PDA

View Full Version : SN1987A in triplicate?



Hat Monster
2005-Dec-30, 08:56 PM
I'm not so sure about this (http://www.geocities.com/peaceharris/sn1987a/index.html) since it seems just a little shaky and relies on a pretty hefty coincidence. Thoughts from more experienced data crunchers?

Jerry
2005-Dec-31, 01:57 AM
Really shaky - the link does not work.

andyschlei
2005-Dec-31, 02:07 AM
The link worked fine for me, but I am in no position to comment, so I won't.
--Andy

trinitree88
2005-Dec-31, 05:07 AM
I'm not so sure about this (http://www.geocities.com/peaceharris/sn1987a/index.html) since it seems just a little shaky and relies on a pretty hefty coincidence. Thoughts from more experienced data crunchers?


I believe the video..NOVA "Death of a Star" circa 1987-8 shows astrometry done in Baltimore after SN1987a. Kirshner(Harvard) and the EUVE imaged two of the trio still there after the explosion..there was initially some controversey because there was a close double that required digital separation in the glare of SN1987a. Sanduleak 69 202 is gone, that we know.

dgavin
2005-Dec-31, 06:33 PM
V388 Mon a new red giant formed in 2000(?) also emited three light pulses before expanding into a red giant (rather suddenly). These pulses had different spectograph lines.I suspect that when any star evolves, that the transitions that take a while to build, happen quicker then thought, and involved the core destabilizing then compressing and expanding in pulses which are reflected in light curves.

The Bad Astronomer
2006-Jan-01, 05:47 AM
I may be considered something of an expert on 1987A (http://www.badastronomy.com/info/sin.html).

His thesis is trivially easy to show wrong: Stars 2 and 3 have been observed for years and have not changed in brightness at all. Also, spectra have been taken showing them to be typical supergiants in the Large Magellanic Cloud.

His idea about 3 light echoes is immaterial. The LMC is a dusty place, and we expect to see multiple rings from a single source.

His idea about the light curve having three peaks is just odd. Most supernovae peak at different times in different colors. The one in M51 this past year (2005cs) did, for example.

The three rings have nothing at all to do with the supernova, so he's way off base there. They existed for thousands of years before the supernova. If the were ejecta from the explosion they would appear to expand, and they don't. Also, we see similar rings from a star that is similar to the progenitor of 1987A. This star is called Sher 25, and I wrote an essay about it (http://www.badastronomy.com/info/logo.html) (I also refereed a journal paper about it some years ago). Since Sher 25 has not yet exploded, and the rings are there, that is further evidence that the rings around 87A were there before the star blew up.

Finally, there's good old parsimiony. It seems a bit extreme to posit three supernovae, which begs a LOT of questions, rather than assume one, which fits the data pretty well.

peaceharris
2006-Jan-09, 09:36 AM
Happy New Year!!!

I just found out that you guys have been discussing my views.


Stars 2 and 3 have been observed for years and have not changed in brightness at all.

This statement is not correct. Refer PASP V 105, pg 1240 (http://adsabs.harvard.edu/cgi-bin/nph-bib_query?1993PASP..105.1240W&db_key=AST):

In that article, take note that the authors assumed star 2 to be constant and also look at fig 6.

ngc3314
2006-Jan-10, 04:25 AM
Happy New Year!!!

I just found out that you guys have been discussing my views.



This statement is not correct. Refer PASP V 105, pg 1240 (http://adsabs.harvard.edu/cgi-bin/nph-bib_query?1993PASP..105.1240W&db_key=AST):

In that article, take note that the authors assumed star 2 to be constant and also look at fig 6.

They do not assume star 2 to be constant - their Fig. 4 shows that at least 27 observations in each passband are all consistent with it being constant. Star 3 does vary by about 0.5 magnitude. This is utterly inconsistent with it having been a supernova; the data are from 1990-92, 3-5 years after SN 1987A was observed, and the star is not substantially fainter than seen in archival images from before SN 1987A. It didn't blow up (and by the same token neither did Star 2).

