHE neutrino flux. For each phase, specialised pulse detection hardware will be built, analogous to the ‘Bedlam Board’ produced for the LUNASKA experiments at Parkes (Bray, Ekers & Roberts 2013), and the ‘ultra-high-energy particles’ (UHEP) mode of LOFAR developed by the NuMoon collaboration (Singh et al. 2012) — for further details, see the co-contribution by Bray et al. (2015). |
Processed_The_lunar_Askaryan_technique_with_the_Square_Kilom.txt | The low-frequency component of the SKA will be a powerful instrument for performing ultra- high-energy particle astronomy. In order to enable these observations with a giant radio array, members of previous lunar experiments have joined forces to form the SKA High Energy Cosmic Particles focus group, together with an experiment to perform precision measurements of EAS. Using the lunar Askaryan technique, phase 1 of SKA-low will be able to test models of top-down particle production, while phase 2 will be able to detect an unprecedented number of cosmic rays at energies above 56 EeV where indications of anisotropy are observed. |
Processed_The_lunar_Askaryan_technique_with_the_Square_Kilom.txt | to promote the lunar technique, and as founding member of the LUNASKA project, he is in large part responsible for the existence of the HECP group. |
Processed_The_lunar_Askaryan_technique_with_the_Square_Kilom.txt | Aab, A. et al. (Pierre Auger Collaboration), 2015, Phys.Rev.D, 91, 092008. Aab, A. et al. (Pierre Auger Collaboration), 2010, ApJ, 804, 15. Abbasi, R.U. et al. (Telescope Array Coll.), 2014, ApJ, 790, L21 Abraham, J. et al., 2010, Phys.Lett.B, 685, 239. Abraham, J. et al. (Pierre Auger Coll.), 2007, Science, 318, 938 Allard, D., Ave, M., Busca, N. et al., 2006, JCAP, 9, 005. Allison, P. et al. (ARA Collaboration), 2012, Astropart.Phys., 35, 457. Alvarez-Muñiz, J., Marqués, E., Vázquez, R.A., & Zas, E., 2006, Phys.Rev.D, 74, 023007 Alvarez-Muñiz, J., Zas, E., 1998, Phys.Lett.B, 434, 396. Armengaud, E., Sigl, G., & Miniati, F., 2005, Phys.Rev.D, 72, 043009. Askaryan, G.A., 1962, Sov. Phys. JETP, 14, 441 Barwick, S.W. et al. (ARIANNA Collaboration), 2014, arXiv:1410.7352. Beresnyak, A.R., Dagkesamanskii, R.D., & Zheleznykh, I.M. et al., 2005, Astron.Rep., 49, 127 Berezinsky, V., Sabancilar, E. & Vilenkin, A., 2011, Phys.Rev.D, 84, 085006 Bray, J.D. et al., 2015, APh, 65, 22. Bray, J.D. et al., 2015, aska.conf, 144 [arXiv:1408.6069B]. Bray, J.D. et al., 2015, these proceedings (I.D. 597). Bray, J., Ekers, R.D., & Roberts, P., Exp.Astronomy, 36, 155. Buitink, S.W. et al., 2010, A&A, 521, A47 Dagkesamanskii, R.D. & Zheleznykh, I.M., 1989, Sov.Phys.JETP Lett., 50, 259. Dewdney, P.E. et al., 2013, “SKA1 System Baseline Design”, SKA-TEL-SKO-DD-001 Dolag, K. et al., 2005, JCAP, 0501, 009. Gayley, K.G., Mutel, R.L. & Jaeger, T.R., 2009, ApJ, 706, 1556 Gorham, P.W. et al., 2004, Phys.Rev.Lett., 93, 041101 Gorham, P.W. et al. (ANITA Coll.), 2010, Phys.Rev.D, 82, 022004 Gorham, P.W. et al. (ANITA Coll.), 2012, Phys.Rev.D, 85, 049901 Huege, T. et al., 2015, these proceedings (I.D. 309). Jaeger, T.R., Mutel, R.L. & Gayley, K.G., 2010, Astropart.Phys., 43, 293 James, C.W. & Protheroe, R.J., 2009, Astropart.Phys., 30, 318 James, C.W. et al., 2010, Phys.Rev.D, 81, 042003 Kravchenko, I. et al., 2012, Phys.Rev.D, 85, 062004. Lunardini, C. & Sabancilar, E., 2012, Phys.Rev.D, 86, 085008 Olhoeft, G. & Strangway, D., 1975, Earth Planet.Sci.Lett., 24, 394. Ryabov, V.A., Gusev, G.A. & Chechin, V.A., 2013, J.Phys.Conf.Series, 409, 012096. Scholten, O. et al., 2006, Astropart.Phys, 26, 219. Singh, K. et al. (LOFAR Coll.), 2012, NIMA, 664, 171. Spencer, R.E., Macfarlane, A., Mills, O. & Piccirillo, L., 2010, Proc. 10th EVN Symposium, 097. Takahashi, Y. et al. (JEM-EUSO Coll.), 2009, New J.Phys., 11, 065009. |
Processed_Australia_Telescope_Compact_Array_Observations_of_.txt | We report results of the first phase of observations with the Australia Telescope Compact Array (ATCA) at 5 and 9 GHz of the fields around 411 γ-ray sources with declinations < +10◦ detected by Fermi but marked as unassociated in the 2FGL catalogue. We have detected 424 sources with flux densities in a range of 2 mJy to 6 Jy that lie within the 99 per cent localisation uncertainty of 283 γ-ray sources. Of these, 146 objects were detected in both the 5 and 9 GHz bands. We found 84 sources in our sample with a spectral index flatter than -0.5. The majority of detected sources are weaker than 100 mJy and for this reason were not found in previous surveys. Approximately 1/3 of our sample, 128 objects, have the probability of being associated by more than 10 times than the probability of being a background source found in the vicinity of a γ-ray object by chance. We present the catalogue of positions of these sources, estimates of their flux densities and spectral indices where available. |
Processed_Australia_Telescope_Compact_Array_Observations_of_.txt | Analysis of the first two years of Fermi/LAT observations yielded a catalogue of 1872 γ-ray sources (Nolan et al. 2012, hereafter 2FGL). Of these, 70 per cent have associations with blazars, pulsars, supernova remnants or other objects. The procedure for assigning associations is described in full detail in Ackermann et al. (2012). However, the 2FGL catalogue still does not provide associations for 573 γ-ray sources. The positional accuracy of Fermi ranges from 0.1 ar- cmin to 16.6 arcmin with a median 1σ uncertainty of 2.0 ar- cmin. These large position errors prevent finding associa- tions by direct matching the 2FGL against optical or in- frared catalogues. |
Processed_Australia_Telescope_Compact_Array_Observations_of_.txt | likely the same object. Therefore, systematic high resolution radio observations of unassociated Fermi objects promise to find all γ-ray sources associated with radio-loud AGNs and shrink the list of objects that remain unassociated with an astrophysical source. |
Processed_Australia_Telescope_Compact_Array_Observations_of_.txt | It was shown by Lister et al. (2009) and Kovalev (2009) that γ-ray fluxes and 8 GHz radio flux densities from re- gions smaller than 5 mas are correlated. Source variability confirms that both radio and γ-ray emission comes from parsec-scale regions. Therefore, a flux-limited sample of γ- ray AGNs, such as 2FGL, should be contained within a flux-limited sample of compact radio sources. Compact radio sources from a flux-limited catalogue found within the posi- tion error ellipses of sources from a γ-ray flux limited cata- logue can be considered with a high confidence as the same objects, provided that the probability of finding a back- ground radio-loud source within the error ellipse is small. As the search area becomes smaller, weaker radio-loud AGNs can be associated with a γ-ray source. |
Processed_Australia_Telescope_Compact_Array_Observations_of_.txt | In order to evaluate the probability of finding a back- ground source in a given search area, we investigated the cumulative all-sky catalogue of compact radio sources detected with VLBI in the absolute astrometry mode (Petrov and Kovalev, in preparation). This catalogue1, as of December 2012, had 7215 objects detected in numer- ous VLBI surveys over the last several decades: Very Long Baseline Array (VLBA) Calibrator Survey (Beasley et al. 2002; Fomalont et al. 2003; Petrov et al. 2005, 2006; Kovalev et al. 2007; Petrov et al. 2008), Long Baseline Ar- ray (LBA) calibrator survey (Petrov et al. 2011b), VLBA Galactic Plane Survey (Petrov et al. 2011a), European VLBI Network (EVN) Galactic plane survey (Petrov 2012), VLBA imaging and polarimetry survey (Taylor et al. 2007; Petrov & Taylor 2011), regular VLBA geodetic observations (Petrov et al. 2009; Pushkarev & Kovalev 2012), and ongo- ing VLBI observations of 2MASS galaxies (Condon et al. 2011). Figure 1 shows the dependence of the logarithm of the number of sources in the whole sky N with a flux den- sity greater than S, as a function of the logarithm of S de- termined as the median correlated flux density at baseline projection lengths in a range of 100–900 km at 8 GHz. We see that the dependence can be approximated as N (S) ≈ −1.237 in a range of [0.18, 5] Jy. We interpret the de- 327 · S viation of log N (log S) from the straight line below 180 mJy as evidence of incompleteness. Assuming the parent popu- lation remains the same for sources as weak as 1 mJy, we can extrapolate the number of compact sources in the celes- tial sphere derived for the range [0.18, 5] Jy to the range [1, 180] mJy. |
