The JCMT Transient Survey: Data Reduction and Calibration Methods

TL;DR

This paper details the data reduction, calibration, and analysis techniques used in the JCMT Transient Survey to study brightness variations in protostars, achieving high-precision imaging and identifying sources with potential variability.

Contribution

It introduces novel data processing and calibration methods for submillimetre observations, enabling precise alignment and flux measurement in a large-scale star formation variability survey.

Findings

Achieved better than 1" spatial alignment between observations

Attained 2-3% uncertainty in relative peak brightness

Identified sources with potential brightness variations

Abstract

Though there has been a significant amount of work investigating the early stages of low-mass star formation in recent years, the evolution of the mass assembly rate onto the central protostar remains largely unconstrained. Examining in depth the variation in this rate is critical to understanding the physics of star formation. Instabilities in the outer and inner circumstellar disk can lead to episodic outbursts. Observing these brightness variations at infrared or submillimetre wavelengths sets constraints on the current accretion models. The JCMT Transient Survey is a three-year project dedicated to studying the continuum variability of deeply embedded protostars in eight nearby star-forming regions at a one month cadence. We use the SCUBA-2 instrument to simultaneously observe these regions at wavelengths of 450 $\mu$m and 850 $\mu$m. In this paper, we present the data reduction techniques, image alignment procedures, and relative flux calibration methods for 850 $\mu$m data. We compare the properties and locations of bright, compact emission sources fitted with Gaussians over time. Doing so, we achieve a spatial alignment of better than 1" between the repeated observations and an uncertainty of 2-3\% in the relative peak brightness of significant, localised emission. This combination of imaging performance is unprecedented in ground-based, single dish submillimetre observations. Finally, we identify a few sources that show possible and confirmed brightness variations. These sources will be closely monitored and presented in further detail in additional studies throughout the duration of the survey.

Tables

Table 1. Table 1: A summary of the observed JCMT Transient Survey fields between the first observations on December 22 nd nd {}^{\text{nd}} , 2015 and March 1 st st {}^{\text{st}} , 2017.

Region	Central R.A.
	Central Decl.
	Number of
	Average 850 $\mu$m Noiseb,c
	Std. Dev. 850 $\mu$m Noised
	Noise in the Co-add
	(mJy  beam-1)
Perseus: NGC1333	03:28:54	+31:16:52	10	12.26	0.40	3.92
Perseus: IC348	03:44:18	+32:04:59	9	12.18	0.43	4.30
Orion: OMC 2/3	05:35:33	-05:00:32	9	11.72	0.54	4.19
Orion: NGC2024	05:41:41	-01:53:51	11	11.29	0.40	4.32
Orion: NGC2068	05:46:13	-00:06:05	10	11.75	0.38	3.85
Ophiuchus: Core	16:27:05	-24:32:37	8	13.35	0.75	5.00
Serpens: Main	18:29:49	+01:15:20	9	12.01	0.27	4.54
Serpens: South	18:30:02	-02:02:48	9	14.56	1.18	4.72

Table 2. Table 2: A summary of all JCMT Transient Survey observations obtained between December 22 nd nd {}^{\text{nd}} , 2015 (the beginning of the survey) and March 1 st st {}^{\text{st}} , 2017.

