You need an eReader or compatible software to experiencethe benefits of the ePub3 file format.

We report the serendipitous discovery of a giant mid-infrared outburst from a previously unknown source near a star-forming region in the constellation Monoceros. The source gradually brightened by a factor of 5 from 2014–2016 before an abrupt rise by a factor of more than 100 in 2017. A total amplitude increase of >500 at 4.5μm has since faded by a factor of about 10. Prior to the outburst, it was only detected at wavelengths longer than 1.8μm in UKIDSS, Spitzer, and Herschel with a spectral energy distribution of a Class I young stellar object (YSO). It has not been detected in recent optical surveys, suggesting that it is deeply embedded. With a minimum distance of 3.5 kpc, the source has a bolometric luminosity of at least 9L_☉ in the quiescent state and 400L_☉ at the peak of the eruption. The maximum accretion rate is estimated to be at least a few 10⁻⁵M_☉ yr⁻¹. It shares several common properties with another eruptive event, WISE J142238.82-611553.7: exceptionally large amplitude, featureless near-infrared spectrum with the exception of H₂ lines, intermediate eruption duration, an embedded Class I YSO, and a low radiative temperature (<600–700 K) in outburst. We interpret that the radiation from the inner accretion disk and young star is obscured and reprocessed by either an inflated outer disk or thick dusty outflow on scales >6.5 au during the outburst.

Original content from this work may be used under the terms of theCreative Commons Attribution 4.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.

Star formation takes place in dense cores of molecular clouds. The initial collapse of the cloud core leads to a small protostar, which grows into a typical main-sequence star by accreting material from the surrounding envelope via a protoplanetary disk. On the basis of the envelope, disk, and stellar core status, young stellar objects (YSOs) can be divided into Class 0, I, II, and III. Class 0 represents the earliest stage of their evolution, where a protostar is embedded in a quasi-spherical thick envelope, followed by Class I objects, which possess a disk and envelope, and the gas envelope is largely dispersed in Class II objects, leaving a planetary-forming disk surrounding the central star; Class III represents the last evolution stage of YSOs before it becomes a main-sequence star with only a debris dust disk (Adams et al.1987). Since the typical accretion rate for T Tauri stars (TTS) is on the order of 10^−8–9M_☉ yr⁻¹, far smaller than the requirement for the growth of a solar mass star, it has been proposed that most of the mass is acquired in episodic accretions (Dunham et al.2013). Numerical simulations show that accretion rates may increase by up to 3 orders of magnitude during short episodic accretion phases (Vorobyov & Basu2006,2010; Vorobyov et al.2013; Kadam et al.2021).

Episodic accretion manifests itself as optical and IR eruptions, which are distinguished from more frequent TTS variations by their much larger amplitude and long duration (Fischer et al.2023, see review). The most extreme eruption types are FU Orionis objects (FUors), with amplitudes up to 8 mag and a duration of 100 yr (Herbig1977,1989; Hartmann & Kenyon1996). The amplitudes of the outbursts in EXors are less than in FUors on average with a duration on a timescale of several months to a year (Hartmann & Kenyon1996). But some EXors brighten up to 6 mag (Fedele et al.2007) in a few months. These two classes of objects exhibit distinct spectroscopic properties. FUors in outbursts are characterized by an F/G supergiant spectrum, but with broad double-peak absorption lines in the optical spectrum and K/M supergiants with CO overtones in the near-infrared (NIR) spectrum. In contrast, classical EXors exhibit K-or-M dwarf absorption spectra and TTS emission spectra in quiescence and an additional thermal component with a temperature of 1000–4500 K during outburst. Hydrogen, metallic, or CO lines are seen in both emission and absorption in the IR during the outburst (Fischer et al.2023). Although EXors are Class II YSOs, it is not yet clear whether FUors resemble Class I or Class II YSOs (Miller et al.2011; Cieza et al.2018). The different spectroscopic properties of FUors and EXors can be understood by their different accretion modes. Magnetospheric column accretion works in EXors, while the accretion disk extends to the stellar surface in FUors (Liu et al.2022; Rodriguez & Hillenbrand2022). With the advance of large-sky surveys, some eruptive YSOs have been found to fill the gap between FUors and EXors in the amplitude and duration of the outburst (Fischer et al.2023).