Processed_Australia_Telescope_Compact_Array_Observations_of_.txt | Figure 1. Dependence of the decimal logarithm of the number of compact sources versus the decimal logarithm of their minimum correlated flux density from regions smaller than 50 mas at 8 GHz. |
Processed_Australia_Telescope_Compact_Array_Observations_of_.txt | Figure 2. Maximum flux density at 8 GHz of a background com- pact source to be found with the probability of 10 per cent within an error ellipse with a given 1σ standard deviation. |
Processed_Australia_Telescope_Compact_Array_Observations_of_.txt | Figure 3. Distribution of position offsets normalised by their standard deviations between 770 Fermi sources and their coun- terparts detected with VLBI. The solid blue lines shows Rayleigh distribution with σ = 1. |
Processed_Australia_Telescope_Compact_Array_Observations_of_.txt | and σmaj, σmin, and θ are semi-major, semi-minor axes and position angle of the 2D Normal distribution that describes errors of source localisation. We derived σmaj, σmin from the so-called “95% confidence source semi-major and semi- minor axes” reported in table 3 of the 2FGL catalogue by scaling them by 1/e. The distribution of normalised arcs be- tween the VLBI and 2FGL positions follows a Rayleigh dis- tribution with σ = 1 very closely (Figure 3). The histogram is best fit into a Rayleigh distribution with σ = 1.026. Since VLBI positions are five orders of magnitude more accurate than the positions reported in 2FGL, they are considered as true. We conclude that the 2FGL position errors are realistic and can be used for a probabilistic inference. |
Processed_Australia_Telescope_Compact_Array_Observations_of_.txt | These estimates show that radio observations are an ef- ficient way to identify AGNs associated with γ-ray sources. According to our experience, it is not difficult to detect sources with flux densities of several mJy using radio inter- ferometers. As we have shown in Figure 1, the cumulative catalogue of compact radio sources detected with VLBI is complete only to the level of 180 mJy. Sources with corre- lated flux densities between 2 and 180 mJy may not have been marked as associations with 2FGL objects because they were missing from VLBI surveys. However, not all associa- tions of γ-ray sources in 2FGL are made on the basis of VLBI observations. In the absence of VLBI observations, the source spectrum can be used as a proxy. It was found a long time ago (Kellermann & Pauliny-Toth 1969) that sources with a spectral index flatter than −0.5 tend to be compact as they are dominated by emission from the core. The prob- lem is that existing catalogues are not complete enough to determine the spectrum in the range of flux densities from 1–100 mJy, especially in the southern hemisphere. |
Processed_Australia_Telescope_Compact_Array_Observations_of_.txt | Figure 4. The distribution of 573 2FGL unassociated sources. The area of this survey is shown with grey colour. |
Processed_Australia_Telescope_Compact_Array_Observations_of_.txt | tual goal of the program is to find all AGNs with correlated flux densities brighter than 2 mJy at 8 GHz from regions smaller than 50 mas within the areas of 99 per cent prob- ability of Fermi localisation uncertainties. Association with a VLBI source automatically improves positions of γ-ray objects from arcminutes to milliarcseconds, allowing for an unambiguous association with sources at other wavelengths, for instance, optical. Combined with results of on-going pro- grams to find pulsars associated with Fermi sources, for in- stance Barr et al. (2013), we expect the list of unassociated sources to shrink. At the moment we do not know whether the list will shrink to zero or a population of radio-quiet γ-ray sources will be found. |
Processed_Australia_Telescope_Compact_Array_Observations_of_.txt | In this paper we report early results of the first step of the program of observations of Fermi unassociated sources: observations with the Australia Telescope Compact Array (ATCA) of 411 sources with δ < +10◦. A brief overview of the program is given in section 2. The data analysis proce- dure is described in section 3. Results, the catalogue of 375 potential associations for 2FGL sources is given in section 4 followed by concluding remarks. |
Processed_Australia_Telescope_Compact_Array_Observations_of_.txt | The areas of 99 per cent probability of 2FGL localisation, which corresponds to 3.035σ, are typically 4–6 arcmin. They are still too large for a blind search with VLBI using tradi- tional approaches. Therefore, we organised observations in several stages. In the first stage we observed the sources with connected interferometers, the Very Large Array (VLA) in the northern hemisphere, and ATCA in the southern hemi- sphere. Observations are made in the remote wings of the wide-band receiving system that covers 4.5–10.0 GHz. The sky distribution of target sources is presented in Figure 4. |
Processed_Australia_Telescope_Compact_Array_Observations_of_.txt | We split the source list into two parts. A list of 411 ◦ sources with declinations lower than +10 was observed with the ATCA. A list of 215 sources with declinations higher than 0◦ was observed with the VLA. This list includes 35 ◦ overlapping sources in the declination band [0, +10 ] and 18 sources tentatively associated with supernova remnants and pulsar wind nebulae. Results of the VLA observations will be reported in a separate paper (F. Schinzel et al., (2013) in preparation). |
Processed_Australia_Telescope_Compact_Array_Observations_of_.txt | spectral indices. The spectral index is defined as S ∼ f α, where S is the flux density, and f is the frequency. Sources with a spectral index α greater than −0.5 are classified as flat spectrum sources and are therefore considered as likely associations. Bright detected sources will be followed up with the VLBA and LBA. Detection of emission at par- sec scales will firmly associate them with the compact core regions of AGNs. |
Processed_Australia_Telescope_Compact_Array_Observations_of_.txt | We observed a list of 411 target sources with ATCA on September 19–20 2012 for 29 hours. We used the automatic scheduling software sur sked originally developed for VLBI survey experiments (Petrov et al. 2011b). In order to max- imise the observing efficiency, we reduced the correlator cy- cle time from the usual 10 seconds to 6 seconds, and observed in the mosaic mode to minimise correlator overheads when changing sources. We tried to observe each target source in three scans of 24 seconds each, separated by at least three hours in a sequence that minimises slewing time. In fact, we observed 80 sources in 4 scans, 239 sources in 3 scans and 92 sources in 2 scans. At the beginning of the experiment we observed the primary amplitude calibrator 1934−638 for 20 minutes. The automatic scheduling process checked that for every target source a phase calibrator was inserted into the schedule that satisfied two criteria: that the arc between the ◦ target and the calibrator should be less than 20 and a cal- ibrator should be observed within 20 minutes of the target. ◦ We used a pool of 1464 compact sources with δ < +20 and correlated flux densities > 200 mJy. The scheduling process picked up 207 sources from this pool. Many adjacent target scans reused the same phase calibrator. Each calibrator scan was observed for 18 seconds. |
