Below you will find the work I did for Dr. Heather Morrison looking for K giants in our halo.
contact me!
Sections are organized by month and week:
August:
8/03-8/11
July:
7/03-7/07
7/10-7/14
7/17-7/21
7/24-7/28
7/31-8/04
June:
6/05-6/09
6/12-6/16
6/19-6/23
6/26-6/30
filelist
list of programs
list of important SM macros
What I am Doing:
Color Magnitude Diagrams:
One of the first things I did was take reduced ugriz' data from Clem (Clem 2005) and attempt to copy his Color-Magnitude Diagrams (CMDs). This also served as my introduction to Super Mongo (SM). I was quickly successful, and moved on to overlay points we had proper motions for (from Cudworth 1979). I have also used several cuts based on CMDs, which eliminated all stars outside of the main sequence, giant branch, or blue horizontal branch from my plots. An example of where I made my giant branch cuts for M13 can be seen here. A CMD also came in handy when we needed to see exactly what kinds of stars were on the Sloan photometric plates we were using.
Transforms:
In order to search the volume of stars we need to get a reasonable number of K giants in the halo, we must use Sloan Digital
Sky Survey (SDSS) data. However, the SDSS uses a different set of filters from the standard UBVRI, and we are attempting to determine equations to transform the Sloan data (which uses a filter system named ugriz) to other filter systems. To that end I have been matching photometric
data from both systems together for globular clusters. We are finding this to be harder than anticipated, as difficulties have arisen in the
non-linearity of the transforms and possible reduction problems in the data itself.
These transforms are made even more difficult by the variety of photometric systems involved. There are multiple ugriz systems we
must deal with, and these systems are relatively new and we are still learning about them. We are also using data from globular clusters,
which means that we have crowded fields. Most (if not all) of the work on ugriz transforms has been done for field stars,
which we are also trying to look at to see if they match the our cluster transforms.
ugriz' - ugriz Transforms:
We attempted to make this transformation because we have much better photometric data in the ugriz' system than from Sloan's ugriz. To use our pretty ugriz' data we need equations to convert it, which is where the transforms come in. We used field stars to find the transforms, with the theory that to make them useable in clusters we would simply need to add a metallicity term. We were lucky enough to get ugriz' data from two telescopes: The Photometric Telescope(PT) and the United States Naval Observatory (USNO) Telescope.
*Note: We also got a set of Sloan ugriz data for M13 from Clem(key), but since we used field stars to make our transformed it was not as heavily used. However, it (and its M92 cousin[key]) were used in a lot of my early transform plots, mostly the ones with annuli. This led to the discovery of an apparent correlation between the transform and the distance from the cluster in g'-g for M13, as seen in this graph. Black points are the closest to the center, followed by red, then green, and finally blue. We're not sure why this is happening, and it only happens with g'-g for M13, but it does not appear to be because of the cluster.
Zeljko Ivezic was nice enough to send us a lot of data for a single stripe (stripe 82) of the SDSS (Ivezic SDSS Paper 658), stars that had been observed in ugriz by Sloan's 2.5 meter and also in ugriz' by the PT. Unfortunately we found out that the PT data was not in ugriz' as we had initially thought, but rather had been transformed into ugriz by Douglas Tucker's equations (http://www.sdss.org/dr4/algorithms/jeg_photometric_eq_dr1.html#usno2SDSS). I wrote a program which iteratively back-transformed the PT data into ugriz' to make it useable for making our transforms. Generally a cubic fit
seemed to work the best, but in some cases a quadratic or linear fit sufficed. After toying with different fits for a while, I
finally settled on the ones I thought were the best. They can be found here. I chose a linear fit for
u'-u, because not only does it do a good job as a fit, but the other fits curve too much on both ends of the plot. g'-g is a
quadratic fit. That one wasnt too hard to come up with, the quadratic fits beautifully. r'-r was a little more difficult,
because of the "bounce" around r'-i' = .5. I tried a quartic, but in the end it wasn't straight enough on either end. The cubic does a decent job, even though it smoothes out the bump a little. I also went with a cubic for i'-i. The cubic, while clearly not perfect,
does a better job than the alternatives (quadratic, quartic, etc). z'-z is just a mess. I chose the linear fit because the others
curve far too rapidly for my tastes. However, the cubic does fit the points fairly well, and is worth consideration too. I wish i
had some more blue and/or red stars for this plot, as the extra information would greatly aid in determining the appropriate
transform.
