Nick Stantzos's REU homepage (Summer 2011)

Nick Stantzos
Northern Arizona University
{nws6 at nau .dot. edu}

UW Madison REU program
Univ. of Wisconsin - Madison
Madison, WI 53706

2011 REU program
Data Reduction
Observations/Data sources

Louie the Lumberjack

Characterizing Arecibo and GBT observations of the Outer Scutum-Centaurus Spiral Arm


Galactic astronomy has been a promising field for the better part of the last century, providing us with a wealth of information including results about the formation, evolution, structure, and function of many galaxies in the universe. Astronomers observe outside galaxies and are able to measure physical properties of these other galaxies, such as brightness, mass, velocities, and temperature to name a few. Given this plethora of information on outside galaxies, one would conclude that astronomers have revealed all that there is to know about our own Milky Way galaxy. Well, when it comes to studying our own galaxies, there are many complications.

Though we are able to see internal features of our own galaxy that we would never be able to resolve in other galaxies, there are formidable challenges in identifying properties of our own galaxy. Being located in the galactic midplane, Earth is surrounded by gas and dust in every direction when looking within our own galaxy. This opaqueness (extinction) proves to be cumbersome when trying to make measurements at large galactic distances. To measure distances, astronomers often use doppler velocities of observed gas, along with an assumed rotation curve. Figure 1 is a radial velocity map of the Milky Way with a distance grid overlaid; it shows observed doppler shifts of HI as a function of its position.

velocity contours
Fig. 1. This "artist's conception" image of the Milky Way Galaxy by Robert Hurt (Spitzer Science Center/IPAC/JPL) shows most of the claimed structures in the Galaxy, where the Sun, located at the center of the polar grid, is assumed to be Ro=8 kpc from Galactic center. Superimposed is the line-of-sight velocity that would be measured from gas at a given position, assuming the rotation curve obtained by McClure-Griffiths and Dickey (2007, ApJ 671, 427). The Galaxy is divided into four quadrants. Galactic longitude l=0 to 90 deg (where longitude runs counter-clockwise) is the 1st quadrant, l=90-180 deg is the 2nd quadrant, etc. Click to enlarge.

The best way to approach this map is to think of a topographical map. Like a topographical map, each countour here corresponds to a different velocity value, with red contours representing red-shifted radial velocities, and blue contours representing blue-shifted radial velocities. Negative velocities are assigned to objects that appear to be moving towards the sun. Similar to a topographical map, the spacing between contours represents the steepness of the change in velocity. Large spacings between velocity contours corresponds to a gradual shift in radial velocity in a certain direction.

This map is useful in assigning distances to structures within the Milky Way, however the assumption of a rotation curve is necessary to make these distance predictions. One evident consequence of this graph in the varying relation between radial velocity and distance. Velocity is not a linear map to distance, it changes with galactic longitudes. As an example, along a longitude of 90 degrees the velocity contours seem to maintain a consistent increase in spacing as distance increases. For a longitude of 30 degrees, this is clearly not the case. Tracking from the blue zero line, the velocity contour spacing does not increase as systematically as it does along the 90 degree line. The spacing appears to behave like a higher order polynomial along the 60 degree line as opposed to an almost linear relationship along the 90 degree line.

One problem with this kinematic distance technique is known as the distance degeneracy problem. For example, if one were to look along a galactic longitude of 330 degrees, there would be two different observed sources both having the same radial velocity. This is a huge problem when trying to use this method (kinematic distances) to make maps of the Milky Way interior to the solar circle, or the circle created by the sun's orbit around the galactic center. Due to this distance degeneracy problem with double valued velocities, we prefer to make observations exterior to the solar circle. Figure 2 (below) illustrates an example of observations interior to the solar circle, plagued with the distance degeneracy problem.

double velocity
Fig. 2. A line of sight of an example observation. Note that interior to the Sun's orbit (the "solar circle") in the 1st/4th quadrant, a measured positive/negative velocity can occur for an object at two different distances. This "kinematic distance ambiguity" has complicated efforts to map the structure of gas and star formation in the inner Galaxy. Click to enlarge.


Since the birth of galactic structure decades ago, the specifics of the spiral arm structures of the Milky Way have been hotly debated. Initially, galactic astronomy was a popular field, with new discoveries and fresh publiations a regular occurrence. However when confronted with the gas obstruction and extinction problem, galactic astronomy slowly began to fade into the background until the advent of more sensitive equipment.

In recent years, more sensitive equipment has become available, and galactic astronomy was once more able to make headway in mapping the Milky Way. With even more sensitive telescope, such as Arecibo, the Green Bank Telescope, Spitzer, etc., astronomers are now discovering sections of the Milky Way's spiral arms that have long been thought to exist. One such example is the recent discovery of the Outer Scutum Centarus arm by T. Dame and P. Thaddeus (2011, ApJ Letters) in May of 2011. This outer segment has been predicted to exist purely from assumptions regarding the symmetry of the galaxy.