However, much of the current knowledge on these eruptions is largely biased against Class I/0 embedded objects. To date, approximately 30 FUors or FUor-like objects are known, most of which have been discovered at optical wavelengths. Systematic surveys at longer wavelengths provided a different view of eruptive YSOs. Contreras Peña et al. (2017) found 106 eruptive YSOs in the VISTA Variables in the Via Lactea (VVV) survey of the Galactic plane covering 119 deg². The sample showed that eruptions were 10 times more common in Class I objects than in Class II objects within the sample, and the typical duration was between those of FUors and EXors. A James Clerk Maxwell Telescope transient survey of eight nearby star-forming regions (SFRs) in 4 yr resulted in six sources with a year-long eruption with 40% higher accretion rates than in the quiescent state, including four Class I and two Class 0 sources. (Lee et al.2021). The latter authors concluded that episodic accretion plays a minimal role in the mass accumulation of stars.

Although some optically discovered FUors and EXors are less variable in the mid-infrared (MIR) than in optical (Fedele et al.2007; Hillenbrand et al.2018; Szegedi-Elek et al.2020), Lucas et al. (2020) found an 8 mag outburst with intermediate duration in the MIR of a previously unknown YSO. PTF 14ig displayed an outburst of more than 6 mag in the MIR accompanied by an optical outburst with a similar amplitude (Hillenbrand et al.2019). A similar amplitude eruption was also detected in WISEA J075915.26-310844.6, but its nature is less clear (Thévenot2020; see also remarks by Hillenbrand ⁶ ).

In this paper, we report the serendipitous discovery of a gigantic eruption in a previously unknown YSO (R.A. = 06:47:22.95, decl. = +03:16:44.56) in the MIR. The source was identified in the blind search for large amplitude MIR variables in the Wide Infrared Sky Explorer (WISE) archive. This paper is arranged as follows. We describe the observations and archival data, as well as the data analysis methods, in Section2. We present the main results in Section3. A discussion of the nature of the source and the outburst is given in Section4. In Section5, we summarize our conclusions.

Initially, we examined MIR light curves (Figure1) from the W1 and W2 bands of the AllWISE and NEOWISE single-exposure photometric database in the IRSA IR archive within 6″ of the ALLWISE position. Light curves covered the period from early 2010 to the end of 2022 (Wright et al.2010; Mainzer et al.2011,2014; Wright et al.2019; NEOWISE Team2020). We examined the single-exposure photometric data in each epoch and found that they were consistent, with the exception of the highest flux state in the W2 band when the source was brighter than 7.0 mag in W2, leading to saturation. Therefore, we averaged the single-exposure measurements at each epoch. We noticed that the location of the erupting source was close to an optically bright star (hereafter referred to as S1, 06:47:22.98+03:16:39.9) in Pan-STARRS before mid-2014, but shifted northward from S1 by 45 since the giant eruption. This indicates that the outburst comes from another source (06:47:22.95+03:16:44.56) that was not detected by Pan-STARRS, rather than due to a large proper motion.

Zoom InZoom OutReset image size

Fortunately, the region was covered by the Spitzer GLIMPSE 360 survey (Benjamin et al.2003; Churchwell et al.2009; GLIMPSE Team2020). We retrieved calibrated images in theI1 andI2 bands, and indeed there is an IR source (J064722.95+031644.6, hereafter J0647) 45 north of S1 in these images (Figure2). These images were taken from 2013 November 6 to 2013 November 12, before the outburst. J0647 has a brightness comparable to S1 in theI2 band, but is much fainter in theI1 band. The color of J0647 is very red. We also examined the optical images at the Zwicky Transient Facility (ZTF), which were taken between March 2018 and November 2021, and no source was detected at the position of J0647.

