OIII arc in Gemini

An arc-shaped emission nebula in the constellation Gemini, with dominant [OIII] lines and an apparent length of about 17°, was detected in data from the Northern Sky Narrowband Survey. The object could be a fragment of a supernova remnant from the same progenitor as the pulsar Geminga, and it may interact with the Monogem Ring. This possibility is discussed below, after a presentation of the images and object data in the next section.

Object data

The center ot the OIII arc is located at a right ascension of 7:41:48 h and a declination of 19:56°. Its size is about 17° in the north/south direction and 2.5° in the east/west direction.

[OIII]	Hα
[SII]	Luminance measurement regions

The following table contains the results of intensity measurements for the regions marked in the upper image. The average intensity in Rayleighs is given in each case. The region labeled bg was used as a background reference.

Region	Hα	[OIII]	[SII]
K1a	0.02	2.04	0.06
K1a'	1.68	1.53	1.14
K1b	2.20	0.63	2.27
K1c	0.24	0.72	1.54

For the determination of the intensity, some parameters had to be estimated. The transmission ratio of the optics and filters was assumed to be 75% for [OIII] and Hα and 70% for the [SII] doublet. The atmospheric extinction under best conditions was assumed to be 20%. The errors resulting from these estimations only distort the absolute measured values by a factor, but have no influence on the intensity ratios between the wavelengths. More relevant are the errors caused by the background calibration, as they distort the intensity ratios between the individual emission lines. The accuracy of the background reference (the two regions marked bg in the image above) is limited by the fact that the entire region is full of faint background nebulae.

Classification and discussion of the superstructure

The arc seems to be a part of a larger structure. To better assess the situation, a wide field view has been created from data of DR0.2 of the Northern Sky Narrowband Survey.

Visual light

A closer look at the faint filaments east of the arc (left in the images) reveals that they seem to cross the arc, as sketched in panel (a) of the next figure. Furthermore, unlike the arc, these filaments have a structure similar to the other SNR (red and blue) near the galactic plane. Connecting them seems natural. The yellow dotted ellipse in panel (b) of the following figure shows the outline of the projection of a hypothetical bubble containing the SNR filaments that appear to belong together.

(a)

(b)

X-rays

In X-rays, a SNR denoted as the Monogem Ring can be observed. Its progenitor is most likely the same as that of the Pulsar PSR B0656+1. This ring is plotted in white in panel (b) of the figure above. Size and position are taken from Knies et al. (2024), who analyzed the region using the latest X-ray data from the eROSITA telescope array. The same authors also propose the existence of a second SNR in a similar distance and with a similar size. This two SNR scenario is plotted in magenta in the image above, where the eastern (left) SNR is designated G205.6+12.4 while the western one is still denoted as Monogem Ring.

Both scenarios are a good match to the visual structures, including the faint filaments east of the arc. However, the arc does not appear to be part of the Monogem Ring or G205.6+12.4.

Far UV

The arc can also be seen in Galex far-UV (134 nm to 179 nm) images as a thin filament. However, since the data is incomplete and the region is full of similar structures, the arc cannot be recognized as a separate object in Galex images.

Far UV spectroscopy was performed by Kim et al. in the range of 90 nm to 175 nm. In the low resolution CIV (154.8 nm and 155.0 nm) emission lines image (Fig. 1 in their work), only the region corresponding to the southern part of the arc (at a galactic longitude from 201° to 208°) becomes visible. In the paper, this region is denoted as R1, and it approximately coincides with the region where the arc apparently overlaps with the Monogem Ring or G205.6+12.4, respectively. Due to the low resolution, it is uncertain whether these emissions originate from the SNR or are the result of some interaction. It is very unlikely that the emissions come directly from the arc, because the northern part (at a galactic longitude from 193° to 201°) is not visible. Furthermore, the authors detected only CIII and CIV emissions, but no OIII. In visible light, however, [OIII] is dominant. The authors interpret this structure as “the blast wave [of the Monogem SNR] with an isolated cloud”.

Interaction between Monogem Ring and OIII arc?

There is a noticeable similarity between the emission line intensities of the arc and the position of the Monogem Ring or G205.6+12.4:

• In the southern part, where the arc and the Monogem Ring apparently overlap, the arc is visible only in [OIII].
• Near the apparent boundary of the Monogem Ring, there is a region where only Hα and [SII] emissions are detected. In this region, the arc also appears distorted.
• In the northern part, which lies outside the Monogem Ring, all three observed emission lines are visible.