Processed_Australia_Telescope_Compact_Array_Observations_of_.txt | The array was in the H214 hybrid configuration with baselines ranging from 31–214 meters between the inner 5 antennas and ∼4.4 km between CA06 and the inner anten- nas. We observed simultaneously in two bands, both 2GHz wide, centred on 5.5 and 9.0 GHz. During data reduction the band edges were excluded and the ranges [4.58, 6.42] GHz (hereafter 5 GHz band) and [8.09, 9.92] GHz (hereafter 9 GHz band) were used. The data were recorded in both polarizations with the Compact Array Broadband Backend (Wilson et al. 2011). |
Processed_Australia_Telescope_Compact_Array_Observations_of_.txt | The task uvsfit estimates source parameters through the minimisation of the least-squared differences between the measured visibilities and those computed from a model source or sources. |
Processed_Australia_Telescope_Compact_Array_Observations_of_.txt | Given a set of observed visibilities, each measured at a point (u, v) that is the projected baseline vector ex- pressed in wavelengths, and at a frequency f , we write Vi = V (ui, vi, fi). There are n = nb × nf visibility mea- surements, where nb and nf are the number of baselines and the number of frequency channels respectively. |
Processed_Australia_Telescope_Compact_Array_Observations_of_.txt | Given a model source, we can compute model visibilities Mi = M (ui, vi, fi) at each point in (u, v, f ) space for which we have visibility measurements. The model visibilities can be calculated as the sum of visibilities for several model sources. The fitting process is the adjustment of model |Vi − Mi|2. source parameters to minimise the quantity Both tasks use the Levenberg-Marquardt algorithm for func- tion minimisation. The diagonal terms of the covariance ma- trix are returned as the variances in fitted parameter values. For a point source with flux-density S and position (l, m) relative to the phase centre, uvfit computes each model visibility as: M (u, v) = S(cos θ + i sin θ), where θ = 2π(ul + vm). |
Processed_Enabling_the_Multi-User_Generalized_Degrees_of_Fre.txt | Abstract—There has been major progress over the last decade in understanding the classical interference channel (IC). Recent key results show that constant bit gap capacity results can be obtained from linear deterministic models (LDMs). However, it is widely unrecognized that the time-invariant, frequency- flat cellular channel, which contains the IC as a special case, possesses some additional generalized degrees of freedom (GDoF) due to multi-user operation. This was proved for the LDM cellular channel very recently but is an open question for the corresponding Gaussian counterpart. In this paper, we close this gap and provide an achievable sum-rate for the Gaussian cellular channel which is within a constant bit gap of the LDM sum capacity. We show that the additional GDoFs from the LDM cellular channel carry over. This is enabled by signal scale alignment. In particular, the multi-user gain reduces the interference by half in the 2-user per cell case compared to the IC. |
Processed_Enabling_the_Multi-User_Generalized_Degrees_of_Fre.txt | methods and insights into the Gaussian counterpart include [3], [7], [5], and [8]. In many cases, the special properties of the LD models provide a possibility of interference alignment on the signal scale. The investigation of the Many-to-One channel [3], was the first to convert this interference alignment to the Gaussian case. Here, lattice coding was used in addition to a layered encoding and successive decoding strategy. The strategy splits the total usable power in many power packages, where each one gets encoded separately by a lattice code. Also each package is viewed as a single Gaussian channel in the decoding process. Since lattice codes obey certain group properties, the addition of many interferers is seen as another codeword at the receiver. This enables interference alignment on the signal scale. The idea was picked up in [5] and [8] which investigated the mixed Many-to-One and One-to-Many models. In these investigations LDM interference alignment strategies were also translated to the Gaussian channel, re- sulting in new achievable rates. Another interesting model is the interfering MAC (IMAC) and interfering BC (IBC). These terms were defined in [9], were it was shown that, using delay properties, a form of multi-user gain can be achieved. Furthermore, with k−users, an interference-free transmission is proven by means of degrees of freedom. Another notable investigation is [10], where the capacity region of the IMAC was characterized for several strong interference cases. Also an upper-bound was provided for the whole interference range. A different strategy was used in [11], where the LD IMAC and LD IBC were investigated. Based on the LD MAC-P2P investigations in [12] and [13], multi-user gain was shown and the sum capacity has been found for the weak interference case. |
Processed_Enabling_the_Multi-User_Generalized_Degrees_of_Fre.txt | Contributions: In this paper we convert the achievable scheme of [11] to the Gaussian models by modifying the methods of [3] and [5]. We show that with layered lattice codes, signal scale alignment can be achieved in the IMAC and IBC. We demonstrate that multi-user gain can be enabled which basically reduces the interference by half (2-user weak interference case) in comparison to the IC. To the best of our knowledge, this is the first demonstration of signal scale alignment and multi-user gain in time-invariant and frequency- flat cellular channel models. |
Processed_Enabling_the_Multi-User_Generalized_Degrees_of_Fre.txt | where i and j denote the transmitter and receiver cell, respec- tively and k is the user index. Z j ∈ CN (0, 1) is assumed to be zero mean and unit variance Gaussian noise. Also each transmitted signal has an associated average power constraint E{|Xik|2} ≤ P . |
Processed_Enabling_the_Multi-User_Generalized_Degrees_of_Fre.txt | where Z j k ∈ CN (0, 1) is again a zero mean and unit variance Gaussian random variable and X has an associated average power constraint E{|Xi|2} ≤ P . |
Processed_Enabling_the_Multi-User_Generalized_Degrees_of_Fre.txt | We also introduce two parameters αi, βi which combine these ratios with SNRi1 = Pi, SNRi2 = P βi i = i P αj . These parameters correspond to α, β which are used i in the LDM channel model [11]. Now we can restrict the investigation to the weak interference regime defined through αi ∈ [0, 1 2 ] and αi < βi. Note that in the case of αi ≥ βi the IMAC as well as the IBC relapse into the IC in terms of capacity. In both models we assume Pi > 1. For convenience we use the standard terms: common and private signal, for the part which is seen at both cells and the part which is only received in the intended cell, respectively. Also note that the following techniques work for all channel parameters in the defined regime, except for a singularity at βi = 1 in which the additional gain for that corresponding cell will be zero. |
Processed_Enabling_the_Multi-User_Generalized_Degrees_of_Fre.txt | where l(p, q) := (cid:98) p q (cid:99) for q > 0 and l(p, 0) = 0. The optimal construction of the coding matrices is a direct extension of the MAC-P2P scheme in [12]. The maximum sum rate is achieved by orthogonal coding. This means that there is either an independent bit or nothing send from a specific level. At the receiver there is no overlap with levels used by another transmitter. This technique can be interpreted as interference alignment, because the levels for the LD IMAC, are chosen in the following way. Both interference signals of a cell align at the receiver as much as possible in the unused levels of the direct transmission. In the LD IBC, levels of the common part are chosen such that they align at the receivers of the unintended cell as much as possible on levels, which receive the private message intended for the opposite receiver. See figure 2 for an example. |