Zoom InZoom OutReset image size

With a separation of 45, the S1 source and J0647 can be easily resolved in images. We retrieved a NIR image and photometry from UKDISS (Lawrence et al.2007) and found a weak source in theK band at the position of J0647 (K = 17.35 ± 0.11 mag). The source appears extended in theK band and is not detected in theJ andH bands with upper limits of 20.0 mag and 19.2 mag, respectively. There is no significant variation in the source between the two observations made in 2007 and 2012. The region was observed by Herschel PACS (PHPDP Team2020) with blue and red filters on 2012 April 29. A weak source is visible at the location of J0647 in both images and is blended with another relatively bright source (Y1). We estimate the image flux by fitting two PSFs to a 9 × 9-pixel image centered on the two sources. The point-spread function (PSFs) in the two bands were constructed by Elbaz et al. (2011), and the centers of the PSFs are fixed to the positions of the IRAC sources. J0647 is detected marginally in the blue band at a signal-to-noise ratio (S/N) of 2.71.

J0647 is blended with S1 in the W1 and W2 images, and pre-eruption photometry from the WISE catalog is dominated by S1. To obtain the complete MIR light curves of J0647, we downloaded the time-resolved unWISE coadds (Meisner et al.2018; UnWISE Team2021) and constructed differential images by subtracting the first-epoch WISE image from the rest of the images. The source has been visible in the different images since epoch 4, and the centroid coincides with J064722.95+031644.6 (Figure3), which supports that S1 did not vary significantly. We measured the differential magnitude of the source in the differential image using forced photometry. The light curves in the W1 and W2 bands were obtained by adding a constant, which was obtained from the Spitzer flux, to the difference fluxes. Note that the seventh release of unWISE does not include images of the last four epochs, and instead, we used the average magnitudes of NEOWISE single-exposure photometry. S1 contamination does not have a significant effect on our results because J0647 was much brighter than S1 during the last four epochs.

Zoom InZoom OutReset image size

An echelle spectrum was obtained using the Immersion GRating INfrared Spectrometer (IGRINS) mounted on the Discovery Channel Telescope at Lowell Observatory (now Lowell's Discovery Telescope) with an exposure of 4 × 1200 s on 2018 November 27, resulting in a spectrum with a resolutionλ/Δλ = 45,000 covering the IRH andK bands (Mace et al.2018). The magnitude measured from the acquisition image in theK band is approximately 12.5 mag, about 4.9 mag brighter than that in the UK hemisphere survey carried out in 2012 March 14 or 2007 October 29 (Dye et al.2018). The IGRINS spectrum was reduced following a standard procedure described in Park et al. (2014) and Lee & Gullikson (2016). The S/N of the spectrum is very low in theH band but increases with wavelength, reaching ∼50 in theK band due to the steepness of the spectral energy distribution (SED).

Three low-resolution IR spectra were obtained with the Triple Spectrograph (Herter et al.2008; hereafter TSpec) on board the Palomar 200-inch telescope (P200) on 2020 September 30, 2020 October 6, and 2022 February 18, respectively. On-source exposure times were 120, 436, and 400 s in BAAB mode. TSpec simultaneously covers theJ, H, andK bands from 1–2.4μm with a spectral resolution of around 2700. Due to the steepness of SED, the S/N in theJ band is less than 1, increases to about 3 in theH band, and reaches up to 30 in theK band. The flux-calibrated spectra are shown in Figure4. For the sake of clarity, the TSpec from 2020 September 30 is not shown because it is almost identical to the spectrum from 2020 October 6, but with slightly lower S/Ns.

Zoom InZoom OutReset image size

3.1. Environment of the Source

The main star formation region (SFR) near J0647 is IRAS 0644+0319, which was identified as an intermediate-mass SFR with a blob-shell morphology (Lundquist et al.2014). Figure5 shows the 56 ×72 images in the Herschel's PACS red (160μm) and blue (70μm) bands, Spitzer's IRAC 1 (4.5μm) and Pan-STARRS'i band. IRAS 0644+0319 is seen in Figure5 as a large bright clump. J0647 is located close to the outskirts of the SFR. Several point sources are visible on both the main SFR clump and the surroundings. They are likely YSOs, probably fragmented from the same parent molecular cloud. Stars 0, 2, 4, 5, and 6 line up in a narrow strip as shown in the middle panel of Figure4, and the 160μm image shows a weak tidal stream stretching from the main star formation area to stars 7 and 8. We collect broadband photometric data for these nine sources marked in the second panel of Figure5 from the Herschel Archival Catalog, Spitzer Survey, Galactic Plane Survey of UKIDSS, and Pan-STARRS archives (Werner et al.2004; Lucas et al.2008; Pilbratt et al.2010; Chambers et al.2016). Their SEDs are plotted in Figure6. The broadband spectral slope in the IR suggests that they are candidate YSOs of Classes I and II. We also studied the MIR light curves of these objects from the WISE archive, and none showed periodic variations, which would have been the signature of dust-rich thermal pulsating AGB stars (Karovicova et al.2013).

