Principal Component Analysis of MGSTES Data and Comparison with Modeling

1
Principal Component Analysis of MGS-TES Data
andComparison with Modeling

• Guo, Xin
• October 7th 2004
• Advisor Yung, Yuk L.

2
Outline

• Mars, Mars Global Surveyor (MGS), Thermal
  Emission Spectrometer (TES)
• Principal Component Analysis (PCA)
• Results of PCA on TES data
• Results of PCA on synthetic data
• Results of PCA on GCM output
• Conclusions and future work

3
Mars Facts

• A Martian year is 668 sols (Martian days), 687
  Earth days
• A Martian day (sol) is 24 hours 37 minutes 22 sec
  
• Atmospheric gaseous components CO2 (95), CO
  (700ppm), H2O (100ppm), N2, O2,O3, NO, H2, Noble
  Gases
• Major Aerosol Components Dust, Water Ice
• Atmosphere shows annual variation and diurnal
  variation

Pater, I.d. and L. Jack J, Planetary Sciences.
2001, Cambridge Cambridge University Press.
4
MGS TES
http//tes.asu.edu/images/newtesimage.jpg
http//mars.jpl.nasa.gov/mgs/images/mgs-mons.jpg

• MGS (Mars Global Surveyor)
• Orbit covers almost the whole surface of Mars
• One orbiting period of MGS at normal mapping
  phase is 118 minutes
• At normal mapping phase, a global mapping takes 7
  sols 3.78º Ls 172.62 hours

• TES (Thermal Emission Spectrometer)
• Spatial resolution 3 km
• Spectral range 200 cm-1 to 1700 cm-1
• Spectral resolution 10 wavenumbers (cm-1) or 5
  wavenumbers (cm-1)
• SNR around 400 at 1000 cm-1
• Sample rate around 800 per second

5
Principal Component Analysis (PCA)

• Terminology Meteorologists call it Empirical
  Orthogonal Function (EOF) Analysis, Factor
  Analysis I am trying to be a statistician here
• Linearly transforms an original set of variables
  to a substantially smaller set of uncorrelated
  variables that represents most of the information
  in the original set of variables
• Capture the variation of data
• 1st principal component (PC1) captures the
  largest variation
• 2nd principal component (PC2) captures the
  largest variation orthogonal to that captured
  by the 1st principal component

6
Previous work of Huang et al.

• PC1 is associated with surface or near surface
  brightness temperature
• PC2 is associated with atmospheric variability
• Signal from surface emission (surface or near
  surface temperature) is dominant

7
Manipulation of Data

When (nadir view), and
(thus
). Ignore the strong CO2 absorption band
between 510 cm-1 and 810 cm-1 Apply PCA to the
residual spectra.
8
PCA on TES data MY25 Ls 30º-45º
9
PCA on TES data MY25 Ls 90º-105º
10
PCA on TES data MY25 Ls 330º-345º
11
Discussion of Results

• Variability of atmospheric dust and water ice
• Incompleteness of the removal of surface emission

Smith, M.D., J.L. Bandfield, and P.R.
Christensen, Separation of atmospheric and
surface spectral features in Mars Global Surveyor
Thermal Emission Spectrometer (TES) spectra.
Journal of Geophysical Research, 2000. 105(E4)
p. 9589-9607.
Smith, M.D., Interannual variability in TES
atmospheric observations of Mars during
1999-2003. Icarus, 2004. 167 p. 148-165.a
12
PCA on Synthetic Data

• Feed the Radiation Model with temperature
  profile, pressure profile, atmospheric dust
  mixing ratio profile, atmospheric water ice
  mixing ratio profile (12 levels)
• Generate IR radiation spectra with different
  abundance of dust and water ice
• Get rid of the surface emission and CO2
  absorption band
• Apply PCA on data

13
PCA on GFDL Mars GCM Based Data

• Geophysical Fluid Dynamic Laboratory (GFDL) Mars
  General Circulation Model (GCM)
• Spatial resolution
• 6 degrees longitude, 5 degrees latitude, 20
  vertical levels
• Output fields
• eight 3D fields, eleven 2D fields
• Output interval
• 2 sols, 2 Martian hours

14
Comparison between GCM and TES
Smith 2004
15
Conclusion and Future Work

• Atmospheric aerosol variability is well captured
  using this method. It is independent of the
  retrieval.
• Better removal of surface emission would lead to
  better results.
• A better radiation model (such as MODTRAN) would
  improve the understanding of the roles of various
  species.
• PCA is a good way to test the GCM and help to
  improve it. Eventually, we would like to predict
  the weather on Mars.

16
The End
Acknowledgements
Xianglei Huang, Yuk Yung, Michael Smith, Run-Lie
Shia, Xun Jiang, Dave Camp for useful guidance
and discussions Oded Aharonson for the access of
Martian surface emissivity data Mark Richardson,
Shabari Basu, Michael Mischna, Jiafang Xiao for
the access of GCM outputs

• Thank you for listening

17
References

• Pater, I.d. and L. Jack J, Planetary Sciences.
  2001, Cambridge Cambridge University Press.
• Albee, A.L., et al., Overview of the Mars Global
  Surveyor mission. Journal of Geophysical
  Research, 2001. 106(E10) p. 23291-23316.
• Christensen, P.R., et al., Mars Global Surveyor
  Thermal Emission Spectrometer experiment
  Investigation description and surface science
  results. Journal of Geophysical Research, 2001.
  106(E10) p. 23823-23871.
• Weisberg, S., Applied Linear Regression. Second
  Edition ed. Wiley Series in Probability and
  Mathematical Statistics, ed. V. Barnett, et al.
  1985, New York John Wiley Sons.
• Jolliffe, I.T., Principal Component Analysis.
  Springer Series in Statistics, ed. D. Brillinger,
  et al. 1986, New York Springer-Verlag.
• Huang, X., J. Liu, and Y.L. Yung, Analysis of
  Thermal Emission Spectrometer data using spectral
  EOF and tri-spectral methods. ICARUS, 2003. 165
  p. 301-314.
• Smith, M.D., J.L. Bandfield, and P.R.
  Christensen, Separation of atmospheric and
  surface spectral features in Mars Global Surveyor
  Thermal Emission Spectrometer (TES) spectra.
  Journal of Geophysical Research, 2000. 105(E4)
  p. 9589-9607.
• Richardson, M.I. and R.J. Wilson, Inverstigation
  of the nature and stablility of the Martian
  seasonal water cycle with a general circulation
  model. Journal of Geophysical Research, 2002.
  107(E5).
• Smith, M.D., Interannual variability in TES
  atmospheric observations of Mars during
  1999-2003. Icarus, 2004. 167 p. 148-165.

18
Solar Longitude (Ls)

• A Martian year is defined 360 degree of Solar
  Longitude (Ls) or Heliocentric Longitude
• Ls 0, northern hemisphere vernal equinox
• 1 Ls 45.67 hours

19
Manipulation of Data

• where is the surface emissivity at
  frequency , is the surface
  temperature, is the normal
  column-integrated (aerosol) opacity, is
  the cosine of the emission angle.
  is the Planck function, is the
  temperature profile
• Denote

When (nadir view), and
(thus
). Ignore the strong CO2 absorption band between
510 cm-1 and 810 cm-1
20
(No Transcript)
21
(No Transcript)
22
(No Transcript)
23
(No Transcript)
24
(No Transcript)
25
(No Transcript)
26
Blue Mars
(Michael Carrol, space artist)
27
Simplified Geologic Map
28
Epithermal Neutrons
(Boynton et al, Science, 2002)