Tucker was nice enough to give us two sets of ugriz' data on standard stars that had been observed on the Naval Observatory
telescope. We used these stars to check how good our transforms were. These stars were matched against Zeljko's ugriz data in addition to Cas stars I had obtained using a MySQL
query. First, Zeljko's stars were combined with the Cas stars using this program, then matched to
both sets of Tucker's ugriz' data (combined into a single file) using another
program. Unfortunately, as can be seen in these three plots, all of
the matching and error cuts (combined with the relatively small amout of USNO data) left us with precious few points. Because of the small quantity of data, it is difficult bordering on impossible (in my opinion) to make any statements about how well our transforms agree with this data.
After we obtained our ugriz'-ugriz transforms for field stars, we need to see how well they fit stars
with different metallicities. Both M13 and M92 are more metal poor than the field stars, so they were good choices. We matched
Clem's with new data that Jen sent us for both M13 and M92. I used this matched data to make transform plots with ours and Tucker's field star fits overlaid,
making plots of g'-r' Vs g'-g and r'-i' Vs r'-r. From these plots it is clear that there are
metallicity effects for the transforms, but we need redder stars to get a better grip on exactly what the effects are.
ugriz - UBVRI Transforms:
Once we obtained our ugriz' - ugriz transforms, we needed to see how good they were. We decided to compare them to Peter Stetson's
UBVRI photometry (http://www.hia-iha.nrc-cnrc.gc.ca/staff/stetson_e.html) by first transforming Clem's ugriz' photometry to ugriz,
and then transforming that into UBVRI via Zeljko's (Ivezic SDSS Paper 658) and Robert Lupton's (Lupton 2005) published
ugriz - UBVRI transforms. As a control, we also saw how closely the Clem data matched Stetson's data if we used Tucker's ugriz' to
ugriz transforms instead of ours. To do this I made four FORTRAN programs to transform the data: us +
Lupton, us + Zeljko, Tucker + Lupton, and Tucker + Zeljko. These programs output ugriz and UBVRI to two different files, from the original Clem file for M13(key) or M92(key). Then another program matches the transformed Clem UBVRI file to Stetson's photometry. Some of the resulting plots (the ones
for M92) are here: us and Lupton, us and Zeljko, Tucker and Lupton,
and Tucker and Zeljko. If you compare the results of our transform and
Lupton's to Tucker and Lupton's you can see how similar they are. The transforms also look resonably good, with maybe only a slight
offset. Problems arise when using Zeljko's ugriz to UBVRI transform, however. His transforms are not only not constrained for the
bluer stars, but since his equations are cubics they get less accurate much faster than Lupton's linear equations do. The result is
that the BHB stars are very poorly fit, as can be seen in the above plots. In my opinion the best way to get from ugriz' to UBVRI is to use our transforms with Lupton's, although it seems that Tucker's transforms are about as good.
Why we need these transforms:
We are ultimately looking for K giant stars, and we find important information about them (metallicity, distance, temperature, etc) through their spectra. We have very good ugriz' cluster photometry (Clem) and a large ugriz spectra dataset (SDSS). We need a ugriz'-ugriz transformation to reconcile the two.
UBVRI photometry is a widely used photometric system. To provide a way of using the massive Sloan dataset in conjunction with UBVRI would be very useful to a lot of astronomers.
Spectra:
Stars in the SDSS are observed through fiber optic cables mounted on large plates in the known location of the star to be observed. The stars on these plates need to be matched with the stars observed by Clem and Jen to combine their ugriz' and ugriz photometry with SDSS spectroscopy. Each plate only generates a few dozen matches, but hopefully by getting several plates together we can accumulate enough points for a good analysis. The only cluster I was able to get plates for was M13, and there were three plates that I used: 2174, 2185, and 2255. plate 2255 was a double plate; we used a plugmap for the regular stars, and a special plugmap for extra bright stars. The plugmaps were matched with the plates via this FORTRAN routine, and then combined into a single file. That file, along with the other two plate files, were put into another program, which matched them with Jen(key) and Clem(key) files. The output of that program was used as the input of another, which finds all the points with Kyle proper motions.
Once we determined which plate stars we could match to our photometry, I got to start looking at the actual spectra. I wrote an IDL program to plot each spectrum on splot and write it out in a text file. the program outputs good spectra into one file(key) and bad spectra into another (with an optional comment). I then wrote another program to make an indices.in file, which is read in as the input to yet another program, which finds the strength of several important lines. Using the output of the last program, I was able to make some plots of line strength vs color.
