Working with L1688

From CoolWiki
Revision as of 18:35, 5 May 2010 by Rebull (talk | contribs)
Jump to navigationJump to search

There are a LOT of things on this Wiki, very few of which are arranged linearly. The Wiki is a web, after all, so there are links interwoven all over the place. However, we do realize that this can be frustrating for new users, so here is a single analysis chain, start to finish, with the products already done so that you can step into and out of the process as you want or need.

This page uses as an example the Lynds 1688 cloud. L1688 is a star-forming region, and our goal is to find old and new YSOs. So, this same chain of analysis will work for any other Galactic star-forming region. To analyze extragalactic data or for using Spitzer data for purposes other than looking for YSOs, while several of these steps are universal (downloading data, literature searching, creating mosaics), the rest are customized to this goal of finding old and new YSOs.

NOTE THAT this page is not meant to be a cookbook to follow blindly (without thinking) for Lynds 1688 or any other cloud! It is instead intended to provide you the tools to use to find the answers you seek for any step of the analysis for any cloud.

Download the tarball with all my reduced images and data tables for Lynds 1688.

Downloading the data

Go here (well, really specifically here) and follow the tutorial to investigate what observations exist and download the data. Search on L1688, or another object of your choice. Find the AORs, look at the coverage, decide if you need to download the IRAC and MIPS mosaics from more than one AOR or just one AOR. The data I'm using for this example on L1688 are all from pid 177, but you should investigate the coverage of the other programs too.

It turns out, though, that L1688 is part of the c2d Legacy program (pid 177 is part of that), so you may also wish to consult the page on already-reduced Spitzer data.


In the generic case (like for the Lynds cloud project from round 3), if there is only had one AOR per instrument, per target, you can download the mosaics directly (no need to make the mosaics in the next step below, since it's already done).

Important note: In most cases, for most famous objects (objects with names), the mosaics as produced by the pipeline are JUST FINE, and there is NO NEED to make your own mosaics in the next step. Starting from the pipeline mosaics is JUST FINE, and is what many professional astronomers do.

Making the mosaics

IF YOU DECIDE YOU NEED TO MAKE A MOSAIC, go here and follow the tutorial to make the mosaics from the post-BCD pieces you have downloaded.

RESULTS: Mosaics from all 4 IRAC channels and 2 of the 3 MIPS channels (for L1688) are included in the tarball above if you wish to short-cut this process.


In the generic case, if there is one AOR per instrument, per target, you can download the mosaics directly (no need to make the mosaics here, since it's already done).

Important note: In most cases, for most famous objects (objects with names), the mosaics as produced by the pipeline are JUST FINE, and there is NO NEED to make your own mosaics. Starting from the pipeline mosaics is JUST FINE, and is what many professional astronomers do.

Getting data from other wavelengths

Go here and follow the tutorial to find data (images and data tables) at other wavelengths. Think about the questions at the end of that page as they pertain to these data.

RESULTS: A POSS image is included in the tarball above if you wish to short-cut this process. Reduced photometry from 2MASS is included in the data tables as well.

Investigating the mosaics

Go here and follow the tutorial (or use your own software) to explore the Spitzer mosaics as well as the mosaics at other wavelengths. Think about the questions at the end of that page as they pertain to these data.

Additional questions:

  1. Look for image artifacts in the mosaics - they are certainly still there. What are real features of the sky and what is a feature of the instrument and/or reduction? IRAC artifact examples and MIPS artifact examples are online, both as part of the corresponding instrument handbook.
  2. If you have images that someone processed for you, how do they compare to the online mosaics produced by the pipelines? What artifacts are better (or worse)?
  3. Which objects are saturated, in which bands?
  4. How big are any of the features in the nebulosity? (What do I mean by big?) in pixels, arcseconds, parsecs, and/or light years? (Hint: you need to know how far away this thing is, and it's about 125 pc away. If it helps, there are 3.26 light years in a parsec. If you're working with another object, you should look in the literature for distance estimates to it.)
  5. In this case, asteroids matter. The ecliptic latitude of L1688 is just -2.7 deg. (You can use Spot/Leopard or the SHA to convert coordinates for any other target you may be working with.) The IRAC and MIPS maps were taken in two pieces. I gave you just one IRAC map, the combination of all frames, but I gave you 2 MIPS-24 frames, one from each epoch (separated by 3-5 hrs). Can you find the asteroids? (personally, I'd use the blink feature in ds9, but you can use whatever you want.)
  6. For the 70 um data, the SSC pipeline produces two sets of BCDs: filtered and unfiltered. The filtered data attempts to automatically remove some of the instrumental artifacts, but does not conserve flux for extended objects or for very bright point sources. I gave you a copy of both the filtered and unfiltered mosaics. Can you find the sidelobes around bright objects in the filtered image that are an artifact of the filtering process?