Region	Date	Scan	${\tau }^a$	850 $\mu$m Noiseb,c
	Number of Sources Above			(mJy  beam-1)
	Number of
	Number of
	Family Members
IC348	20151222	19	0.06	12.54	9	6	3
IC348	20160115	22	0.07	9.99	12	6	3
IC348	20160205	18	0.04	12.79	12	6	3
IC348	20160226	20	0.05	12.39	13	5	3
IC348	20160318	27	0.05	11.1	12	5	3
IC348	20160417	09	0.04	11.0	13	5	3
IC348	20160826	40	0.08	14.33	12	6	3
IC348	20161126	22	0.05	12.29	14	6	3
IC348	20170209	28	0.09	13.17	12	5	3
NGC1333	20151222	18	0.06	12.22	39	36	7
NGC1333	20160115	10	0.08	11.76	40	29	7
NGC1333	20160205	17	0.04	12.99	39	29	7
NGC1333	20160229	17	0.04	11.46	42	26	7
NGC1333	20160325	11	0.06	11.58	33	25	7
NGC1333	20160802	31	0.09	12.31	43	28	7
NGC1333	20160830	48	0.09	14.05	39	33	7
NGC1333	20161119	88	0.07	9.65	38	28	7
NGC1333	20161126	21	0.05	12.78	36	27	7
NGC1333	20170206	29	0.12	13.85	42	26	7
NGC2024	20151226	49	0.12	12.58	12	10	3
NGC2024	20160116	22	0.06	9.8	15	7	3
NGC2024	20160206	13	0.04	11.88	21	7	3
NGC2024	20160229	22	0.04	13.01	17	7	3
NGC2024	20160325	21	0.06	11.35	19	8	3
NGC2024	20160329	10	0.05	9.1	19	7	3
NGC2024	20160427	12	0.05	11.91	9	6	3
NGC2024	20160826	29	0.09	12.84	23	8	3
NGC2024	20161119	99	0.07	9.13	16	9	3
NGC2024	20161126	53	0.06	10.29	18	8	3
NGC2024	20170206	25	0.11	11.68	15	8	3
NGC2068	20151226	52	0.12	13.4	32	23	8
NGC2068	20160116	27	0.06	9.65	33	21	8
NGC2068	20160206	15	0.05	12.08	31	21	8
NGC2068	20160229	13	0.04	12.21	29	18	8
NGC2068	20160329	11	0.06	10.82	33	22	8
NGC2068	20160427	13	0.05	12.68	28	16	8
NGC2068	20160827	53	0.08	11.8	31	21	8
NGC2068	20161120	88	0.09	11.98	30	20	8
NGC2068	20161126	56	0.06	10.16	30	21	8
NGC2068	20170206	17	0.11	12.76	32	20	8
OMC 2/3	20151226	36	0.11	12.48	55	51	10
OMC 2/3	20160116	19	0.06	9.94	30	22	10
OMC 2/3	20160206	12	0.04	12.03	26	21	10
OMC 2/3	20160229	11	0.04	11.8	45	30	10
OMC 2/3	20160325	15	0.06	10.74	29	20	10
OMC 2/3	20160422	11	0.05	11.4	28	23	10
OMC 2/3	20160826	20	0.11	15.21	25	17	10
OMC 2/3	20161126	52	0.06	9.74	56	40	10
OMC 2/3	20170206	21	0.12	12.13	43	32	10
Oph Core	20160115	84	0.07	11.9	26	23	4
Oph Core	20160205	63	0.04	12.47	22	13	4
Oph Core	20160226	51	0.05	11.08	27	17	4
Oph Core	20160319	65	0.04	12.56	27	16	4
Oph Core	20160417	43	0.04	12.03	24	16	4
Oph Core	20160521	34	0.08	15.03	27	15	4
Oph Core	20160826	11	0.11	17.56	24	14	4
Oph Core	20170206	83	0.11	14.2	22	16	4
Serpens Main	20160202	54	0.09	12.11	23	21	5
Serpens Main	20160223	50	0.05	11.68	22	18	5
Serpens Main	20160317	51	0.04	12.2	21	14	5
Serpens Main	20160415	46	0.04	11.82	22	16	5
Serpens Main	20160521	39	0.08	14.01	22	15	5
Serpens Main	20160722	23	0.1	11.49	23	14	5
Serpens Main	20160827	12	0.09	11.32	24	15	5
Serpens Main	20160929	12	0.09	11.95	18	13	5
Serpens Main	20170222	70	0.1	11.47	26	12	5
Serpens South	20160202	58	0.09	11.27	39	35	9
Serpens South	20160223	65	0.05	18.66	39	32	9
Serpens South	20160317	52	0.04	11.41	34	25	9
Serpens South	20160415	48	0.04	11.57	38	29	9
Serpens South	20160521	44	0.07	12.61	38	27	9
Serpens South	20160721	11	0.08	19.42	41	31	9
Serpens South	20160827	17	0.09	17.05	43	32	9
Serpens South	20160929	18	0.08	11.34	39	30	9
Serpens South	20170222	81	0.1	17.68	36	28	9