Zoom InZoom OutReset image size

Zoom InZoom OutReset image size

Lundquist et al. (2015) calculated the kinematic distance of the SFR using the¹³CO profile obtained with the Onsala 20 m telescope in the direction of IRAS 06446+0319 (Lundquist et al.2015) with a beam size 35″ FWHM. These authors found two peaks in the profile of the¹³CO emission line separated by 2.05 km s⁻¹, which was interpreted as two components of the molecular gas of column densities 3.66 and 2.99 × 10²¹ cm⁻² at distances 3.4 and 3.7 kpc, respectively, assuming that the molecular gas strictly follows the Galactic rotation curve withV_☉ = 240 km s⁻¹ andR_☉ = 8.34 kpc. The small separation in two peaks of CO may be caused by bipolar outflows. In that case, the distance is likely between the two. Finally, we note that the VLSR velocity determined from H₂ lines in the IR spectrum is about 40 km s⁻¹, which is consistent with that from the CO measurement of the SFR, considering the uncertainty of wavelength calibration and possible warm H₂ outflows. This supports the idea that J0647 is associated with SFR IRAS 06446+0319. We adopt a nominal distance of 3.5 with an uncertainty of 0.2 kpc.

3.2. SED in the Quiescent State and Class I YSO

Photometric data before the outburst are plotted in Figure7. Although these data are collected at different times, the SED represents the typical quiescent emission of J0647.K magnitudes obtained in 2007 and 2012 were not significantly different. Little can be seen in the WISE differential images for epoch 1-epoch 0 and epoch 2-epoch 0, suggesting that the source stayed in a quiescent state before epoch 3. The Spitzer and Herschel observations were all performed before epoch 3. S1 and J0647 are mixed in the WISE W3 and W4 bands. Since S1 does not show an infrared excess in W1 and W2, an IR excess in the W3 or W4 bands is unlikely. S1 is not variable in the W1 or W2 bands, and no systematic residuals are seen in the star position in differential images. Thus, we estimated the magnitude of W3 and W4 of S1 by fitting a stellar model to its SED in the optical and NIR as well as Spitzer I1 and I2. The SED is well fitted with an ATLAS9 model. Extrapolating S1 photospheric emission suggests that it contributes ∼1.3% of the observed W3 flux and ∼0.1% of the W4 flux.

Zoom InZoom OutReset image size

The SED of J0647 in the quiescent state increases steeply from 3 to 20μm and then flattens to the far-infrared (FIR). Quiescent WISE colors reside at the location of Class I YSOs on the color–color diagram, far from the AGB star locus (Figure5 and the second panel of Koenig & Leisawitz2014; also Lucas et al.2017). With a spectral slope of and an upper limit of bolometric temperature (Myers & Ladd1993) of 165 K, it is classified as Class I YSO (Figure 3 of Pokhrel et al.2023). However, the upper limit of the bolometric temperature is near the boundary between Class I and Class 0, so Class 0 cannot be completely rejected. The SED is reasonably well fitted by the YSO models in Figure6 (Robitaille2017). Bolometric flux is estimated to be 2.5 × 10⁻¹¹ erg cm⁻² s⁻¹ by integrating the SED from the near to the FIR with a simple log-linear interpolation. This number is within a factor of 2 of the integration of the best-fitted YSO model. The bolometric correction for the luminosity of W2 isC =L_bol/νL_ν(W2) ≃ 20. With the distance in Section3.1, we derived a quiescent luminosity of 9(d/3.5 kpc)²L_☉. This luminosity and the absence of periodic variations eliminate the possibility of J0647 being an embedded AGB interloper.