Previously identified sources

How many of these sources are 'famous' and how many are new that you have just discovered?

Go here and follow the tutorial to explore the literature. ( Move on to these examples and questions when you're ready.) More things to do: Find some old and some very recent papers on L1688. For how many decades have people been studying this region? Find some of the famous objects in your images. Note that this is a pretty famous region, so it's doubtful that you will easily be able to construct a list of ALL the famous objects in this region in a short time, even using SIMBAD. (If you're working on a different region, you may find many fewer papers on your region. Your mileage may vary.) Keep in mind too that people work in different coordinate systems. A paper reporting on radio data may have a slightly different coordinate system than Spitzer (which is tied to 2MASS). A paper from anything earlier than the 90s might be working in 1950 coordinates rather than 2000 coordinates, and may or may not provide high enough precision coordinates for you to easily find the corresponding object in your Spitzer images; you may need to find some archival data (POSS or 2MASS) to 'rediscover' the old object. You will have to use your judgment to decide if any given object is really the same as the object in the Spitzer image, or if the Spitzer image didn't really detect the object, or can't resolve the object (e.g., can't distinguish it from a nearby object).

Doing photometry

Read about units before you get started.... Then read about photometry in general, and finally read the advanced intro to photometry.

The decision you now face is the following. You have a mosaic with 100s to 1000s of sources, over at least 4 bands (and recall if you have HDR IRAC data, there are actually [at least] 2 frames per band, a long and a short one), so that's 8 IRAC frames, and probably you want to at least do MIPS-24 as well, for a total of 9 frames. You need to decide if you are interested in photometry of just a few objects, in which case you might consider doing all of the photometry by hand, one object at a time, one band at a time. On the other hand, you might be interested in having the computer automatically detect and measure the photometry of all the objects in the frame. If this is the case, you need to tune the parameters of your photometry extraction software to the specific application you have here -- the specific band, source density, background, etc., etc. No matter what you do, in order to check the accuracy of your photometry, you will probably end up doing this more than once on the same data set.

OPTION 1: Using MOPEX to do many sources at once (not particularly recommended!): Go here and get the (admittedly still cryptic) doc file that is linked in partway down. Find the part on using Apex-1Frame. Follow that for each channel (twice for HDR data!) to get the photometry.

OPTION 2A: Using APT to do one source at a time: Go here and work through the (somewhat skeletal) page. If you take this option, I recommend you start with a short list of famous objects from the above step. If you have suggestions for specific improvements for this tutorial, please let me know, or just implement them!

OPTION 2B: Using APT to do a list of sources at once: Only AFTER you have thoroughly explored the parameter options in APT and thoughtfully considered what parameters to use, you can now send APT a list of targets, so you're no longer limited to doing one object at a time using APT. I STRONGLY recommend that you try several by hand until you understand what is going on, and optimize your parameters correspondingly for the mosaic and channel and background levels you have in your specific mosaic.

OPTION 3: Using your own preferred software to do the photometry. Be careful; read the stuff on Units first.

IN ANY CASE: Never just trust that the computer has done it right. It probably did what you asked it to do correctly, but you asked it to do the wrong thing. Always make some plots to test and see if the photometry seems correct.

RESULTS: Reduced (and bandmerged) photometry is included in the tarball above if you wish to short-cut this process, and I promise I won't think any less of you.

For the data I gave you, I actually cropped the image and the catalog out of a MUCH MUCH MUCH larger mosaic and catalog. The cropping of the image had to be done in pixel space, so north is not necessarily up, and the cropping is not necessarily strictly in RA/Dec. The cropping in the catalog is strictly RA/Dec. This means that there will be wedges of images without sources. Sorry! If you really want those data, let me know, or go get the data yourself from the c2d delivery on the Legacy pages.

Questions:

  1. If you are doing automatic source detection, compare the source extractions from any one band with its corresponding mosaic. Is it getting fooled by detector artifacts?
  2. If you are doing automatic source detection, overlay the source extractions from, say, IRAC band 3 with the image from IRAC band 1, or the source extractions from MIPS-24 with image from IRAC band 1. Are there a lot of stars (or other objects) in common? How does the nebulosity compare?