Processed_Enabling_the_Multi-User_Generalized_Degrees_of_Fre.txt | region of the lattice has an average power per dimension of θl and is good for channel coding. Moreover, aligning sub-signals use the same code (with independent shifts). |
Processed_Enabling_the_Multi-User_Generalized_Degrees_of_Fre.txt | a lattice code (γΛC + v) ∩ S exists with arbitrary small error probability. This code consists of a lattice ΛC ∈ Rn, a scaling factor γ ∈ R, a translation v ∈ Rn and a spherical shaping region S ⊂ Rn with power P per dimension. N denotes the noise variance per dimension. |
Processed_Enabling_the_Multi-User_Generalized_Degrees_of_Fre.txt | where the result of lemma 1 is used. With the definition of α = 0.5 and nj ik|2P (cid:101) one can see, that the proposed scheme can achieve the upper bound in (6) within a constant gap of 2(7 + log 5) Bits. |
Processed_Enabling_the_Multi-User_Generalized_Degrees_of_Fre.txt | since α ≤ 1, P ≥ 1 and therefore P 1−α ≥ 1. The IBC case follows on the same lines. |
Processed_Enabling_the_Multi-User_Generalized_Degrees_of_Fre.txt | Proof: The total achievable sum rate is the summation over the interference alignment structure RIi , and additional parts RAi and RRi. We show the proof exemplary for the IMAC channel in case that α < 0.5 and (cid:98)L(cid:99) = even. The IBC proof follows on the same lines. Also note that the rates for the alignment structure are solely of the parts below P αj . This is necessary for ensuring decodability of both, direct and interference sub-signals (see lemma 1). |
Processed_Enabling_the_Multi-User_Generalized_Degrees_of_Fre.txt | One can see that the achievable rate of the Gaussian channel is within a constant-gap of 2(2 + log 4) + 6(cid:98)L1(cid:99) + 6(cid:98)L2(cid:99) of the LDM rate using the correspondence nj ik|2P (cid:101). See figure 4 for a graphical comparison. |
Processed_Enabling_the_Multi-User_Generalized_Degrees_of_Fre.txt | Figure 4. Illustration of the achievable rate in comparison to the LD case. Direct signals and interference is assumed to be symmetrical for clear presentation and β = 7/8. At (cid:98)L(cid:99) = L = even, the achievable scheme reaches the LDM upper bound within a constant gap. |
Processed_Enabling_the_Multi-User_Generalized_Degrees_of_Fre.txt | IMAC and IBC channel can be upper bounded by the LDM channel bounds within a constant gap. We believe that this investigation provides crucial insights for cellular networks in general and can be helpful in future investigations. |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | Publications of the Astronomical Society of Australia (PASA) c(cid:13) Astronomical Society of Australia 2021; published by Cambridge University Press. doi: 10.1017/pas.2021.xxx. |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | Abstract One of the major science goals of the SkyMapper Survey of the Southern Hemisphere sky is the determination of the shape and extent of the halo of the Galaxy. In this paper we quantify the likely efficiency and completeness of the survey as regards the detection of RR Lyrae variable stars, which are excellent tracers of the halo stellar population. We have accomplished this via observations of the RR Lyrae-rich globular cluster NGC 3201. We find that for single epoch uvgri observations followed by two further epochs of g, r imaging, as per the intended three-epoch survey strategy, we recover known RR Lyraes with a completeness exceeding 90%. We also investigate boundaries in the gravity-sensitive single-epoch two-color diagram that yield high completeness and high efficiency (i.e., minimal contamination by non-RR Lyraes) and the general usefulness of this diagram in separating populations. |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | The formation and evolution of galaxies like our own Milky Way, and the shape and extent of its dark mat- ter halo, are subjects of significant importance to As- trophysics. In that context, we can improve our knowl- edge of galaxy formation processes by studying the sub- structures and stellar streams in the Galactic halo. RR Lyrae stars (RRLs) make excellent halo tracers for such studies as they have well established absolute magni- tudes and thus measure accurate distances. Moreover, because of their intrinsic luminosity, they are detectable to the edge of the Galaxy with moderate sized tele- scopes. RRLs are old and metal poor, and are relatively common and easy to recognise from their characteristic colours and light curves. |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | Various studies have already been performed using RRLs to aid our understanding of galaxy formation. For example, Akhter et al. (2012, and references therein) have studied the break radius, i.e., the Galactocentric radius at which the density profile changes slope, in the spatial density distribution of RRLs in the Milky Way stellar halo. Overdensities and sub-structures also provide important information and RRLs have been used extensively in this area (e.g., Vivas et al. 2003; 2001; Newberg et al. Ibata et al. 2003; Yanny et al. 2003; Zinn et al. 2004; Belokurov et al. 2006; Vivas & Zinn 2006; Duffau et al. |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | taken with 1-3 days difference (Keller et al. 2007). The multi-epoch data is intended to find variables and here we investigate their usefulness for detecting RRLs. The SkyMapper filter set is also designed to provide stel- lar surface gravity information. Consequently, we also investigate the utility of the gravity-sensitive diagram for separating genuine RRL candidates from potential main sequence contaminants such as eclipsing binaries. This paper demonstrates a method to find RRLs in the SkyMapper survey data and investigates the effi- ciency and completeness by studying an RRL-rich glob- ular cluster. Section 2 describes the observations and the data reduction, including the photometric calibra- tion procedure. In Section 3, we outline the RRL se- lection process which uses variability in the g and r magnitudes in the multi-epoch data together with other colour information. The outcome of our variable selec- tion process in comparison with the known variables in the cluster field is examined in Section 4. In Section 5, we quantify the efficiency and completeness of the RRL selection process. Finally, we discuss our results in Sec- tion 6. |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | The RRL-rich Milky Way globular cluster NGC 3201 was selected as a test object to measure the utility of the SkyMapper Survey in identifying RRL stars. Key parameters for the cluster are provided in Table 1 where “rcore” and “rh” are the core radius and the half-light radius in arcmin. The values in the table are taken from the current on-line version of the Harris (1996) cata- logue2 except for the number of RRLs in the cluster field which comes from the on-line version of the Globular Cluster Variable Star Catalogue (Clement et al. 2001)3. We reproduced the intended three-epoch survey strategy by observing a field centred approximately on the cluster three times. The first two epochs were sepa- rated by 1.93 hours and the second and the third epochs separated by 2.99 days. Observations were made using the g and r filters at all three epochs and with the u, v, and i filters at the first epoch only. |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | were processed (bias subtracted, flat-fielded and amp- combined) with a preliminary version of the SkyMapper Science Data Pipeline System at the ANU Supercom- puting Facility, Canberra. |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | We first separated each frame into the 32 individual CCD images. These were then fit with a World Coor- dinate System (wcs) using custom SkyMapper software (Keller 2012, priv. comm.). The cluster was centred on CCD-31 and since the 34′ × 17′ field of this CCD more than encompasses the cluster, subsequent analysis used only the frames from this CCD. The u filter image was chosen to define the stellar sample, which contains both cluster and field stars, as this image has the bright- est limiting magnitude and thus stars detected on this frame are likely to be common to all the other frames. Use of IRAF task daof ind on the u frame, with a de- tection threshold set to maximise the number of stars and minimise the number of spurious detections, re- sulted in the detection of 817 stars. The same stars were then identified on other frames by using the IRAF tasks xy2sky and sky2xy. |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | We then calculated for each star the radial distance rd from the center of the cluster on the g first epoch frame, which we denote by g1 (more generally we refer to frames by “f n” where f is the filter and n is the epoch). Visual inspection of the g frames showed that within a radius rd < 300 pixels, i.e., 150 arcsec, image crowding and blending was significant, while outside of this radius the majority of the stellar images were rela- tively isolated. Consequently, to minimise the effects of crowding on the photometry, the stars within a radius of 300 pixels from the cluster centre were excluded from the analysis, leaving a sample of 600 stars. Aperture photometry, using the IRAF task phot and an aperture radius of 4 pixels, was then performed on each frame to generate sets of instrumental magnitudes. These out- put lists were then compared to include only those stars with valid measurements on all the data frames. This left a final sample of 517 stars that will be referred to as the “all stars” sample. |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | We then corrected the g2 and g3 magnitudes to the g1 instrumental system, and similarly the r2 and r3 mag- nitudes to r1 system. To do this we plotted g2 − g1 and g3 − g1 against g1 and determined the mean magnitude differences from the g1 system using ∼150 bright stars with g1 < 16 mag (gstd ≈ 14.2). The offsets to bring g2 and g3 on to the g1 system were 0.112 ± 0.001 mag and 0.011 ± 0.001 mag respectively, where the error is the standard error of the mean. Offsets were determined in a similar fashion for r2 and r3 as −0.043 ± 0.001 mag and 0.03 ± 0.001 mag respectively, using ∼180 bright stars with r1 < 16 mag (rstd ≈ 14.0). |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | aEntries are from the current online version of the Harris (1996) catalogue except for the number of RRLs which comes from the online version of the Clement et al. (2001) catalogue. |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | The values of α and ZP , along with the rms about the fit, are given in Table 2. |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | A similar process was then used to determine off- sets to place the instrumental magnitudes v1 and i1 on a standard system. As for gstd, the standard mag- nitudes vstd and istd were calculated from the slopes and intercepts derived from the least squares fits to the magnitude difference-colour relations (B − v1) ver- sus (B − V ) and (I − i1) versus (V − I), applied at (B − V )0 = 0.0 and (V − I)0 = 0.0, respectively. We take E(V − I)/E(B − V ) as 1.25 for (B − V )0 = 0.0 (Dean et al. 1978) and thus (V − I)0 = 0.0 corresponds to (V − I) = 0.30. We note further that since the (B − v1) versus (B − V ) relation is somewhat non- linear, the α and ZP values given in Table 2 have been determined from the fit to only the bluer stars (B − V < 0.8). |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | sequently, we used the magnitude difference-colour re- lations (B − u1) versus (B − V ) and (V − r1) versus (V − I) to determine the offsets to generate standard- ised Skymapper magnitudes ustd and rstd. The process works well for the r magnitudes, as demonstrated by the low rms for the fit given in Table 2. However, even restricting the fit to the bluer stars, the (B − u1) ver- sus (B − V ) relation shows significant scatter, resulting in a larger rms value compared to the other relations and thus larger uncertainties in the α and ZP values. Consequently, while we assert that the calibrations for g, i, r and v are relatively reliable, the standardised u magnitudes derived here may not be exactly on the fi- nal SkyMapper photometric system. Consequently, the magnitudes and colours discussed in this work should be considered as illustrative only, although none of the sub- sequent analysis is affected by our choice of zero points. We now proceed to correct our standardised magni- tudes for the effects of reddening. We assume that the cluster value of E(B − V ) = 0.24 applies to all stars in the field whether they are cluster members or not. Bessell (2012, priv. comm.) has used the response func- tions of the SkyMapper filter system, input spectral en- ergy distributions from model atmospheres, and a stan- dard absorption curve to determine absorption and red- dening coefficients for the SkyMapper filter system. In particular, we have Ag = 1.19 AV , so that with AV = 3.2 E(B − V ), the absorption correction to the stan- dardised g magnitudes is 0.91 mag. |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | g − i and u − v colours are then 0.35 and 0.06 mag, re- spectively. In the subsequent discussion all standardised magnitudes and colours have been reddening corrected. We note in passing that for the SkyMapper survey itself, reddening values will be taken from standard sources such as Schlegel et al. (1998, see also Majewski et al. (2011) and the references therein), and that most field RRLs will be at sufficiently large distances and heights above the Galactic plane that the uncertainties in the reddening corrections will be small. |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | We now describe the process through which the candi- date RRLs are selected. In brief, the first step is identi- fying stars that appear to vary significantly across the 3 epochs of g and r data. The second step uses the fact that the g and r data were taken almost simultaneously: as a consequence the variations in the g and r magni- tudes for genuine variable stars should be tightly cor- related. The third and final step uses the colour of the candidates and their location in a gravity-sensitive dia- gram to separate the RRL candidates from other classes of potential variables. |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | For each of the 517 stars in the “all stars” sample we calculated the mean magnitudes from the g and r ob- servations at the 3 epochs, and also the standard devi- ations of the magnitudes. The mean magnitude values were then grouped into bins (all stars brighter than hgi, hri = 14.0 mag, then bins of width 0.5 mag) and the mean and standard deviation of the individual 3-epoch standard deviations for the stars in each bin calculated. We denote these quantities by hσiin and σn(σi), where n represents the bin number and i the set of stars in each bin. Then, for each magnitude bin, we removed the stars with 3-epoch standard deviations greater than 3×σn(σi), flagging them as candidate variables. The values of hσni and σn(σi) are then recalculated and fur- ther “3σ” deviants classed as candidate variables. The process is continued until σn(σi) ceases to change sig- nificantly. |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | By this process we identified 47 stars as candidate variables from the g frames and 49 from the r frames. In total there are 51 candidates at this stage, of which 45 are in common between the two sets. The output is illustrated in the panels of Fig. 1 in which the indi- vidual 3-epoch standard deviations are plotted against the mean magnitudes for the g (upper panel) and r data (lower panel). The