Equations

$$cor(\text{R.A.},\text{Decl.})=\sum _{{\text{pixels}}_x}\sum _{{\text{pixels}}_y}\mathcal{R}(x,y)\times \mathcal{S}(x-\text{R.A.},y-\text{Decl.})$$

$$\text{Radial offset}=\sqrt{{\left(\Delta \text{R.A.}\right)}^2+{\left(\Delta \text{Decl.}\right)}^2}$$

Full text

techniques: image processing – methods: data analysis – stars: formation – submillimetre: ISM – submillimetre: general

††journal: ApJ††software: Starlink (Currie et al., 2014), Astropy (Astropy Collaboration et al., 2013), Python version 3.5, APLpy (Robitaille & Bressert, 2012), Matplotlib (Hunter, 2007)

1 Introduction

Although there have been many advances made in understanding low mass star formation over the past ten years (see, for example, di Francesco et al. 2007, Ward-Thompson et al. 2007a, André et al. 2014), the manner in which mass assembles onto a forming star remains a crucial open question. As Kenyon et al. (1990) first demonstrated, assuming the mass accretion process onto a young star occurs at a constant rate (steady inside out collapse; Shu et al. 1987) gives rise to “The Luminosity Problem”: the empirical result that the median protostellar luminosity is measured to be approximately an order of magnitude less than the expected value. In recent years, this problem has been confirmed and emphasised by Spitzer Space Telescope observations through which even lower luminosities have been discovered (see Dunham et al. 2008, Evans et al. 2009, Enoch et al. 2009, Dunham et al. 2013, Dunham et al. 2014). One solution to this problem is that the accretion does not proceed at a constant rate. Rather, it occurs during episodic events which may be accompanied by outbursts that can be detected at infrared, submillimetre, and sometimes optical wavelengths (see McKee & Offner 2011, Johnstone et al. 2013, and Scholz et al. 2013). There is indirect evidence that episodic accretion occurring while a protostar is still deeply embedded in its nascent gas and dust is an early phase of the more evolved FU Orionis (FUors; Herbig 1977, see also Hartmann & Kenyon 1985) sources (Dunham et al., 2014; Audard et al., 2014).

The physical mechanism responsible for a continuum outburst detected at the submillimetre wavelengths of interest to this survey is re-radiation from heated dust grains in the surrounding protostellar envelope. Outside of the JCMT Transient Survey, there have already been two millimetre sources (both embedded protostars) that have shown direct evidence of an active burst accretion phase accompanied by a dramatic brightening, HOPS 383 in Orion (Safron et al. 2015; using Atacama Pathfinder Experiment and SCUBA archive data), and MM1 in NGC6334I (Hunter et al. 2017; using Atacama Large Millimeter/submillimeter Array and Submillimeter Array data).