Bandmerging the photometry

Now that you have a measurement of the brightness of each object in each frame, you need to match the same object across multiple bands, called "bandmerging." Maybe, if you used APT, you kept track of that manually. If you did the photometry automatically, you need to have the computer match the sources. If you did it automatically, some of the sources you extracted might not even be real - this is a common problem, and you will have to tune the source extraction... or get rid of the false sources in the bandmerging step, since the false sources will generally not have counterparts at other bands. Of course, some of the most interesting real sources also can be single-band detections...

There is no GUI tool to bandmerge the photometry.

The code I use is written in IDL (expensive to have a site license, even for academics), and for each band, starting with 2MASS JHK, looks for matches within 1 arcsecond in IRAC-1, then IRAC-2, then IRAC-3, then IRAC-4, and then within 2 arcseconds in MIPS-1, and then within 10 arcseconds in MIPS-2. (Bonus questions: Why should I start with the 2MASS data and work up in wavelength? Why should this matching distance change?)

RESULTS: Bandmerged photometry (including 2MASS data!) is included in the tarball above if you wish to short-cut this process, and I promise I won't think any less of you.

Questions:

  1. Make sure that I've merged the right sources across several bands by spotchecking a few of them. (Find an object that the catalog says is detected in at least 3 bands and then overlay the images in a 3-color image or Spot to see if there is really a source there, at exactly that spot, in all bands, or if it's a cluster of objects, or different objects getting bright at different bands.

Working with the data tables

I've provided the data tables in something called "IPAC table format" in the tarball above. There is a long version with many columns, and a short version with many fewer columns. Spot, Leopard, and IPAC's Skyview all understand IPAC table format. You can also import the file into Excel to play with it there. NOTE THAT there are a LOT of sources in this relatively small mosaic. You will want the computer to do as much work for you as possible - you don't want to be counting sources manually.

You can use Spot, Leopard, or IPAC's Skyview to overplot the sources onto an image of your choice. NOTE AGAIN THAT there are a LOT of sources in this relatively small mosaic. Skyview (and maybe Spot/Leopard) don't understand the full table, with all those columns; that's why I provided the "catshort" version with just positions.

Import the data table into Excel, as that's probably the easiest way for you to work with it. (DO I NEED TO WRITE A TUTORIAL ON THIS?) There should even be a (contrived) way to get Excel to cough back up a file that is close enough to IPAC table format that you can manually edit it to be compatible - you might wish to do this for example, if you want to take just the IRAC-4 detections and overlay them on an IRAC-1 image. (see questions in 'photometry' above.)

A few words on working with IPAC's skyview

(skyview is also free and available for a variety of platforms. There is a manual too.) Get into Skyview and "paint" ("pa") the image into a window. Use the left mouse button to pull up a "zoom window" in the upper left. Click the right button to zoom in and the left button to zoom out. Click the middle button to reposition the view. Read in the table and mark every source with a red circle of diameter 6 px. Use the "gpi" ("graphics pick") feature to click on individual sources and have the parameters from the tbl file printed in the terminal window. (Left button to select source, right button to quit out of it.) Use "er" to erase window.

      unix% skyview
      skyview version  02/28/00  (R3.4)
        Copyright (C) 1992, California Institute of Technology.
        U.S. Government Sponsorship under NASA Contract NAS7-918 is acknowledged.
      > pa mosaic3.6long.fits
       reading file . . done
       scanning for max min . . done
       building histogram . . done
       redoing histogram . . done
       stretching from -2.190281e+00 to 2.686462e+00
      > zoom_pan
     Centered on RA = 5h07m35.69s  Dec = -6d20m44.4s
      > table extract3.6long.tbl mark $RA $Dec red circle 6
      > gpi
      mark came from line 145 of file [...]
      > er

Questions and things to try

The final catalog should include only real detections, dropping false detections that are due to cosmic rays or other instrumental effects. (If you are constructing your own catalog, you will probably want to drop objects seen in just one band.)

  1. How many stars are detected in each band? Is this about what you expected based on your answer to the questions in the mosaic section above? HINT: you can do this using Excel, you don't need to count these manually!! Ask if you need a further hint on exactly how to do this.
  2. Which stars in the catalog are the stars identified in the literature?
  3. Read about Spitzer units and then make sure that you understand the difference between flux and magnitudes. Check my unit conversion - or do the unit conversion yourself if you are working on your own. Make sure you understand how I got the magnitudes from the fluxes (or the fluxes from the magnitudes). You will need magnitudes for the next step, and fluxes for the step after that.