candidate variables are shown as filled circles while the stars classified as non-variable are represented by plus-signs. |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | Figure 1. The standard deviation of the magnitude values from the 3 observation epochs is shown as a function of the mean mag- nitudes for the g filter (upper panel) and for the r filter (lower panel). Using an iterative sigma clipping process 47 g and 49 r candidate variables are identified. These are shown as blue solid circles while the non-variables are plotted as red plus-signs. |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | the magnitude differences r1 − r2 and g1 − g2 for the first and second epochs, and similarly the first and third epoch magnitude differences r1 − r3 and g1 − g3. The relationships between these magnitude differences are shown in the panels of Fig. 2. It is clear that most of the candidate variables identified in the previous section are true variables as they exhibit tight linear relations be- tween the magnitude differences. We note though that for RRLs, the substantial changes in surface gravity that occur at particular phases in the pulsation cycle might result in less well-defined relations if SkyMap- per u or perhaps v magnitude differences are compared to the r magnitude differences for a similar ensemble of photometric measurements. It is also possible that the scatter about the ∆(g), ∆(r) relation might be in- creased if there is a significant metallicity spread among the RRLs in the sample. |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | In Fig. 2 there are, however, a small number of stars identified as candidate variables by the previous selec- tion process that lie away from the correlations de- fined by the majority of the candidate variables. These stars were inspected individually on the frames and it was concluded that the apparent variations were due to varying amounts of contamination of the aperture photometry by close neighbours. These stars are thus probably not true variables and have been discarded from the candidate variables list. Indeed their positions do not match with any of the catalogued variables in the NGC 3201 field. |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | As a result of this process and that of the previ- ous section a final list of 47 candidate variables was generated. For these stars the best fit line for both (∆(g), ∆(r)) data sets indicates a relation ∆(g) = 1.30 ∆(r). A similar relation, ∆(B) =1.32 ∆(R), was found for a sample of 119 MACHO RRLs in the LMC (Keller et al. 2008). |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | Relative to the SDSS, the SkyMapper system possesses an extra filter, v, and a modified u filter. As shown in Keller et al. (2007) using model atmosphere based cal- culations, an appropriate colour combination involving these filters, with (g − i) as a temperature indicator, can provide surface gravity information (see Fig. 11 of Keller et al. 2007). We now investigate whether we can use this colour index to distinguish candidate RRLs (log g ∼ 2.5) from potential main-sequence (log g ∼ 4) con- taminants such as eclipsing binaries. We first note that of our 47 candidate variables, 3 have notably red colours (first epoch standardised and dereddened (g − i) colours exceeding 1.0) and are therefore unlikely to be genuine RRLs. The other 44 variable candidates have relatively “blue” colours, potentially consistent with an RRL clas- sification. |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | Figure 2. Magnitude difference relations for g and r for epochs 1 and 2 (upper panel) and epochs 1 and 3 (lower panel). The red plus-symbols are stars classified as non-variable from their magnitude dispersion across all 3 epochs (see Fig. 1), while the black symbols represent the candidate variables. Most (the filled circles) follow a well defined correlation ∆(g) = 1.30 ∆(r), which is shown as the black solid line. Four stars do not follow this relation and are plotted as triangles – 2 of the four stars (filled triangles) lie off the relation in both panels while the other two (open triangles) are discrepant only in the upper panel. These four stars have been removed from the candidate variables list. |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | Figure 3. The gravity-sensitive index (u − v)0 − 0.2(g − i)0 is plotted against (g − i)0 for the “all stars” sample. It is evident that for (g − i)0 less than ∼0.9 most of the stars fall in one of two distinct branches, which we interpret as indicating the gravity dif- ference between main sequence stars (upper branch) and evolved stars with lower gravities. Based on this diagram we classify 43 of our candidate variables as RRLs (filled black circles) and one as a main-sequence variable (e.g., an eclipsing binary) shown as the filled black square. The three remaining variable candidates (blue filled squares) have redder colours and are not readily classified. |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | In Fig. 3 we plot the gravity-sensitive index (u − v)0 − 0.2(g − i)0 against (g − i)0 for the “all stars” sam- ple using the epoch 1 photometry. For the stars bluer than (g − i)0 ≈ 0.9 there is a clear separation into two distinct branches. Based on Fig. 11 of Keller et al. (2007)5 we interpret the upper branch as being domi- nated by stars with main sequence gravities while the lower branch is presumed to be dominated by lower gravity evolved stars. Of the 44 RRL candidates, 43 fall on this lower branch while one variable candidate falls among the upper branch stars. |
Processed_Finding_RR_Lyrae_Stars_with_SkyMapper:_an_Observat.txt | We can then summarise the outcome of variable star selection processes as the following: using an observa- tional strategy similar to that for the SkyMapper 3- epoch survey, we have identified 47 candidate variables. Of these we classify 43 as candidate RRLs, 1 as a candi- date main sequence variable, while the remaining 3 are redder objects, which cannot be readily classified. We now compare this outcome with the catalogue of known variables for this cluster. |
Processed_Low_Frequency_Marsquakes_and_Where_to_Find_Them:_B.txt | NASA’s InSight mission on Mars continues to record seismic data over 3 years after landing, and to date, over a thousand marsquakes have been identified. With only a single seismic station, the determination of the epicentral location is far more challenging than on Earth. The Marsquake Service (MQS) produces seismicity catalogues from data collected by InSight and provides distance and back azimuth estimates when these can be reliably determined - when both are available these are combined to provide a location. Currently, MQS do not assign a back azimuth to the vast majority of marsquakes. In this work we develop and apply a polarization analysis method to determine the back azimuth of seismic events from the polarization of observed P and S-wave arrivals. The method is first applied to synthetic marsquakes, and then calibrated using a set of well-located earthquakes that have been recorded in Tennant Creek, Australia. We find that the back azimuth is estimated reliably using our polarization method. The same approach is then used for a set of high quality marsquakes recorded up to October 2021. We are able to estimate back azimuths for 24 marsquakes, 16 of these without MQS back azimuths. We locate most events to the east of InSight, in the general region of Cerberus Fossae. |