The figure below, Figure 3 is a map of spiral arm segments where we have been able to observe HI/CO emissions. The marked segments are where we have data that directly points to the existence of these arm sections. Most of the arm segments were discovered before 1970, the distant arm in 2003, the far 3-kpc bar segment in 2009, and the Outer Scutum-Centarus arm in June 2011.

Figure 3. HI/CO emission map of spiral arm segments. The marked lines indicate segments that have been observed and are known to exist. The dashed lines correspond to arm segments that are predicted to exist, though no data supports these predictions.

With the serendipitous discovery of the existence of the Outer Scutum-Centarus arm, our project was highly motivated to characterize physical properties of this previously unknown spiral structure. All of the observations of this project were made on 21cm neutral hydrogen (HI) emissions in the galactic midplane region. As depicted in Table 1, the combined area of longitude observed from all three surveys was 0 < L < 90 and latitude was -15 < B < 15.

Back to Top

Data Reduction

Observations/Data sources

In total, this project utilized three different sets of data. Figure 4 (below) shows instruments used, along with longitude, latitude, and velocity ranges of our data.

Figure 4. Instruments, and parameter ranges of our data sets. GBT and LAB are acronyms for Green Bank Telescope and Leiden Argentine Bonn, respectively. The LAB survey was a result of the merging of the Leiden/Dwingeloo survey with the Instituto Argentino de Radioastronomía survey. The boxes list the instruments' resolution, and each color corresponds to the respective longitude range in the diagram.

The data reduction procedure was straightforward, but required a fair amount of IDL programming. In this case, we created multiple routines in IDL for our initial Arecibo data set. We read in the data cube (as a fits file), and then proceeded to run routines to plot the data cube in different orientations. We created spectra of specific latitude and longitude slices, and then proceded to do some fits. By sweeping through the data cube in longitude and velocity while keeping the latitude variable, we were able to effectively characterize the galactic midplane. Figure 5 (below) shows a typical latitude profile, which is useful for finding where the galactic midplane is located.

Fig. 5. This figure is a typical latitude profile of the galactic midplane. We made these latitude profiles for each longitude and velocity for our data cube and applied different fitting techniques to the profiles. The red line is a gaussian fit made for this latitude profile.

After creating several routines to loop through our data cube and plot these latitude profiles, we applied fitting techniques to begin characterizing the midplane. Fitting techniques/parameters such as:

were all used to reduce our cube into interpretable data. Figure 6 (below) is an example of the different fitting techniques/parameters used with our arecibo data cube.

arecibo data
Fig. 6. This image shows five of the different fitting techniques we used on our Arecibo data cube. The lines plotted over the images are galactocentric radial measurements from the galactic center, with units in kpc. Note that each colorbar scale is as marked, with boxes 3 and 4 having a logarithmic scale. Click to enlarge.

A gaussian fit routine was written to manage the enormous quantities of spectra that were present in our data cube. Though this scripted method has its flaws, it was able to further quantify the location of the midplane. In cases where spectra were dominated by noise, the gaussian fitting procedure would ignore these spectra. A quality check was implemented to skip spectra where the mean intensities were less than zero, and only contained noise.

The flaws of this gaussian fitting program are quite evident in Figure 6. The not only did the gauss fitter have trouble with low signal spectra, it also had difficulties with spectra containing multiple peaks. Another source of error was the range of observations made in the data. The latitude range was not broad engough to collect all of the signal emitted from the source. For instance, Figure 7 (below) shows a latitude profile in which the peak of the signal falls out of the range of the observation. Though this is particular troublesome for the gauss fitter, is does provide more evidence for the warping of the galactic midplane.

Fig. 7. This latitude profile is an example of a flaw with the gauss fit program. Since our data sometimes edged out of our latitude survey range, the gauss fit failed at fitting a reasonable gaussian to each spectrum where this edging out occurred.

Once the fitting was completed, we used a routine to mark the specific areas of the Outer Scutum-Centarus arm to better characterize this specific region of the galaxy. The goal was to mark a set range of longitudes, latitudes, and velocities that we could then put back into a gaussian fitting program to fit gaussians to this specific region. We anticipated that this would give rise to more accurate fits which would enable us to precisely define the arm of this study.

Once this process was completed with the Arecibo data, it was then applied to the remaining two data sets. Figure 8 (below) shows the same fits used on the Arecibo data, though applied to the GBT data.

gbt data
Fig. 8. This image shows five of the different fitting techniques we used on our GBT data cube. The lines plotted over the images are galactocentric radial measurements, with units in kpc. Note that each colorbar scale is as marked, with boxes 3 and 4 having a logarithmic scale. Click to enlarge.

Back to Top


After these thorough analysis processes, clear features were noticed in the structure of our spiral segment. Firstly, the galactic midplane exhibits a warping feature, that is it rises above a latitude of 0 degrees. This warping feature is known, though our data displays this in a much higher resolution than most previously analyzed data sets.