Making color-color and color-magnitude plots

Read about color-magnitude and color-color plots in general here first. Get Excel to cough up the magnitudes and learn how to make it generate a plot. (DO I NEED TO HAVE SOMEONE WRITE A TUTORIAL ON THIS?) The best way to pick which colors to plot, to be brutally honest, is to do a literature search, look at what other people did and how they interpreted it, and then make the same plot for your own data. Try some of the plots from the intro presentation, or papers mentioned in the IR section of Finding cluster members, but you should feel encouraged to branch out and try other papers that you find in the literature, since there are LOTS of color-color combinations, and they all tell you different things. Color-color plot ideas (Don't peek until you're stumped!)

  1. Pick at least one color-color or color-magnitude plot to make. Figure out a way to make it ignore the -9 (no data) flags. Where are the plain stars? Where are the IR excess objects?
  2. Where are the famous objects in the plot? Where are the new YSO candidates?
  3. Make a new column in your Excel spreadsheet with some colors. Is there a way you can get Excel to tell you automatically which objects have an IR excess?
  4. Bonus but very important question: How do you know that some of these sources aren't galaxies? Can you find something that is obviously a galaxy on the images? Can you think of a way using public data that already exist to check on the "galaxy-ness" of some of these objects? (Read about Finding cluster members.)

Making SEDs

Read about SEDs in general here first. Get Russ Laher's spreadsheet from me, or write your own. Make sure you understand the difference between plotting up a flux density as a function of wavelength versus an energy density as a function of wavelength. (You may need to read about Spitzer units first.) Note that you probably want to make it a plot of log energy vs. log wavelength, not linear values.

  1. Pick some objects to plot up, maybe some of the famous ones from above would be a good place to start, or the ones you flagged above as having an IR excess.
  2. What do the IR excesses look like in your plots? Do they look like the SEDs on this page or in the intro presentation?
  3. Which ones of these sources have optical (or any other) data as well, from your web search above, or from the literature? Add these to your bandmerged catalog (not sure if Russ' spreadsheet can handle this...)
  4. If using Russ' spreadsheet, fit some SEDs with some blackbodies. How many blackbodies does it take to make a good fit to some of the sources with the most detections (like all 3 2MASS, all 4 IRAC, and MIPS-24)?

Analyzing SEDs

Add a new column in Excel to calculate the slope between 2 and 8 microns in the log (lambda*F(lambda)) vs log (lambda) parameter space. This task only makes sense for those objects with both K band and IRAC-4 detections. (For very advanced folks: fit the slope to all available points between K and IRAC-4 or MIPS-24. How does this change the classifications?)

  • if the slope > 0.3 then the class = I
  • if the slope < 0.3 and the slope > -0.3 then the class = 'flat'
  • if the slope < -0.3 and the slope > -1.6 then class = II
  • if the slope < -1.6 then class = III

These classifications come from Wilking et al. (2001, APJ, 551, 357); yes, they are the real definitions ( read more about the classes here)!

  1. How many class I, flat, II and III objects do we have in this set of extractions? Read about finding cluster members and then decide how many of the objects in this set of extractions are really cluster members, which aren't, and which ones are ambiguous.
  2. Does the ratio of Class I, flat, II, and III objects change when considering just the objects you think are members? Assuming that the classes are really representative of ages, do you see how cluster membership can affect the age you estimate for a cluster?
  3. Where are the objects with infrared excesses located on the images? Are all the Class Is in similar sorts of locations, but different from the Class IIIs?


For very advanced folks: suite of online models from D'Alessio et al. and suite of online models from Robitaille et al.. Compare these to the SEDs we have observed.

Writing it up!

Write up your results. Be sure to include:

  1. How the data were taken.
  2. How the data were reduced.
  3. What the Spitzer properties are of the famous objects, including how the Spitzer observations confirm/refute/resolve/fit in context with other observations from the literature.
  4. What the Spitzer properties are of other sources here, including objects you think are new YSOs (or objects you think are not), and why you think that.
  5. How this region compares to other regions observed with Spitzer.

Take inspiration for other things to include from other Spitzer papers on star-forming regions in the literature.