3.3. MIR Outburst and Evolution of SED

3.4. Estimate of the Mass Accretion Rate

The main source of energy for the outburst is the gas accretion onto the YSO, although the detailed physical process that caused such a sudden mass inflow is not yet clear. The accretion power can be written as

whereβ < 1 is the radiation efficiency, is the mass accretion rate in units of 10⁻⁵M_☉ yr⁻¹, andM_* andR_* are the mass and radius of the YSO, respectively. We will adopt the meanR_*/M_* = 3.9R_☉/M_☉ with a scatter of 1.9 for TTS (Lucas et al.2020). At a high accretion rate, we might underestimateR_*/M_* due to a possible expansion of the upper layer of the star caused by the advection of the disk energy (Baraffe et al.2012). Taking into account the uncertainty in the bolometric correction,M_*/R_* andβ, a mass accretion rate of at least a few 10⁻⁵M_⊙ yr⁻¹ is required to explain the observed luminosity. This is well within the range for FUors.

3.5. Emission and Absorption Lines in theK Band

The S/Ns obtained in theJ andH bands are low, and so we will focus on theK-band spectra. Several H₂ emission lines, 1-0 S (1), 1-0 S (2), 1-0 S (3), and 1-0 Q (1), are clearly visible in the IGRINS spectrum at 1.95756, 2.03376, 2.12183 and 2.40659μm. Additional H₂ lines 1-0 Q(2) (2.41344μm), 1-0 Q(3) (2.42373μm), 2-1 S(1) (2.24772μm) and 1-0 S(0) (2.22329μm) are also present. The 1-0 S(1), 1-0 S(2) and 1-0 S(3) lines are also seen in the P200 spectra. These lines are single-peaked and can be fitted by a Gaussian function withv = 18.9 ± 0.3 andσ = 12.9 ± 0.3 km s⁻¹ in the IGRINS spectrum (see Figure8). To detect weak lines, we have tied the centroid and width for all lines. H₂ is a factor of 2–4 narrower than those in classical TTS (Takami et al.2007) but broader than emission from the outer disks in Herbig Ae/Be stars (Carmona et al.2011). The line width suggests that it comes from the circumstellar material, rather than from the interstellar medium. Taking into account the motion of the Sun and Earth, the velocity with respect to the LSR is 40.1 km s⁻¹, which is consistent with the LSR velocity of CO lines from IRAS 06446+0319 (Lundquist et al.2015) considering all uncertainties. The measured line fluxes are listed in Table1. The weak lines 1-0 S (0) and 2-1 S (1) are sensitive to the exact subtraction of the continuum, which is affected by telluric OH absorption lines. Their flux errors can be as large as 50%. No other emission or absorption lines are seen. Neither recombination lines Brγ at 2.1661μm and He I nor CO lines are seen in absorption or emission. Photospheric metal absorption lines such as from Na I and Ca I, commonly seen in the IR spectra of low-mass YSOs (e.g., Faesi et al.2012), are not seen in the IGRINS spectrum.

Zoom InZoom OutReset image size

Table 1. Flux of Emission Lines from IGRINS

Line ID	Wavelength	Intensity
	(μm)	10−16 (erg s−1)
1-0 S(0)	2.22329	0.37 ± 0.05
1-0 S(1)	2.12183	1.89 ± 0.05
1-0 S(2)	2.03376	0.68 ± 0.05
1-0 S(3)	1.95756	1.34 ± 0.13
1-0 Q(1)	2.40659	1.98 ± 0.08
1-0 Q(2)	2.41344	0.75 ± 0.11
1-0 Q(3)	2.42373	1.12 ± 0.13
2-1 S(1)	2.24772	0.30 ± 0.06

Download table as: ASCIITypeset image

H₂ 1-0 Q(3) and 1-0 S(1) share the same upper level, as do 1-0 Q (2) and 1-0 S (0). The ratios of these lines can be used to estimate their extinction. Due to the weakness of 1-0 S(0), we will not use the latter line ratio. The observed ratio 1 − 0Q(3)/1 − 0S(1) = 0.77 ± 0.09 is consistent with the theoretical value of 0.70 (Turner et al.1977). Line ratios of H₂ 10 S (1) to 21 S (1) and 32 S (3) have traditionally been used to diagnose the excitation mechanism of H₂ molecules. In J0647, 10 S (1)/21 S (1) = 6.3 ± 1.3 is higher than the value 2–3 expected for UV excitation, but is consistent with shock excitation (Hollenbach & Natta1995). The presence of narrow H₂ emission lines is also consistent with a YSO.