Processed_Low_Frequency_Marsquakes_and_Where_to_Find_Them:_B.txt | NASA’s Interior Exploration using Seismic Investigations, Geodesy and Heat Transport (InSight) mission landed on Mars in November 2018 and successfully completed the deployment of the Seismic Experiment for Interior Structure (SEIS) in February 2019 (Banerdt et al., 2020). It consists of two sensors, the Very Broadband (VBB) and Short Period (SP) seismometers (Lognonn´e et al., 2019). The lander is located in western Elysium Planitia. Figure 1 shows a global map of Mars, with a zoom inset of a region close to the lander. North of InSight lies the volcanic region Elysium Mons, and in north-eastern to eastern direction is the extensive and young Cerberus Fossae graben system (Berman and Hartmann, 2002). The dichotomy boundary, separating the highlands of the southern hemisphere from the lowlands of the northern hemisphere, is near the landing site to the south. For an overview of the geology around the InSight landing site, see Golombek et al. (2020). |
Processed_Low_Frequency_Marsquakes_and_Where_to_Find_Them:_B.txt | Earth, back azimuths are not routinely determined by polarization, but similar methods are used to confirm sensor orientation, specifically for ocean bottom seismometers, whose orientation is otherwise unknown (e.g. Anderson et al., 1987; Stachnik et al., 2012). Such methods typically use either the P- or the Rayleigh wave motion, since both are expected to occur in the plane pointing towards an earthquake. Deviations of both P-wave (Fontaine et al., 2009) and Rayleigh wave azimuth (Laske, 1995) from great circle direction due to three-dimensional structure are observed, but are typically below 5◦. Having been deployed on the surface, the orientation of the SEIS instrument was determined using a sundial on its cover (Savoie et al., 2021) with an accuracy of 2.5◦. The instrument coordinate system is therefore known well enough to use polarization for marsquake location. |
Processed_Low_Frequency_Marsquakes_and_Where_to_Find_Them:_B.txt | As described in Clinton et al. (2021), since no surface waves have been identified on Mars and the waveforms are affected by strong scattering (Lognonn´e et al., 2020), Rayleigh wave-based methods cannot be applied to Mars. For 8 marsquakes, the P-wave direction of motion could be used to determine the back azimuth, but the majority of marsquakes only have distances (obtained from S-P times), and subsequently, lack a location. Any analysis of seismicity and tectonics on Mars would greatly benefit if more marsquake locations were available. Similarly, moment tensors require a back azimuth to provide results (Brinkman et al., 2021). In St¨ahler et al. (2021), SM 4, a method based on the correlation and instantaneous phase coherence between vertical and horizontal components as a function of azimuth was used to locate a number of additional marsquakes. |
Processed_Low_Frequency_Marsquakes_and_Where_to_Find_Them:_B.txt | In this paper we present an alternative way to estimate the back azimuth for marsquakes, which is based on the polarization of the signal obtained from eigenanalysis of the spectral matrix (Schimmel et al., 2011; Samson and Olson, 1980), allowing the identification of polarized signals even in presence of strong, complex noise. A similar method was previously used for orientation of ocean-bottom seismometers focusing on Rayleigh waves (Scholz et al., 2016). We show that it can be used jointly on P- and S-waves, yielding a robust and consistent back azimuth estimate. |
Processed_Low_Frequency_Marsquakes_and_Where_to_Find_Them:_B.txt | We briefly summarize the existing approach used by the Marsquake Service (MQS, Clinton et al., 2018) to determine the back azimuth, and then describe the specifics of our proposed method. The method is then tested on synthetic martian data as well as a number of earthquakes. We finally apply this method to a set of high quality events recorded by InSight until end of September 2021 (Marsquake Catalog Version 9, InSight Marsquake Service, 2022) to determine their back azimuth and create a new and significantly extended set of marsquake locations. |
Processed_Low_Frequency_Marsquakes_and_Where_to_Find_Them:_B.txt | Figure 1: Map of Mars using Mars Orbiter Laser Altimeter (MOLA, Smith et al., 2001) elevation data. The red triangle marks InSight’s location in Elysium Planitia. The colored ellipses show the estimated locations of a set of high-quality marsquakes. Events close to the InSight lander are marked and labeled with colors, those further out are marked and labeled white. Distances with uncertainties are taken from the MQS catalog V9 (InSight Marsquake Service, 2022). A zoom inset shows the area close to InSight where many quakes are located. |
Processed_Low_Frequency_Marsquakes_and_Where_to_Find_Them:_B.txt | The seismic data from Mars (InSight Mars SEIS Data Service, 2019) is routinely analysed by MQS which catalogs any seismic activity (InSight Marsquake Service, 2022). MQS categorises the seismic events recorded by InSight in different event families depending on their frequency content - low frequency families, and high frequency families. These two event families are then further divided into sub-categories. The low frequency family consists of low frequency (LF) and broadband (BB) events. LF events have energy below 1 Hz, whereas BB events have energy up to and around 2.4 Hz. This specific threshold is due to a subsurface structure effect resulting in a strong Airy phase at 2.4 Hz (Hobiger et al., 2021), which is visible in scattered energy excited by marsquakes (Clinton et al., 2021; Dahmen et al., 2021). The low frequency family are considered ‘classical’ marsquakes, comparable to earthquakes. They occur at teleseismic distances and in depths of 20–50 km (Brinkman et al., 2021; St¨ahler et al., 2021) and have provided insight into the Martian interior. The high frequency family consist of high frequency (HF), very high frequency (VF), and 2.4 Hz events. All three have energy predominately around 2.4 Hz and above. They are the more numerous group, with 882 out of 951 recorded events (InSight Marsquake Service, 2022). In contrast to the low frequency family, their origin is less well constrained. So far, no back azimuth for a high frequency family event has been determined. Their predominantly high frequency content means that the signal is highly scattered and finding a stable back azimuth has proven challenging (van Driel et al., 2021). In this study, we focus on the LF and BB events recorded by InSight. For Catalog Version 9, only six out of the 69 recorded low frequency family events have been located with the standard MQS method: five in the general Cerberus Fossae region (events S0173a, S0235b, S0809a, S0820a, and S0864a), and one in the distant Valles Marineris region (event S0976a) (InSight Marsquake Service, 2022). Two additional events have back azimuths but lack distances due to an uncertain (event S0183a) or absent (event S0899d) second phase. S0183a has a back azimuth consistent with Cerberus Fossae, but the second phase suggests the event is located further away in the Orcus Patera region. S0899d lies in northward direction, where the volcanic region of Elysium Mons can be found (InSight Marsquake Service, 2022). Khan et al. (2021) argue that the S-wave is not observed due to an S-wave shadow zone produced by a lithospheric negative gradient in vS. Following that argument, the distance of these events would be in said shadow, between 45 and 55°. |
Processed_Low_Frequency_Marsquakes_and_Where_to_Find_Them:_B.txt | Throughout the paper, for consistent presentation, we use the high quality low frequency family marsquake S0235b as an example. It is cataloged as a quality A broadband event with Mw of 3.5 and a distance of 28◦. Its large signal to noise ratio (SNR) makes it one of the best events recorded by InSight so far. Figure 2 presents the waveforms for S0235b. Rows (a)-(c) show the velocity timeseries for the vertical, north, and east components respectively, filtered between 10 s and 9.5 Hz. The P and S arrivals identified by MQS are marked with a dashed line. Marked with gray bands are so-called glitches in the data (Scholz et al., 2020), which we will explain in more detail later. Figure 2(d) shows the acceleration spectrogram of the vertical component between 20 s and 10 Hz, calculated with a continuous wavelet transform. Both the P and S arrival are clearly visible in the spectrogram. The P-wave has more broadband energy and excites the 2.4 Hz resonance. On the other hand, the S-wave has lower frequency energy and is predominately visible below 1 Hz. |