If the ionization of the arc is caused by the cooling (and subsequent recombination) of hot plasma, this correlation could be explained by differences in temperature. It should be mentioned that interaction can only take place within the Monogem Ring or on its shell, because the Monogem Ring expands with hypersonic velocity.

3D structure of the arc

The OIII arc is probably (a patch of a) thin shell. What we see is the two-dimensional projection of this structure. Even if the shell is assumed to be spherical, it is difficult to reconstruct it from only a small fragment. However, due to the high aspect ratio, the projected arc must lie close to the perimeter of the projected sphere.

A rough estimate of the region where the center of a spherical shell could be located is indicated by the yellow dotted ellipse in the figure below. The yellow square marks the center that would be expected if the arc lies exactly on the apparent boundary of a spherical shell.

Ghost object?

A search for similar fragments along the approximate position of the spherical shell yielded no results. One possible explanation is that the object is moving rapidly, so that hypersonic velocities relative to the ambient interstellar medium (ISM) only occur in the forward direction of motion. A shock front would then form only in that direction. At least a fraction of the ionization energy would come from the kinetic energy inherited from a high-velocity progenitor.

Planetary nebulae with these properties are described by Ogle et al. (2025) and are referred to there as ghost planetary nebulae (GPN), because they would be invisible without a high velocity relative to the ISM. A SIMBAD search for nearby (<200pc, corresponds to an arc size of <60pc) white dwarfs or hot subdwarfs with high proper motion (>100 mas/yr, corresponds to 10 km/s at 200 pc) within a 60°-diameter region around the equatorial coordinates (RA,DEC) = (99.7°, 15.5°) yielded no candidates that could plausibly have formed the arc as a GPN.

Geminga (ghost) SNR?

Another object, however, does move in the right direction: the pulsar Geminga. The magenta arrow in the previous figure traces its motion, beginning at its birthplace (Pellizza et al. (2019) and Faherty et al. (2007)) and ending at its current position.

Could the OIII arc be a SNR formed by Geminga’s progenitor? Could its shape be explained by Geminga’s motion? To address these questions, more data must be gathered from the literature.

Data

Faherty et al. (2007) estimate Geminga's current distance at 250 (+120  / -70) pc based on HST measurements. Pellizza et al. (2019) reported an age of 342 kyr, a maximum progenitor mass of 15 solar masses, and a birthplace distance between 90 and 240 pc. Unfortunately, these distance estimates rely on an outdated and inaccurate parallax measurement for Geminga’s current position – 157 (+59 / -34) pc (from Caraveo et al., 1996; see Faherty et al., 2007 for a discussion). I therefore adopt the same distance for the birthplace as for the current position, since this would also explain the possible interaction with the Monogem Ring (see below). In addition, a 0° inclination of the bow shock appears to provide the best fit in Figure 3 of Caraveo et al. (2003). With these parameters, Geminga's apparent motion of 17° corresponds to a travel distance of about 74 pc and a velocity of 210 km/s.

Modelling

The upper limit of the kinetic energy inherited from the progenitor is therefore less then 6.0·10⁴⁸ erg (velocity: 210 km/s; mass: 15 solar masses of the progenitor minus 1.4 solar masses of the neutron star). This value is far below the characteristic explosion energy of 10⁵¹erg. Thus, the motion should have no significant impact on SNR evolution until most of the explosion energy is radiated away in the snowplough phase.

It is further assumed that the kinetic energy inherited from the progenitor is preserved (i.e., not radiated away through shock ionization) and can be treated independently of the SNR energetics. This implies that the SNR as a whole will be slowed down as more mass is swept up.

These assumptions, combined with the analytic SNR evolution approximation of Blondin et al. (1998), form the basis of a simple simulation written in C. The underlying mathematics and usage instructions are documented in the comments of the source code (see link).

Simulation results

The remaining parameters used for the simulation are a characteristic explosion energy of 10⁵¹ erg and a particle density of 1.7·10^-2 cm^-3. The density was fitted to match the position of the OIII arc, as described later in this section. The results are shown in magenta in the figure above. The small square marks the simulated center of the SNR, located 31 pc from its birthplace and halfway between the birthplace and Geminga’s current position.

Because the mass swept up by the motion (not by the expansion) is large compared to the initial mass, the SNR is decelerated to a velocity of about 49 km/s, or 23% of the initial speed. Although this is probably greater than the sound speed in the ISM (depending on temperature), this does not affect visibility, since the expansion speed remains much higher – about 130 km/s. Thus, the resulting velocity relative to the ambient ISM is likely supersonic across the entire shell, meaning the SNR cannot be categorized as a ghost object.