H₂ emission lines are considered tracers of outflows in YSOs, and were detected in only three of nine FUors with S/N > 3σ (also V346 Nor, Kóspál et al.2020; V960 Mon, Park et al.2020) and a few EXors (Kóspál et al.2011; Hodapp et al.2019,2020; V899 Mon, Park et al.2021a). Most of these detections are made only in 2.218μm transition with the exception of two objects. When high-resolution spectra are available, the line widths are in the range of 20–40 km s⁻¹, much narrower than hydrogen lines or other metal emission lines whenever observed, consistent with being a large-scale outflow origin, rather than high-velocity outflows from the inner disks (e.g., Herbig2009).

We have argued that J0647 is a deeply embedded Class I protostar according to the pre-outburst SED, location at the edge of an SFR, consistency of H₂ velocity with CO₂ of the SFR, low quiescent luminosity, and absence of periodic fluctuations. The distance of the SFR, around 3.5–3.7 kpc, derived from the CO kinematic measurement, is also consistent with the constraint of extinction of foreground stars with known distances. The quiescent luminosity of 9L_☉ obtained in Section3.2, which is close to that of V1647 Ori (Aspin et al.2008), places it in the upper quarter of the Class I YSOs in the Spitzer and Herschel survey samples (Furlan et al.2016; Pokhrel et al.2023). If the accretion power dominates the total radiative output in the Class I phase, the bolometric luminosity provides an estimate of the quiescent mass accretion rate () to be a few 10⁻⁶M_☉ year⁻¹ for the typical mass and radius of a protostar. However, if the radiation is mainly powered by the protostar, the mass of the star is estimated to be around 0.7–1.3M_☉ for a stellar of an age less than 10⁷ yr according to the models of Siess et al. (2000). A lower limit on stellar mass of 0.58M_☉ is derived from the luminosity evolution models for exponentially declining infall rates (Fischer et al.2017).

J0647 exhibited a 500-fold increase in MIR brightness over a 2 yr period, followed by a slow decline. This eruption amplitude is the second largest recorded among all known YSO eruptions in the MIR (Lucas et al.2020). The tenfold decay within 6 yr is much slower than that of EXor eruptions but considerably faster than FUor eruptions. From the light curve alone, J0647 can be classified as an intermediate-type eruption with an exceptional amplitude such as WISEA 1422.

IR-only eruptions have been observed in several YSOs over the past decade, such as OO Ser (Kóspál et al.2007) and WISEA 1422. The most extreme example of this type is Hops 383, a Class 0 object, which had a 35-fold eruptive amplitude at 24μm (Safron et al.2015) and was not detected in theJ, H, andK bands up to 23.1, 21.8, and 20.4 mag, respectively (Fischer & Hillenbrand2017). Other sources without detection in theJ andH bands were reported in Nikoghosyan et al. (2017,2023). These sources, together with J0647, all had eruptions of intermediate duration, which is consistent with the finding that intermediate-type eruptions are the most commonly seen type in NIR time-domain surveys (Nikoghosyan et al.2017).

J0647 does not display detectable stellar photospheric absorption lines or CO absorption bandheads of a typical M supergiant in the NIR spectrum during the outburst, which is the benchmark for classical FU Orion objects (Greene et al.2008; Connelley & Reipurth2018). Neither shows Brγ or Pachen emission lines or CO bandheads in emission that are usually seen in EXor eruptions. Featureless spectra have been reported in a handful of YSO outbursts so far. OO Ser is the first object of this type (Kóspál et al.2007). The group includes three other objects ⁷ : 2MASS 22352345+7515076 (Kun et al.2019), WISEA 1422 (Lucas et al.2020), and VVVv815 (Contreras Peña et al.2017). All of them are deeply embedded sources.