The JCMT Transient Survey (Herczeg et al. in preparation) is a three year project dedicated to observing continuum variability in deeply embedded protostars at submillimetre wavelengths with the Submillimetre Common User Bolometer Array 2 (SCUBA-2; Holland et al. 2013). To this end, we are monitoring eight regions selected from the JCMT Gould Belt Survey (GBS; Ward-Thompson et al. 2007b) that have a high density of known protostellar and disk sources (Young Stellar Object Classes 0 to II and flat spectrum; see Lada 1987, Andre et al. 1993, and Greene et al. 1994) at an approximate 28 day cadence whenever they are observable in the sky. SCUBA-2 uses approximately 10,000 bolometers subdivided into two arrays to observe at both 450 $\mu$m and 850 $\mu$m simultaneously. While we expect sources undergoing an accretion burst event to show a stronger signal at 450 $\mu$m (Johnstone et al., 2013), in this paper we focus only on the 850 $\mu$m data. The noise levels in 450 $\mu$m maps are much more dependent on the weather than their 850 $\mu$m counterparts, causing the signal-to-noise ratio (SNR) to fall dramatically when there is more water vapour in the atmosphere. In addition, the beam profile is less stable than at 850 $\mu$m (as shorter wavelengths are more susceptible to dish deformation, and focus errors. For more information, see Dempsey et al. 2013) requiring careful attention in order to make precise measurements of compact objects. We thus start here by defining the 850 $\mu$m calibration and we will use this knowledge to calibrate the 450 $\mu$m data at a later date. As the survey matures and precise 450 $\mu$m data calibration is achieved, these simultaneous observations will provide further confirmation of significant variations.

In order to track the peak brightnesses of submillimetre emission sources over each epoch, we test and employ a robust data reduction method and use multiple observations of the same regions to derive post-reduction image alignment and relative flux calibration techniques. Reducing SCUBA-2 data is a complex process with several user-defined parameters that affect the final image produced (for detailed information on SCUBA-2 data reduction procedures, see Chapin et al. 2013). A large amount of work has been invested in understanding the optimal data reduction parameters to use for differing science goals (see, for example, Mairs et al. 2015) depending on the scan pattern of the telescope and the amount of large-scale structure that needs to be robustly recovered. In all cases, the nominal JCMT pointing error is 2-6$\arcsec$ (East Asian Observatory staff, private communication) and the flux calibration is uncertain to $\sim$5-10% (Dempsey et al. 2013; see also, Section 4.2). While this is sufficient for most projects which use JCMT data, both of these uncertainties can be improved upon when there are multiple observations of regions with bright sources taken in a consistent manner. In this work, we seek to improve both the spatial alignment and the flux calibration of the JCMT Transient Survey data by approaching the problem from a relative point of view.

Properly matching faint, potential protostellar sources over the observed epochs and co-adding those observations with high precision for the highest SNR requires sub-pixel accuracy (<<3$\arcsec$ at 850 $\mu$m) in the spatial alignment. Similarly, if we were to use the nominal flux calibration, where the uncertainty is taken to be $\sigma\sim 10\%$, the flux would need to vary by 30-50% for a transient event to be deemed significant (3-5$\sigma$). Thus, our goal is to reduce this uncertainty by a factor of 3 to 5 (i.e. sigma $\sim 2-3\%$) by considering relative brightness changes over time and ignoring the absolute flux calibration. We will then be able to measure flux variations of $\sim 10\%$ to be statistically significant (>$3\sigma$). Several models predict smaller flux variations due to episodic accretion over few year timescales to occur much more frequently than large flux variations (see, for examples, Bae et al. 2014, Vorobyov & Basu 2015, and Herczeg et al. in preparation). Observations like those performed throughout the JCMT Transient Survey will help constrain the current models. The techniques we have developed here can be applied to any JCMT data obtained in a similar manner, including archival data obtained by the GBS (follow up analysis by Mairs et al., in prep.). Thus, we are able to successfully align and relatively flux calibrate archival data such as those which were obtained by the GBS and we include these data in a follow-up analysis (Mairs et al. in preparation).

This paper is organised as follows: In Section 2 we summarise the details of our SCUBA-2 observations. In Section 3 we outline our data reduction methods and showcase four tests we performed which altered the amount of large-scale structure recovered in a given map and the initial priors offered to the map-making pipeline in order to select the most robust techniques for our purpose of detecting transient events in deeply embedded protostars. In Section 4 we detail our source extraction, post-reduction spatial alignment, and relative flux calibration methods applied to all current JCMT Transient data. In Section 5, we present an analysis on the recovered compact emission sources and highlight objects of interest including the first demonstrably variable source in our survey (Yoo et al. in preparation). Finally, we present our conclusions in Section 6.

2 Observations