Scavenger Hunt #1   (the 2nd one...; IDL arrays start at 0)

NGC 5701 is a nearly face-on spiral. Here are some links to images of it. the SDSS Navigator tool image R-band image from EGG archive NED search NGC 5701 has been mapped in HI with the VLA. The image here comes from the paper by Kornreich et al. 2000, Astro J. 120, 139.

larger view

1. Look up the Kornreich et al. paper in the Astronomical Journal. Explain in detail how you found the paper (what sites you visited, any searches you performed). There are several answers to this; let's see what route(s) you each take...

2. Notice that the coordinates in the HI map displayed in the Kornreich et al. paper are in the 1950 epoch reference frame, B1950. For comparison with the A2140 data, we need to precess those coordinates to J2000 (or vice versa). We noticed last week that "The HI from NGC 5701 is detected in beams 1 and 6 at 14h39m11s.", and if fact, Adrienne noted, looking at the actual A2140 dataset: "HI is visible... in the Beam 6 drift from 14h38m32.7s to 14h39m56s."
   Figure out how to precess those J2000 coordinates to B1950, explain how you did this, and then comment on how Adrienne's comment compares to what is seen in the Kornreich et al. map. (But note that we have to deal with the issue of spatial smoothing; see below.)

3. Smoothing in the frequency domain
   

   First, here is a global HI profile, from the Cornell EGG digital archive. It was obtained with the (now deactivated) Green Bank 42m telescope by Haynes et al. 1997. The beam width (full width at half maximum power = FWHM) of the 42m telescope at 1.4 GHz is 21 arcmin.

   42m telescope HI spectrum

   You can download the digital HI spectrum; note the columns indicated at the top of the file. Can you guess the units?
   Try to plot the raw spectrum using some handy program like IDL, SM, etc. You are given the raw (as observed) spectrum, the spectrum after smoothing (usually Hanning; see below), the polynomial baseline fit to the signal free parts of the spectrum, and the final (baseline subtracted) spectrum.
   Try to see what happens if you smooth the raw spectrum using different methods:
   
   • 3-channel Hanning smoothing:   replace each channel (or pixel) with a weighted average of that channel and its neighbors on either side, assigning weight 0.5 to the given channel and 0.25 to the other two. You can also apply more complicated versions Hanning smoothing over 5, 7, etc channels (with appropriate weights). Here, try at least the 3 channel version.
   • Boxcar N smoothing:   replace each channel with the average of N channels around it, where N = 3, 5, etc. Try at least 3 and 5.

   What is the impact of smoothing on the appearance of the profile?
   Do you notice any difference in the appearance among the different smoothings you tried?

4. Smoothing in the spatial domain
   In reducing the A2140 data, we apply smoothing to the drift data in the time domain. Remember that we record data every second and the FWHM of the ALFA beams is about 3.5 arcmin. How long does it take a source to drift through the beam (at half power)?
   The default option in the ALFA_IDL routine "strip_pv" for smoothing is NGAUSAV=11. The routine performs the convolution of the input drift with a Gaussian weighting function with a FWHM of 11 records (seconds). Explore the effect of smoothing by convolving a 1-D model HI distribution (of your choice; explain what you do, but use NGC 5701 as the context for scaling) with such a Gaussian weighting function. Explain what you do and what you learn. This problem will require some thought and some programming skill!

5. What questions do you have for the rest of us?