There are also issues with how accurately spatial centroids can be retrieved from IUE line-by-line files, but I'll have to pull out numbers when back at work where there is a set of old IUE newsletters with picky details on the processing steps.

peaceharris
2006-Jan-11, 03:08 AM
Based on the average of images taken during the period January - April 1989, Walker and Suntzeff computed the V magnitude Star 2 and 3 to be 14.88 and 15.60 respectively. Refer PASP V 102, pg 131 (http://adsabs.harvard.edu/cgi-bin/nph-bib_query?bibcode=1990PASP..102..131W&db_key=AST)

Based on these images they had computed the magnitude of another 35 stars close to SN1987a

Towards the end of that paper, they have a note added in press:

CCD photometry for stars 2 and 3 (Table 4) on 4 nights in October - November 1989 gives the following improved magnitudes ...

The V magnitude of Star 2 reduced to 14.96, while Star 3 reduced to 15.82.

They didn't make any correction to the other 35 stars.



They do not assume star 2 to be constant - their Fig. 4 shows that at least 27 observations in each passband are all consistent with it being constant.

To compute the magnitude of a star from a photo, you need to compare that star to a reference star which is assumed to be constant. This is neccessary since the exposure time of the photo and atmospheric conditions can vary.
Since you claim that they didn't assume Star 2 to be constant, please tell me which star they used as their reference. Let me quote from page 1241 (http://adsabs.harvard.edu/cgi-bin/nph-bib_query?1993PASP..105.1240W&db_key=AST):

"The success of the adopted procedure can be gauged by the lack of any significant systematic trends in the magnitudes of Star 2 (assumed constant) during the time that SN1987a declined by some 6 magnitudes."



Star 3 does vary by about 0.5 magnitude. This is utterly inconsistent with it having been a supernova;

The fact that we have observed Star 2 and Star 3 decline in brightness might suggest that Star 2 and 3 are plateau type SNe, where after an initial rapid decline, they reach a plateau stage where the light curve declines very slowly. This observation has been made for other plateau type SNe (http://www2.arnes.si/~gljsentvid10/supn1.html)

ngc3314
2006-Jan-11, 01:53 PM
To compute the magnitude of a star from a photo, you need to compare that star to a reference star which is assumed to be constant. This is neccessary since the exposure time of the photo and atmospheric conditions can vary.
Since you claim that they didn't assume Star 2 to be constant, please tell me which star they used as their reference. Let me quote from page 1241 (http://adsabs.harvard.edu/cgi-bin/nph-bib_query?1993PASP..105.1240W&db_key=AST):

"The success of the adopted procedure can be gauged by the lack of any significant systematic trends in the magnitudes of Star 2 (assumed constant) during the time that SN1987a declined by some 6 magnitudes."




Your statement is seriously incomplete for photometry using digital detectors. The reference stars can be anywhere in the sky, not necessarily in the same field, as long as one makes adequate measurements to determine the atmospheric extinction by looking at reference stars (of which there are several extensive networks around the sky) at a range of altitudes above the horizon. In section 2.2, they state that their calibration traces to the (high-quality) reference star list by Arlo Landolt (they reference a 1992 publication). The remark about star 2 being assumed constant means, in context, that not only are its measurements consistent with it being constant, but if they assume it is for the purpose of giving an outside range to their error bars, the trend in star 3 remains. If they simply assumed it was constant, noting a lack of systematic trends in their data would be a complete tautology.

peaceharris
2006-Jan-12, 04:08 AM
Also, spectra have been taken showing them to be typical supergiants in the Large Magellanic Cloud.

Which are the other supergiants in the LMC that have the same spectra as Star 3?

Here's the spectra of Star 3:

http://archive.stsci.edu/cgi-bin/coadd_plot?FOS=Y0WY0806T

One of the differences between a normal star and a supernova remnant is that fusion reactions in the core of a SNR do not generate enough heat to prevent the star collapsing inward. If the core is colder than the surface, we will observe emission lines. Normal stars exhibit absorption lines.

ngc3314
2006-Jan-12, 01:49 PM
Which are the other supergiants in the LMC that have the same spectra as Star 3?

Here's the spectra of Star 3:

http://archive.stsci.edu/cgi-bin/coadd_plot?FOS=Y0WY0806T

One of the differences between a normal star and a supernova remnant is that fusion reactions in the core of a SNR do not generate enough heat to prevent the star collapsing inward. If the core is colder than the surface, we will observe emission lines. Normal stars exhibit absorption lines.