Processed_Low_Frequency_Marsquakes_and_Where_to_Find_Them:_B.txt | Figure 2: Waveform timeseries of marsquake S0235b and the vertical acceleration spectrogram. Depicted are (a) the vertical (Z), (b) north (N), and (c) east (E) components of the velocity waveforms, 0.1 Hz–9.5 Hz bandpass. The MQS P and S arrival are indicated by the dashed blue lines. Glitches in the data are marked with gray boxes. (d) shows the vertical acceleration spectrogram between 0.05 Hz–10 Hz. |
Processed_Low_Frequency_Marsquakes_and_Where_to_Find_Them:_B.txt | Figure 3: MQS polarization analysis of marsquake S0235b. (a) Velocity waveforms for vertical, north, and east components filtered between 2–8 s. P (red) and S (blue) picks are marked with vertical lines. Picking uncertainties are given by a horizontal error bar. A glitch in the data is marked with the gray box. The light red box after the P pick marks the time window from which the hodograms in (b) are calculated. The normalized probability density function (PDF) is shown in the bottom right plot. The MQS-assigned back azimuth for this event is 74◦, indicated by the dashed black line. |
Processed_Low_Frequency_Marsquakes_and_Where_to_Find_Them:_B.txt | Figure 4: Analysis of maximum amplitude of the rotated envelopes for marsquake S0235b. The normalised maximum amplitude of the radial component is given for a (a) P and (b) S time window. The correlation with the vertical component is given in (c) for P and (d) for S. The dashed lines in each plot show a pre-event noise window. Window lengths are each 15 s for the noise, P, and S window. The signal is bandpass-filtered for three frequency windows: 0.18–0.25 Hz (blue), 0.25–0.36 Hz (pink), 0.36–0.5 Hz (green). |
Processed_Low_Frequency_Marsquakes_and_Where_to_Find_Them:_B.txt | the event. Conversely, the S-wave amplitude should be close to or at a minimum at the same time if assuming primarily SH-waves. There is a 180◦ ambiguity with this method, which can be resolved by calculating the correlation of the radial with the vertical component. The correlation is positive at the correct angle for the P-wave. There is also a sensitivity of the method to the frequency range of the bandpass filter applied to the seismic signal before rotating and calculating the envelope. The polarization of the signal can be (and routinely is) frequency dependent. We show three different frequency bands (0.18–0.25 Hz, 0.25–0.36 Hz, 0.36–0.5 Hz) for marsquake S0235b in Figure 4. The MQS back azimuth is marked with a gray dashed line. In addition to a short 15 s window around the P and S-wave arrivals, we show a 15 s pre-event noise window to compare the signal to environmental noise. The P window maximum amplitude in Figure 4 (a) shows two maxima, around 60–70◦ and 240–250◦. The first coincides with a positive correlation with the vertical component (Fig. 4 (c)). This is in agreement with the results provided by MQS. The S window in Figure 4 (b) is more complex. In contrast to the P window, the frequency bands have less agreement and the minima diverge from the MQS results by about 60◦. Different frequency bands only converge to a common back azimuth for high quality events. For all other events, the back azimuth from the maximum amplitude varies strongly between frequencies. This makes it impossible to say which, or if any, of the estimated back azimuths is correct. Both the MQS method and the rotated-amplitude method briefly presented here struggle with the vast majority of recorded marsquakes. Also, it would be preferable, if the tradeoffs in selection of time and frequency windows were immediately visible in the results. |
Processed_Low_Frequency_Marsquakes_and_Where_to_Find_Them:_B.txt | In addition we compare our results to those from St¨ahler et al. (2021) SM 4, currently under revision in Drilleau et al. (2022), who estimate back azimuths for a set of events based on Marsquake Catalog Version 8 (V8). On top of the method described above, the instantaneous phase coherence between vertical and horizontal components as a function of azimuth. For P-waves, they expect the energy to be maximum and both correlation and phase coherence to be minimum at the correct angle. Since the interference of horizontal and vertical shear wave components (SH and SV, respectively) affects the energy, it cannot be used directly in the case of S-waves. However, the SV-wave component should produce a weak maximum of both correlation coefficient and phase coherence along the correct back azimuth. Drilleau et al. (2022) analyse the signals in different time windows around their phase picks as well as a noise window. They use fixed frequency ranges for the analysis, 0.4–1.0 Hz for P-waves and 0.3–1.0 Hz for S-waves, owing to S-waves’ lower frequency content. |
Processed_Low_Frequency_Marsquakes_and_Where_to_Find_Them:_B.txt | We propose to apply a complete polarization analysis of P and S body waves to determine the back azimuth of marsquakes. The method is based on the work of Samson and Olson (1980) and Samson (1983), and was further developed for seismological applications by Schimmel and Gallart (2003) and Schimmel et al. (2011) and has been applied to martian ambient noise before (Stutzmann et al., 2021). The 3-component seismogram is first transformed into time-frequency domain to produce a time-frequency dependent spectral matrix. For each time-frequency pixel, the matrix is decomposed into eigenvectors to obtain information on the polarization of the seismic signal. In contrast to Stutzmann et al. (2021), we use the continuous wavelet transform as described by Kristekov´a et al. (2009) for the t − f transformation. |
Processed_Low_Frequency_Marsquakes_and_Where_to_Find_Them:_B.txt | Polarization describes the three-dimensional motion at the receiver location. At any particular time, this motion can be approximated by an ellipse with semi-major axis and semi-minor axis. From these axes in 3-D space, we can get the azimuth and inclination for each. We also obtain the ellipticity of the signal from the ratio of the semi-minor to the semi-major axis. For an incoming P-wave, we expect the particle motion to be rectilinear and the azimuth of the semi-major axis of the polarization ellipse points in the direction of the event location (Sollberger et al., 2020). The S-wave is more complex. To the first order and assuming an identical incidence vector between P and S, we expect the azimuth of the S-wave semi-major axis to be either ±90◦ (for SH) or −180◦ (for SV) with respect to the P-wave. The ratio of SH and SV-wave energy depends on the source mechanism. We propose an analysis that does not rely on correct discrimination between SH and SV-waves. Any incoming S-wave should lie in a plane perpendicular to the P-wave vector in 3-D space, see Figure 5. Assessing P and S-wave polarization together reduces the influence of noise in any back azimuth estimation. Further, we can use the inclination of the signal to differentiate between the expected seismic signals and noise (e.g. wind or glitches in the data). For a P-wave, the inclination should be steep while a S-wave is expected to arrive with lower inclination. |
Processed_Low_Frequency_Marsquakes_and_Where_to_Find_Them:_B.txt | Figure 5: Schematic showing wave polarization in 3-D space. The P-wave is represented by a vector with an inclination and azimuth relative to the horizontal plane (red ellipse) and north component (N), respectively. The S-wave consists of both SH- and SV-waves, which lie in a plane perpendicular to the P-wave vector (cyan ellipse). |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.