In addition to the warping feature, a corrugation effect is also apparent in our results. This behaves much like a sinusoidal function, with repetitive crests and troughs. For a better visualization, see the description and Figure 9 below.

Fig. 9. This figure is a Velocity Longitude plot depicting the intensity-weighted average for Latitude of the galaxy for the specified longitude regions. The colors towards the red end of the color bar indicate that the observed gas (HI) is at higher latitudes. The white regions are areas where either no signal was detected, or too much noise was present to obtain a signal. The lines plotted over the image are galactocentric radial measurements from the galactic center, with units in kpc.

Figure 9 (above) characterizes distinct features of the galactic midplane. The y axis tracks radial velocities of the observed HI gas. Keeping in mind that radial velocities correspond to distance, we can effectively determine the distance to the observed HI regions. The white lines intersecting the figure are galactic radial lines, in units of kpc. With these radial lines, we can easily determine a relationship between the warp of the galactic midplane, and the galactic radius.

First, there is a noticeable "corrugated" feature (60 < L < 70, -80 < v < -25) stretching from the inner radius of the galaxy outward. For an easy way to visualize this, imagine taking a latitude slice perpendicular to the corrugation. The side view would resemble that of a sine function with a polynomial component, though not necessarily with continuous crest of trough features. This particular feature has been known to exist, though it has not been displayed in such high resolution before. For a visualization of this feature, see Figure 10 (below).

In addition to the corrugation feature present in this latitude plot, there is also clear evidence for the warping of the midplane of the galaxy. At a radius of about 13kpc, the latitude measurements begin to show an upward trend as the radius increases. The warping of the midplane has been long studied, though with this high resolution data it becomes easier to quantify the specifics of the warp.

Fig. 10 The sine function with a polynomial component. Approximate side view of a slice taken perpendicular through the corrugation feature. The white sine wave represents the corrugation of the galactic midplane, while the red line represents what a warp would look like on a non-corrugated galactic midplane. Note that the galactic radii and latitudes are not physically accurate of the Milky Way.

Not only did we measure features dealing with the latitude of the galactic midplane, we were also able to quantitatively measure the width of the galactic midplane. Figure 11 below shows a latitude velocity plot of the galactic midplane looking at a galactic longitude of 72.5 degrees.

Fig. 11 A latitude velocity plot of the galactic midplane. The solid red lines represent the edges of the galactic midplane (the distance between the two red lines is a measure of the total width of the midplane) and the dashed red line is an estimation of where the center of the midplane would exist. The vertical lines represent distances of 1 kpc, with the solid lines being 5 kpc. The horizontal lines represent distances of 200 pc with the solid lines being 1 kpc.

This shows a latitude-velocity plot for galactic longitude L=72 degrees. The vertical black lines show distance from the sun, every 1 kpc, with solid lines every 5 kpc. The left-most line is for a distance of 25 kpc. The horizontal black lines show the height above the Galactic midplane, every 200 parsecs, with solid lines every kiloparsec. The grid narrows to the left because more distant objects subtend a smaller angle. The red lines show the calculated HI midplane and thickness. Note that though the Outer Arm and Outer Sct Cen Arm appear to be different sizes, this is not physically accurate. This apparent size difference is achieved because the Outer Sct-Cen arm is further away, and thus subtends a smaller angle.

These latitude-velocity plots were made for every longitude position we surveyed in our data sets. Movie 1 below is a step through of the entire survey's latitude-velocity plots. You will see the galactic midplane for every step in longitude space the movie makes.

Movie 1. Click to play movie of latitude-velocity plots for whole Arecibo longitude range.

A clear pattern in this movie is the banded regions representing cross cuts through different spiral arms. This banded pattern is consistent throughout much of the movie, though an interesting effect is also present. It would appear that the width of the galactic midplane is greater for those areas between the spiral structures. One reason for this behavior could be the absence of star forming regions. With the lack of stars, the gas between the spiral arms could be unconstrained by gravitational forces. Though this is just a rough speculation, with no studies done to support this hypothesis. An interesting future project could be to relate star maps of spiral structure with out observed HI maps of the same structures.

Back to Top


  • We clearly see the HI warp in all three data sets, with the midplane shifting to ~1kpc above b=0, starting at RG=13 kpc.
  • We see coherent patterns of corrugation in the HI midplane.
  • Outer-Scutum-Centarus is also very clear in all data sets, and we are measuring its vertical width vs. longitude.

Further work is required to place more accurate bounds on the width of the midplane, to define the warped region of the midplane, and the corrugated features observed in our segment of the Outer Scutum-Centarus spiral arm.

Back to Top


The following links are useful for looking up info on UNIX, web page making, astrophysical data and journals, and fun entertainment.

SIMBAD (Stellar/Galactic database)

NED (Extragalactic database)

UNIX tutorial

Web page basics

NASA Astrophysics Data Service



Leisure Dive

Universe Today

Back to Top