The emission lines in that aperture spectrum are utterly dominated by so-called nebular lines, those arising at very low density. Notably these include O++ at 4959, 5007 A and the N+ lines flanking H-alpha at 6548,6583 A. This spectrum therefore has an important contribution from the nebula known to surround SN 1987A, and those lines tell you nothing about the star itself. That's not a huge surprise, since the spectrum was obtained in April 1992 - before the first Hubble servicing mission - when the Faint-Object Spectrograph could not have isolated a star in such a crowded region no matter how small an aperture was used.

There are other reasons stars can show emission lines, as in many O stars and Wolf-Rayet stars. Chief among these are stellar winds, where the lower density allows different radiative processes to influence what we see (compared to the dense photosphere).

Can you name a single plateau supernova which remained anything like as bright as 10 magnitudes below its peak 15 years after the explosion? That seems to be what you're claiming for stars 2 and 3, which are demonstrably still there at brightnesses not very different from their pre-1987 data.

peaceharris
2006-Jan-13, 05:12 AM
The emission lines in that aperture spectrum are utterly dominated by so-called nebular lines, those arising at very low density. Notably these include O++ at 4959, 5007 A and the N+ lines flanking H-alpha at 6548,6583 A. This spectrum therefore has an important contribution from the nebula known to surround SN 1987A, and those lines tell you nothing about the star itself. That's not a huge surprise, since the spectrum was obtained in April 1992 - before the first Hubble servicing mission - when the Faint-Object Spectrograph could not have isolated a star in such a crowded region no matter how small an aperture was used.

The 3 stars were brighter than the 3 nebulous rings in 1992, so it is natural to assume that the spectra of the star is dominant.

The central equatorial ring was much brighter than both the outer rings. The size of the FOS slit was not big enough to include both star 3 and the central equatorial ring. To get an idea of the size of the FOS slit, refer to Fig 1 of Apj V 466, pg 998.

Only the southern outer ring could potentially contaminate the spectra of star 3. However if you want to argue that emission lines from the southern outer ring is dominant in the spectra of star 3, then using that same logic you can argue that all astronomers who have written papers about the emission lines from SN1987a are mistaken too since the northern outer ring intersects the central SN, and it is not possible for any slit to isolate the SN from the NOR. Refer to Fig 1 of ApJ V 459, L 17.

The spectral image of SN1987a taken in April 1997 also proves that star 3 was an emission line object. Refer ApJ V 492 L139 (http://adsabs.harvard.edu/cgi-bin/nph-bib_query?bibcode=1998ApJ...492L.139S&db_key=AST). In Fig 3A, at around 4860A, the emission of star 3 is enchanced above the continuum. Refer Fig 1b: It's a pity that star 2 wasn't included in the aperture.


Can you name a single plateau supernova which remained anything like as bright as 10 magnitudes below its peak 15 years after the explosion? That seems to be what you're claiming for stars 2 and 3, which are demonstrably still there at brightnesses not very different from their pre-1987 data.

I do not know of any study regarding the variation of stars 2 and 3 done after 1992. Please give me a reference to a similar study done after PASP V105 pg 1240.

ngc3314
2006-Jan-13, 05:20 PM
The 3 stars were brighter than the 3 nebulous rings in 1992, so it is natural to assume that the spectra of the star is dominant.

The central equatorial ring was much brighter than both the outer rings. The size of the FOS slit was not big enough to include both star 3 and the central equatorial ring. To get an idea of the size of the FOS slit, refer to Fig 1 of Apj V 466, pg 998.

Only the southern outer ring could potentially contaminate the spectra of star 3. However if you want to argue that emission lines from the southern outer ring is dominant in the spectra of star 3, then using that same logic you can argue that all astronomers who have written papers about the emission lines from SN1987a are mistaken too since the northern outer ring intersects the central SN, and it is not possible for any slit to isolate the SN from the NOR. Refer to Fig 1 of ApJ V 459, L 17.