The radius of the SNR according to the evolution model from Blondin et al. (1998) is 116 pc, shown as the dotted magenta circle in the figure above. The three-dimensional situation is sketched in the next figure, which also illustrates the possible interaction with the Monogem Ring. With the assumed distances of 300 pc for the Monogem Ring (Knies et al., 2024) and 250 pc for both the birthplace and current position of Geminga (and thus also the center of the Geminga SNR), this is a perfect match: The [OIII]-only part lies within the Monogem Ring, the Hα+[SII] part lies on the boundary, and the region emitting all three wavelengths lies outside. However, this alignment depends strongly on the uncertain distances. For example, adopting the (unlikely) birthplace distances from Pellizza et al. (2019) would place the entire arc outside the Monogem Ring.

The perfect match was also obtained by fitting the ISM density to 1.7·10^-2 cm^-3. With a density of 5·10^-3 cm^-3 (density around the Monogem Ring derived by Knies et al., 2024), the SNR radius would be considerably larger – about 142 pc – and the travel distance increase to 40 pc.

The simulation results depend mainly on ISM density and distance (and only weakly on explosion energy, progenitor mass, and velocity). The critical parameters are highly uncertain. Within reasonable limits, a change in one parameter can usually be compensated by adjusting the others. The same applies to the models themselves: a more advanced simulation might yield different results, but as long as the parameters remain uncertain, compensating adjustments should be possible.

If the sketch in the following figure correctly represents the 3D situation, the distance of the OIII arc would be about 260 pc. At this distance, the length of the arc would be almost 80 pc.

Discussion and conclusion

There are two Hα+[SII] filaments that cannot be assigned to any known structures in the region and may therefore belong to the Geminga SNR. These are marked by a white dotted line in the figure above. The longer structure crosses the galactic plane and thus cannot be associated with the Orion-Eridanus superbubble, which is located below the galactic plane.

It is also possible that some filaments usually attributed to the Orion-Eridanus superbubble actually belong to the Geminga SNR. Depending on its 3D orientation (see Pon, 2016), the Orion-Eridanus superbubble may overlap with the Geminga SNR as constructed here. However, the intersection is too small to explain why only a small part of the Geminga SNR is visible.

According to the simulation in Blondin et al. (1998), the Geminga SNR (as constructed here) is in transition between the Sedov phase and the radiative (or snowplough) phase. This transition is characterized by instabilities. I could not found analytical models for the temperature in this phase, but the 2.7·10⁵ K predicted by Padmanabhan (2001) for the Sedov phase can likely be treated as an upper limit. At this temperature, plasma recombines and the SNR should become visible at optical wavelengths. Could the instabilities account for the discrepancy between prediction and observation? Cloud the velocity inherited from the progenitor make the difference, even if it is small compared to the expansion velocity?

Another possible explanation for the discrepancy between prediction and observation is a non-uniform ISM. In denser regions, SNRs evolve faster and remain smaller. If the SNR exploded into a low-density region toward the Monogem Ring, the OIII arc may be the only part still visible, while the rest has already cooled and possibly merged with the ISM.

An interaction between the Monogem Ring and the Geminga SNR could explain the differing emission line ratios, but not the OIII arc as a whole, since part of the arc always lies outside the Monogem Ring.

In general, the size (about 80 pc for the OIII arc and >200 pc for the entire SNR if uniform density is assumed) should not be surprising. A look at other galaxies, e.g. M101 and NGC 2303, reveals [OIII]-only structures in outer (low-density) regions with diameters of more than 400 pc.

In conclusion, it seems plausible that the OIII arc belongs to the Geminga SNR. At present, this is also the only explanation for the object. Nevertheless, for a secure classification, many questions remain open – especially why only a small fragment is visible.

Further observations

Weinberger et al. discovered a [SII]+Hα filament. which is also visible in the wide field image presented above. A few similar objects are located in the same region. Some of them are bright in [SII], while others are bright in [OIII]. They appear to be extensions of the elongated southern filament of the Monogem Ring or G205.6+12.4, respectively. The mentioned objects can be found in this Javascript presentation.

Image data

Images where captured with a camera array which is described on the instruments page.