Another interesting feature is that the W1–W2 color stays almost constant from the pre-outburst to the outburst at J0647. The constant shape can extend to the NIR band by comparing the NIR spectra obtained at different epochs. We can compare this with other YSOs in outbursts. Two FUors Gaia 17bpi and Gaia 18dvy showed a pattern of bluer when brighter in the MIR band during an eruption with an amplitude >6 mag in the W1 bands (Hillenbrand et al.2018). WISEA 1422 becomes redder from quiescence to the eruption, and with a short duration of turning blue after the peak, becomes even redder afterward (Lucas et al.2020). The V1647 Ori analogy ASASSN-13db has a color variation very similar to that of WISEA 1422. However, a wavelength-independent decay was observed for the OO Ser outburst between 1996 and 2005, while the pre-outburst MIR color is not available (Kóspál et al.2007).

In summary, J0647 is a deeply embedded YSO that has four distinct features: a medium-length eruption, an unusually high outburst intensity in the MIR, a featurelessK-band spectrum, and wavelength-independent changes in the NIR and MIR bands. Apart from the last one, it is similar to WISEA 1422. We will show that these peculiarities can be explained by a dusty disk outflow or an inflated outer disk, which also transforms primary radiation from stars and disks into longer IR wavelengths.

CO absorption bandheads in an outburst FUor are formed in the hot inner accretion disk (Connelley & Reipurth2018; Liu et al.2022), whose inner radius is pushed by high gas pressure to the surface of YSO when the accretion rate is high. The Brγ and Paschen emission lines in EXors are produced by gas photoionized by UV photons, generated in the shock of the magnetospheric accretion column for a low to moderate accretion rate (Guo et al.2021). In V899 mon, whose properties lie between FUor and EXor, CO absorption lines were observed in the bright outburst phase (Ninan et al.2015), while CO emission lines were detected when the source faded (Park et al.2021b), supporting this idea. Photospheric absorption lines are absent in J0647 because the YSO and the inner accretion disk are completely obscured. Recombination lines are not observed because the emission line region is obscured as well or because the accretion rate is high, so magnetospheric accretion is suppressed as in the case of a FUor. The high accretion rate in this object is supported by its high outburst luminosity.

The common trait of spectroscopically featureless outbursts is that the sources are deeply embedded. This is consistent with the above scenario. The obscurer could be the inflated outer disk, a dusty wind, or even dust in the envelope when the system is viewed edge-on. This same obscurer may also absorb the radiation from the inner accretion disk and the central star and reprocess it into MIR emission with a low effective temperature. As estimated in Section3.3, the MIR emission region has a size of at least 6.5 au, which is in the range of protoplanetary disks around Class I stars (Tobin et al.2020), but much smaller than the envelope. The high luminosity in the IR suggests that the reprocessor should have a large covering factor. The constant observed W1–W2 in J0647 implies that the reprocessing region expands (shrinks) as the source brightens (dims) because it is expected that the temperature of the dust will vary with the luminosity ofT ∝L^1/4 in a steady structure. This indicates that the obscurer is more likely a disk wind, as the timescale to adjust for the outer disk should be long. We hypothesize that this is the result of the increasing launching radius of the dusty outflow with the increase of disk luminosity.

Lucas et al. (2020) proposed a second possibility that the eruption of WISEA J142238.82−611553.7 could be caused by the fragmentation of the disk or the falling material from the envelope. Because the specific gravitational binding energy in situ (a size of > 6.5 au) is so small, it requires an unrealistically high mass infall rate (M_☉ yr⁻¹) to explain the high luminosity and large emission size for J0647. The total mass accreted during the outburst far exceeds the range for the disk or envelope in Class 1 YSOs. Therefore, the in situ scenario can be rejected here.

In summary, we anticipate the dust-obscuring and reprocessing scenario to explain the size of the emission region of the IR outburst, the featurelessK-band spectrum, and the constant IR colors. To confirm the feasibility of this, further work is needed, including detailed radiation transfer modeling and numerical simulations of disk winds under extreme changes in the accretion rate of the inner disks.