The spectral image of SN1987a taken in April 1997 also proves that star 3 was an emission line object. Refer ApJ V 492 L139 (http://adsabs.harvard.edu/cgi-bin/nph-bib_query?bibcode=1998ApJ...492L.139S&db_key=AST). In Fig 3A, at around 4860A, the emission of star 3 is enchanced above the continuum. Refer Fig 1b: It's a pity that star 2 wasn't included in the aperture.



I do not know of any study regarding the variation of stars 2 and 3 done after 1992. Please give me a reference to a similar study done after PASP V105 pg 1240.

For those who have forgotten, HST was launched with serious spherical aberration traced to the primary mirror. Even the smallest FOS apertures used before the Dec. 1993 servicing mission sampled a blurred focal plane, so the most you could manage with a tiny aperture was to maximuize the contract of a star with surrounding nebulosity.

As to variability: I just grabbed a set of 6 different epochs' worth of WFPC2 blue (4390 A) imagery from the archive, this being one from each date with a short enogh exposure not to saturate the core of the image of star 3 too much for accurate photometry. These are the results of simple Gaussian-fitting photometry, using an approximate conversion from that filter to standard B magnitudes:

Date star 3 counts/100s B magnitude
1994 Sept 24 6948 18.69
1995 Mar 05 8577 18.46
1996 Feb 06 7229 18.55
1997 July 10 6559 18.76
2001 Mar 23 6326 18.79
2001 Dec 07 9337 18.37

There is no monotonic trend in these numbers, just erratic variability consistent with the early-1990s data. Star 2, being brighter, was saturated on all but 2 of the images, which gave B=17.80+/-0.08 from these measurements (whch were not all that careful).

peaceharris
2006-Feb-03, 09:19 AM
These are the results of simple Gaussian-fitting photometry, using an approximate conversion from that filter to standard B magnitudes:

Date star 3 counts/100s B magnitude
1994 Sept 24 6948 18.69
1995 Mar 05 8577 18.46
1996 Feb 06 7229 18.55
1997 July 10 6559 18.76
2001 Mar 23 6326 18.79
2001 Dec 07 9337 18.37


I have a couple of questions:

1)Could you please tell which software you used to produce the above numbers?

2)When you say Gaussian-fitting, are you trying to fit the magnitude and the standard deviation of the Gaussian curve? I mean a 1-dimensional gaussian is y=A*exp (-x^2/B). Are you trying to vary 'A', keeping 'B' constant, or are you varying both A and B until the data matches the Gaussian curve?

I need to understand what you are doing before I reply to your post.

Blob
2006-Jul-15, 12:37 AM
In 1987 a massive star (Sanduleak -69202) exploded in the Large Magellanic Cloud, a neighbouring dwarf galaxy., an event called a supernova. It was the closest supernova to Earth since the invention of the telescope centuries ago. Now, a team using the Spitzer Space Telescope and the 8-meter Gemini South infrared telescope in Chile have probed the supernova remnant and found the building blocks of rocky planets and all living creatures.

Using infrared telescopes, said Dr. Eli Dwek, a cosmic dust expert at NASA Goddard Space Flight Centre and his colleagues have been following this supernova for a year, and have detected silicate dust created by the star from before it exploded. This dust survived the intense radiation from the explosion. Nearly 20 years onward, the supernova shock wave blasting through the debris that was shed by the star prior to its fiery death is now sweeping up this dust, making the material "visible" to infrared detectors.

Dust -- chemical particles and crystals finer than beach sand -- is both a frustration and a fascination for astronomers. Dust can obscure observations of distant stars. Yet dust is the stuff from which all solid bodies are formed. This is why dust research, as bland as it sounds, is one of the most important topics in astronomy and astrobiology.

Read more (http://www.spitzer.caltech.edu/Media/happenings/20060714/index.shtml)

Position(2000): RA 05h 35m 28.30s Dec -69 16' 11.10

Jerry
2006-Jul-15, 02:46 PM
Do we see rings because there is only enough optical density in the dust cloud to lighten the limb, or is the distribution of the dust radially skewed?

The Bad Astronomer
2006-Jul-15, 06:18 PM
The rings are gas. They were there before the star blew up, and the ultraviolet flash lit them up in the same way a current lights up a fluorescent bulb. I have lots of info about this in a series of essays on my Bitesize pages (http://www.badastronomy.com/bitesize/sn87a_threering.html).

peaceharris
2006-Jul-16, 01:12 AM
They were there before the star blew up, and the ultraviolet flash lit them up in the same way a current lights up a fluorescent bulb.