Image data are:

	View #1	View #2
Center position:	RA: 7:44h, DEC: 18:30°	RA: 6:12h, DEC: 14°
FOV:	8.75°×14.31° (RA×DEC, through center)	60°×60° (RA×DEC, through center)
Orientation:	North is up	North is up
Scale:	10 arcsec/pixel (in center at full resolution)	10 arcsec/pixel (in center at full resolution)
Projection type:	Stereographic	Stereographic
Exposure times (Sum of exposure times of all single exposures used to calculate the image):	Hα: 80 h OIII: 55 h SII: 52 h	Data from DR0.2 of the Northern Sky Narrowband Survey

Image processing

Both images were processed differently. The second image (60°×60° FOV) was calculated from DR0.2 of the Northern Sky Narrowband Survey. (Follow the link for a detailed description.) The first image (8.75°×14.31° FOV) is older and is not continuum-subtracted (in contrast to the newer one). Its processing is described as follows.

All image processing steps are deterministic and none of the algorithms use machine learning (often referred to as “AI”), which tends to generate plausible looking fake details. The software used can be downloaded here.

The image processing steps were:

1. Bias correction, dark current subtraction, flatfield correction, noise estimation
2. Alignment and brightness calibration using stars from reference image
3. Stacking with outlier rejection, background estimation and optimal weighting based on noise estimation
4. Star subtraction where star positions and intensities are extracted from continuum images
5. Denoising and deconvolution of both components (stars and residual)
6. Dynamic range compression using non-linear high-pass filter
7. Color composition and tonal curve correction

Citing / persistent resources

If referencing a volatile webpage is not appropriate, a persistent snapshot of this page is available via the DOI 10.5281/zenodo.17215566.

References

 

John M. Blondin, Eric B. Wright, Kazimierz J. Borkowski, and Stephen P. Reynolds. Transition to the Radiative Phase in Supernova Remnants. ApJ, 500(1):342–354, June 1998. [ DOI ]

 

Patrizia A. Caraveo, Giovanni F. Bignami, Roberto Mignani, and Laurence G. Taff. Parallax Observations with the Hubble Space Telescope Yield the Distance to Geminga. Astrophysical Journal Letters, 461:L91, April 1996. [ DOI ]

 

P. A. Caraveo, G. F. Bignami, A. De Luca, S. Mereghetti, A. Pellizzoni, R. Mignani, A. Tur, and W. Becker. Geminga's Tails: A Pulsar Bow Shock Probing the Interstellar Medium. Science, 301(5638):1345–1348, September 2003. [ DOI ]

 

Jacqueline Faherty, Frederick M. Walter, and Jay Anderson. The trigonometric parallax of the neutron star Geminga. Astrophysics and Space Science, 308(1-4):225–230, April 2007. [ DOI | http ]

 

I. J. Kim, K. W. Min, K. I. Seon, J. W. Park, W. Han, J. H. Park, U. W. Nam, J. Edelstein, R. Sankrit, and E. J. Korpela. Far-Ultraviolet Observations of the Monogem Ring. The Astrophysical Journal, 665(2):L139–L142, August 2007. [ DOI | http ]

 

Jonathan R. Knies, Manami Sasaki, Werner Becker, Teng Liu, Gabriele Ponti, and Paul P. Plucinsky. A new understanding of the gemini-monoceros x-ray enhancement from discoveries with erosita, 2024. [ arXiv | http ]

 

Patrick Ogle, Mark Petersen, Tim Schaeffer, Lewis McCallum, Alberto Noriega-Crespo, R. Michael Rich, Biny Sebastian, Carl Bjork, Steeve Body, Sendhil Chinnasamy, Marcel Dreschsler, Tarun Kottary, Yann Sainty, Patrick Sparkman, and Xavier Strottner. SDSO1 is a Ghost Planetary Nebula Bow Shock in Front of M31. arXiv e-prints, page arXiv:2507.15834, July 2025. [ DOI | arXiv ]

 

T. Padmanabhan. Theoretical Astrophysics, Vol 2. 2001. [ DOI ]

 

L. J. Pellizza, R. P. Mignani, I. A. Grenier, and I. F. Mirabel. On the local birth place of Geminga. Astronomy and Astrophysics, 435(2):625–630, May 2005. [ DOI | arXiv | http ]

 

Andy Pon, Bram B. Ochsendorf, João Alves, John Bally, Shantanu Basu, and Alexander G. G. M. Tielens. Kompaneets Model Fitting of the Orion-Eridanus Superbubble. II. Thinking Outside of Barnards Loop. ApJ, 827(1):42, August 2016. [ DOI | arXiv | http ]

 

R. Weinberger, S. Temporin, and B. Stecklum. Detection of an optical filament in the Monogem Ring. Astronomy and Astrophysics, 448(3):1095–1100, March 2006. [ DOI | arXiv | http ]