Introduction

Objective: Analyze spectra of the planets Mars, Venus and Jupiter and look at absorption regions to discern key features of their atmospheric content

Motivation: This is an application of a standard technique for determining the chemical characteristics of stars. With planets we should truly be seeing the absorption spectra of the atmosphere, since there is less distance and extinction of light from us to the source. So this project really aims to gauge how much we can tell about these planets from their atmospheres.

Observing Conditions: We (myself and  Wilson Cauley – a Rice graduate student) took the images at the campus (Rice Univeristy) 16” Meade telescope in Houston, Texas (29°45′46″N  95°22′59″W) to take the spectra on Monday, March 26, 2012. The night was clear, with a humidity of less then 5%. The seeing was close to average in Houston – about 3”. We observed Mars at 8:57 , Venus at 9:02 and Jupiter at 9:07 Central Time.

Tools: Again, we used the Rice University 16” Meade Telescope mounted on the roof of Brockman Hall.We used “The Sky” software to locate the planets and automatically adjust the telescope and the CCDOPS software to take the images.

 

 

Method

Observing Procedure

First using “The Sky” software, we simply typed in the planet we wanted to image and we designated it to move the telescope to that field of view.  The we manually checked that the object was in the field of view using the finder attached to the scope – all three times we could see the object but if we needed to we could go back and adjust the telescope more. Then viewing the image in the CCDOPS software we manually adjusted the focus on the telescope. Next we ensured that the program would take the spectra from the light directly coming from the slit by setting the region of interest in our over all image to the location of the slit (heuristically this is the region where the image became most like a bright vertical line – ie the shape of the slit). Then in the Track menu, we brought up the SlfG or self-guide tool to automatically track and image the planet (though we really weren’t exposing that long for there to be significant deviation in the position of the planet). We wanted to get at least 10,000 counts of integrated brightness for each image ( to get a quality spectrum) and to do this we use a 40 second exposure for all three (there was plenty of light coming from the planets often you have to go upwards  of 500 seconds for an exposure of a star).  We saved the resulting spectra.

We subsequently took a Neon lamp spectrum. This amounted to essentially holding a neon lamp  up to a white screen for about 10 minutes and taking an image of it, using a similar procedure. These will be our Neon calibration images for processing the spectra.

 

Spectra Processing Procedure

The Software we used to process the spectra was IRAF.

I found the following parameters for each of the planet images

Mars:                     ycen =  90    dy = 30

Venus:                  ycen = 100   dy = 24

Jupiter:                 ycen = 81      dy = 34

where ycen was the y position of center of the spectra on the image (which appears as a horizontal band) and dy was the width of the band

I then used the “twod” and  “longslit” packages. For each star, I then change the “nsum” value for  the “longslit” command to my dy value and the “aperture” value to the “scopy” command to ycen, .  Then I ran the “scopy” command on each spectra, using the dispersion axis along lines. I then took a  background spectra from each respective image (just repeating the same procedure but with a ycen significantly away from the actual specta). Finally I subtracted the background spectra from the actual spectra.

From my Neon calibration image  I used “scopy” on the same ranges as I did for each of the planets (to ensure that the calibration is specifically for that row of pixels – we noticed that the calibration lines in the image were tilted so we needed to account for this variation in the y dimension). I used  the “identify” command  with the Chebyshev fit on each of the spectra images to bring up a Intensity vs  X-Position (pixel) graph. Then I manually defined the wavelengths of each peaks in the spectrum using the relative positions of the Neon lines in the calibration image ( I identified the wavelength of these lines using a previously taken Neon spectrum – done by the professor of this course Dr. Johns Krull). The procedure  is to scroll over each peak, hitting ‘m’, then type in the wavelength. The I looked at the residuals from the Chebychev fit  using the ‘f’ key. There were two lines that I had already seen were blurred together in the calibration image and thus I had only used designated the stronger of the too; so when I looked at the residuals they were pretty close to zero deviation already at the second order (linear fit). To reduce some of the oscillations in the residuals, I used a third order (quadratic) fit for the Mars spectra and a fourth order (cubic) fit for Jupiter and Venus. In the end my RSME values were

Mars:       2.284

Venus:    1.925

Jupiter:    2.275

After applying the fit, I got plots of now Intensity vs Wavelength as wells as an “id” file that hold all the detail of the fit. Finally I edited the header to assign REFSPEC1 to each plannets’s spectra to use the “dispcor” command  to each planet to actually apply the fit to the spectra.

 

 

Results and Analysis

The resulting Intensity vs. Wavelength plots are shown here

 

 

 

 

 

 

Technical Notes

Ignore the y-axis on these graphs, it is a result of the algorithm and not a true unit of intensity – all we care about is the shape of the spectra and the dips. The steep Heavyside function behavior at about 6125 angstroms again is a feature of our imaging process and not an actual characteristic of spectra itself. The same is true for the over all trend in decreasing intensity are value higher than about 6900 angstroms.

Observations

The Lyman alpha line is visible in all three around 6500 angstroms – so the presence of H2 is detected.  The is a slight dip at about 6250 angstroms in both the Venus and Jupiter spectrum which may indicate O2. Jupiter has a strong dip at around 7200 angstroms which may actually be the prominent Methane line at 7250 that Jupiter is known to have (this may mean that the calibration at higher wavelengths is slightly off).  There also might be a faint H2O dip around 7200 for Venus ( which should appear as a series of two subsequent dips a narrower one followed by a wider one)

There would possibly be more lines to identify at wavelengths less than 6000 angstroms but due the limitations of our equipment we could not analyzed them here.

 

Sources of Error

Clearly the fit in each case is a large cause of the error. Though the calibration lines themselves are easy enough to identify applying the fit to them is not as straightforward.  The spectroscopy itself could be a source of the error as well due to any number of biases in the CCD imaging.

 

Conclusion

Even with all the processing, I could only discern a few trace elements in the atmospheres of these planets. More sophisticated spectroscopic equipment is required to really get some of the finer details of the spectrum and identify more lines. If we had a Hg calibration image we could also possibly sample over a wider range and pick up more lines towards the bluer side of the spectrum, possibly some metallic elements and nitrogen lines as well.

 

 

References

Kirby, K.P & Goldfield, E.M. 1991. ,J. Chem. Phys., 82, 4720

van Dishoeck, E 1987, J. Chem. Phys., 86, 196