If ultraviolet light from the SN had caused the equatorial ring to light up, then when the UV source stopped emitting UV photons, we would expect the ring to fade off.

Using your analogy of a light bulb, when we remove the power supply, the fluorescent bulb would stop emitting light.

When we saw the light echo of sn1987a, the light echo was present at a particular position only for a few days. Then the light from the SN would lighten up dust at a further distance as the images below show.
http://www.aao.gov.au/images/image/echo_composite_sm.jpg

But the equatorial, and the other 2 outer rings were seen even many years after the UV source had faded.

Secondly, no one who studied the spatial distribution of UV light from sn1987a based on the ELBL files available at the IUE archive has ever proven that the central star emitted the initial UV flash. It is wrong to claim that

1) The UV flash lit the ring up.
2) the rings were there before the star blew up.
3) The UV flash originated from the central star.

peaceharris
2006-Jul-17, 05:33 AM
To understand why astronomers believe that the initial UV flash came from the central star, you need to read: AA Vol 177 L25 (http://adsabs.harvard.edu/cgi-bin/nph-bib_query?1987A%26A...177L..25P)

Here are some quotes from that paper:

"At epochs later than 1st March 1987, the SN spectrum does not vary appreciably any more."

"Moreover, a close inspection of the line by line spectrum of the 3rd of March reveals that it is in fact contributed by 2 different stars."

"One may identify the northern component with star 2, and the southern component with either star 3 or star 1 or both."

They went on to argue that the southern component is most likely due star 3.

The reason why I think this approach is wrong is because the usual procedure adopted in SNe detection is the presence of an abnormally bright source and not the absence of light.

In ApJ Vol 323 L35 (http://adsabs.harvard.edu/cgi-bin/nph-bib_query?1987ApJ...323L..35S), the authors showed that for SWP data taken on March 4 1987, there is excellent fit of 2 PSFs (Point Source Functions) superposed, hence concluded, that it is unlikely for a 3rd point source to contribute.

They also showed that for SWP data taken on Feb 27, 1987, there was unsatisfactory results when 2 PSFs were superposed, but obtained an excellent fit when 3 PSFs were superposed.

They concluded that the 3rd point source was in between stars 2 and 3, in excellent agreement with the position of Star 1.

The PSFs they used were skewed gaussians. For a skewed Gaussian, there are 4 variables: The Gaussian width, the skewness, the peak intensity and position. The more variables we have, the more accurate our 'fit' would be. The skew parameter even changes for different data sets, and is a function of wavelength. Refer Fig 10 of A&A Vol 144 page 335 (http://adsabs.harvard.edu/cgi-bin/nph-bib_query?1985A%26A...144..335C). The authors of that paper compared the skew parameter for SWP 18067 and SWP 18881, which were 2 single point sources.

So just because the superposition of 3 PSFs give better fitting than 2 PSFs we cannot conclude that there is indeed 3 sources contributing. The better fit may be due to the fact that there were 3 sources, or may be due to the fact that we have more variables to play with.

Here's an analogy: If we have 2 data points (x1,y1) and (x2,y2), we can easily fit a linear equation ax + b. But if we have 3 data points, a linear equation may not work, but a quadratic equation, ax^2 + bx + c will. In a quadratic equation, we have 3 parameters to play with: a, b and c.

snarkophilus
2006-Jul-17, 08:57 AM
If ultraviolet light from the SN had caused the equatorial ring to light up, then when the UV source stopped emitting UV photons, we would expect the ring to fade off.

Using your analogy of a light bulb, when we remove the power supply, the fluorescent bulb would stop emitting light.

Well, it doesn't quite work like that. Think of the original UV emission as charging a battery, rather than turning on a switch. Even after the power to the battery is cut off, a light attached to the battery can still run. That's what's happening.

I don't really know the composition of the rings, but if you really want to calculate how long it will take for them to "turn off," you can look up the composition and probably find decent rate coefficients for all of the electronic relaxation transitions, and for major species, vibrational/rotational transitions (if there is molecular matter in the ring). I think Turner et al 1977 has H2 Einstein A coefficients, which describe the rate at which photons are spontaneously emitted from a vibrationally/rotationally excited molecule.