We discovered a long-lasting and massive outburst in J0647, which was only detected at wavelengths longer than theK band when quiescent. The source was identified as an embedded Class I YSO from its association with an SFR, the SED, and the bolometric luminosity in quiescence. Its NIR spectrum is different from that of classical FUors, EXors, or many other known intermediate-type outbursts in YSO due to its lack of absorption or emission lines other than H₂. Its SED during the outburst was dominated by MIR emission with an effective temperature of 600–700 K and a weak NIR excess component. We interpret the observed properties in the context of an obscured accretion outburst. Both the YSO and the inner accretion disk in J0647 are obscured by a thick dusty outflow or cold outer disk/envelope, which reprocesses the stellar and disk emission into MIR light. The weak NIR excess is probably the scattered accretion disk emission. The size of the reprocessing emission region is >6.5 au. As a result, photospheric absorption lines from stellar and disk photospheres are not directly seen. We estimate a maximum mass accretion rate of the outburst of at least a few 10⁻⁵M_☉ yr⁻¹. The fading rate (0.5 mag yr⁻¹) is steeper than FUors, but much shallower than EXors. The variability amplitude in the MIR is only one magnitude smaller than the outburst in another serendipitously discovered WISEA 1422 (Lucas et al.2020), whose NIR spectrum is also similar to J0647 with a nearly featureless continuum with the exception of the H₂ emission line, but the spectrum was taken when the source was 6 mag below the peak. This source is also identified as an embedded Class I YSO. The color changes in the two sources are different. The color of W1–W2 remains nearly constant in J0647 during the outburst, while it changes dramatically in WISEA 1422 and outbursts in other YSOs. This may indicate the presence of a population of heavily embedded outbursts. The true number of such sources and their contribution to our understanding of the mass growth of YSOs have yet to be fully explored.

The authors thank Dr. Gregory Herczeg for the thorough reading of the first manuscript version and constructive suggestions. We thank the referee for the helpful comments. This work is supported by NSFC funding (NSFC-11833007), the China Manned Space Project (No. CMS-CSST-2021-A13), the Ministry of Science and Technology (2022SKA0130102), and the Cyrus Chun Ying Tang Foundations.

This work used IGRINS, which was developed under a collaboration between the University of Texas at Austin and the Korea Astronomy and Space Science Institute (KASI) with the financial support of the Mt. Cuba Astronomical Foundation, the US National Science Foundation under grants AST-1229522 and AST-1702267, McDonald Observatory of the University of Texas at Austin, the Korean GMT Project of KASI, and Gemini Observatory. These results were obtained using the Lowell Discovery Telescope (LDT) at Lowell Observatory. Lowell Observatory is a private, non-profit institution dedicated to astrophysical research and public appreciation of astronomy and operates the LDT in partnership with Boston University, the University of Maryland, the University of Toledo, Northern Arizona University, and Yale University. TSpecs are obtained through the Chinese TAP program.

This research has used the NASA/IPAC Infrared Science Archive, which is funded by the National Aeronautics and Space Administration and operated by the California Institute of Technology.”

The Pan-STARRS1 (PS1) Surveys have been made possible through contributions of the Institute for Astronomy, the University of Hawaii, the Pan-STARRS Project Office, the Max-Planck Society, and its participating institutes, the Max-Planck Institute for Astronomy, Heidelberg, and the Max-Planck Institute for Extraterrestrial Physics, Garching, The Johns Hopkins University, Durham University, the University of Edinburgh, Queen's University Belfast, the Harvard-Smithsonian Center for Astrophysics, the Las Cumbres Observatory Global Telescope Network Incorporated, the National Central University of Taiwan, the Space Telescope Science Institute, the National Aeronautics and Space Administration under grant No. NNX08AR22G issued through the Planetary Science Division of the NASA Science Mission Directorate, the National Science Foundation under grant No. AST-1238877, and the University of Maryland.

Facilities: IRSA - , Spitzer - Spitzer Space Telescope satellite, WISE - Wide-field Infrared Survey Explorer, Herschel - European Space Agency's Herschel space observatory, Hale - , LDT - , PO:1.2m - .