If you want to really do this right, you need to incorporate the interactions within the cloud and model the full evolution of the thing. For simple collisions, Peter Martin (and CITA (http://www.cita.utoronto.ca/index.php/research_cita/interstellar_medium) in general) has some pertinent info (http://www.cita.utoronto.ca/~pgmartin/h2.html).

One thing you'll notice is that some of the transitions are pretty slow. Atoms/molecules in those states will take quite some time to radiate their energy away. Also, with any significant amount of radiation, a lot of the energy given off will be re-absorbed by the cloud, further extending the time it is lit up.

I want to suggest a paper by Dove and (I think) Raynor, detailing a master equation study of interstellar H2, but I can't remember if that's the one I'm thinking about specifically. It should be from the early to middle 1980s (maybe even late 1970s). Anyway, you're looking for a solution to the master equation describing shocked interstellar clouds.

The full calculation isn't trivial, but it's not really that hard, either. I can probably even procure some 25 year old Fortran code for you if you're really interested in checking it out. :) Or you can just grab some data from a similar system and blithely apply the conclusions here. Either way, I can guess what you're going to end up with....

Edit: It looks like someone might have done this calculation already (http://www.badastronomy.com/bitesize/sn87a_innerring.html), or something like it, saving you some time. Even if it's not a full theoretical model of the cloud, you can use it as a basis, and compare the theoretical calculations for the triplicate SN event, those for the single event, and the actual observations. The BA's paper on this (http://adsbit.harvard.edu/cgi-bin/nph-iarticle_query?1995ApJ...439..730P) is neat, too. Page 741 has a pertinent plot, and page 744 shows how fast the ring is fading.

peaceharris
2006-Jul-19, 05:08 AM
Well, it doesn't quite work like that. Think of the original UV emission as charging a battery, rather than turning on a switch. Even after the power to the battery is cut off, a light attached to the battery can still run. That's what's happening.

I don't think UV photoionization can explain why the equatorial ring is luminous. One of the gases present in the ring is hydrogen. Hydrogen emission lines have been detected from the ring for many years, but the initial UV flash lasted for only a few days.

If the density of gas is low, Einstein's A coeffecients can be used to determine how long the gas would remain in its excited state. According to this lecture (http://www.astro.psu.edu/users/rbc/a480/lec22n.pdf), the A values for hydrogen are A ~ 10^6 sec^-1, implying that those atoms can remain excited for only about 1 micro second.

If the density is very high, we may have to also consider collisional excitation. Maybe neighboring particles are colliding and these are causing the emission spectrum. Imagine a solid piece of iron taken from a furnace, glowing red hot, being put out in space. It will just take a few hours or minutes for that piece of iron to cool down. Maybe if that piece of iron is very large, it would retain its heat for a few days.

Phil Plait wrote (http://www.badastronomy.com/bitesize/sn87a_innerring.html): "We were able to figure out the density of the ring from the rate at which it faded; dense gas fades faster, low density gas fades slower. We found the gas to have a density of about 10,000 atoms per cubic centimeter. Sound like a lot? On Earth, our atmosphere has about 3 quadrillion times that density! So the ring is really a pretty good vacuum compared to gas like we have here on Earth!"

I don't understand why Phil says that low density gas fades slower. Based on my reasoning stated above, I think dense gases can retain their heat longer. Phil, Can you please explain why you say that.

Blob
2006-Jul-24, 09:11 PM
Despite its importance, scientists still know very little about star dust. How much dust does a star produce throughout its lifetime? How much survives a star's death? And how do rings of dust coalesce to form stars and planets?

1987A's newly detected stardust, found using an infrared telescope at the Gemini South Observatory in Chile, could help astronomers answer these questions. The dust particles are intermixed with superheated, X-ray emitting gas and found within an equatorial ring around SN 1987A. About a light-year across, the ring of gas and dust is expanding very slowly.

This suggests that the ring was created about 600,000 years before the star exploded, the researchers say. It is therefore unlikely that the ring was created by a supernova blast during the star's death, but rather by stellar winds when the star was still alive.

Read more (http://www.msnbc.msn.com/id/14013695/)