Sunday, March 19, 2017

Discerning the Details from the Big Picture: Determining the Geometry of Ice-Crystals in Martian Clouds Through Analysis of Orbiter Data

PVL is headed to Houston! Today nine of our group (!!) arrive at the Lunar and Planetary Science Conference (LPSC) in The Woodlands, Texas. For many it will be their first international conference (and for others their first conference, period!). I remember my first LPSC ('04 - if you are counting) well and hope everyone has as much fun and learns as much as I did. To kick things off, here's undergraduate Brittney Cooper. She has provided a MARCI Composite Image of Mars, (Image: NASA JPL) reproduced above, to add some visual interest and context - enjoy, and come visit us in the poster sessions and in our talks!

By Brittney Cooper

This week will be my first time attending the Lunar and Planetary Science Conference (LPSC) just outside of Houston, Texas. I’m very excited to have the opportunity to not only go to what is known as the largest planetary science conference in the world, but to also be able to present a poster on my research.
            For about a year now, I’ve been working on project analyzing images taken by the Mars Colour Imager (better known as MARCI). MARCI is a camera fixed to the Mars Reconnaissance Orbiter (MRO) satellite, which as you may have probably gathered, orbits Mars. MRO reached Mars in 2006, and began its Primary Science Phase (PSP) in November of 2006, for about 2 Earth years.
The specific images that I’m looking at were taken during this phase, when MARCI was locked in a Sun-synchronous 3am-3pm orbit. This type of orbit is named as such because it’s a polar orbit in which MARCI sees all of MARS at essentially the same local solar time (LST), with MRO crossing Mars’ equator at an LST of 3 pm (or in other words, when the Sun is at a 45 degree angle from what it would be at noon). MRO also crosses Mars’ equator at 3am on what would be the dark side of Mars at that time, but of course MARCI does not image that part of the orbit.

A figure I produced with an approximated orbital path and imaging swath of MARCI over Mars, showing the range of emission angles MARCI observes in each image.  

As a result of this orbit, MARCI takes 12-13 images a day, working as a push-broom imager. The term push-broom refers to the rectangular area used on the charge coupled device (CCD) that MARCI uses to capture images, with 5 different visual wavelength filters physically adhered to it. As MARCI orbits Mars it takes an image every few seconds, just as if you were pushing a broom along a glass-floor and capturing what was below it every-few seconds. MARCI does this for each orbit going from pole-to-pole, and assembling a long composite image that can then be separated into 5 individual images for each filter.  MARCI also has a separate part of its CCD cordoned off for imaging in the ultraviolet wavelengths, set up in a similar way as described for the visual wavelengths, but for only two UV filters.

(a)A small section of a raw MARCI image showing a few frames each divided into 5 filter framelets, which eventually become separated and consolidated into 5 images,

 (b) is an example of a section of the image in (a) compiled for the red filter.

Now that a little bit of background on the MARCI instrument has been provided, I will go on to discuss what it is that I hope to do with these images of Mars.
You may be aware that like Earth, Mars also has clouds, and those clouds resemble what we know to be cirrus clouds on Earth. They are typically optically thin, wispy clouds comprised of ice-crystals as opposed to water droplets. Work is being done from both the surface and orbit to try to understand a great deal more Martian ice-water clouds. We want to be able to characterize their effect on the climate on Mars, as well as answer smaller questions, such as whether or not rainbows can be observed on Mars.
              In my work specifically, I’m hoping to isolate the dominant ice-crystal geometries of Martian water-ice clouds through the analysis of these MARCI images taken during the PSP. I set out to do this by using the known angular field of view of the MARCI imager to assign angles of observation for each column of pixels in a MARCI image, and use that information in combination with the knowledge of the angles at which each pixel receives radiation from the sun (known by the LST) to determine a phase angle. The phase angle tells us how the clouds in any image pixel receive and reflect light, and that knowledge is key to understanding the geometries of its ice crystals because each shape of crystal scatters incident light in different ways for various scattering angles. The parameter used to describe the way these crystals scatter light is known as the phase function. When the phase function is plotted with respect to scattering angles, it can be observed that different ice-crystal shapes have their own individual plot shape analogous to a unique fingerprint. Chepfer et al made a number of laboratory measurements to demonstrate this, and their resultant plot can be seen in Figure 2 (paper found here:  

Our goal is to run through the thousands of images in the PSP and isolate the phase functions of Martian clouds to ultimately perform a comparison, determining the dominant ice-crystal geometries in Martian clouds.
In order to do this, I set up a computational pipeline which reads in the MARCI images from the Planetary Data System (PDS), calibrates them, separates each image into individual filter-consistent images and extracts spectral radiance (or radiometrically calibrated brightness values) from each of the image pixels. The pipeline also calculates the observation and emission angles for each image pixel, calculates reflectance values by dividing the spectral radiance by the flux (of solar radiation that Mars was receiving at the time the image was captured), and uses these values in combination with a few other parameters to determine the scattering phase function for each pixel, in each filter.
In order to ensure that we isolate only those pixels with clouds, we make the assumption that the brightest pixels (which end up becoming the upper boundary of the plots we produce) in each filter will be those corresponding to Martian clouds. This is a safe assumption as we confine our computation range in the image to equatorial regions to exclude the bright polar caps, and we know that ice water clouds scatter equally in all wavelengths, which is why they appear white.
Once we have these plots of scattering phase functions versus scattering angle, we can begin the process of comparing the upper bounds in our plots to previous work done, such as that by Chepfer, et al., in 2002. We can also look at how the scattering phase function of the clouds varies over Mars’ orbit, along with comparisons of the phase function in the red filter (which can largely be attribute to Martian dust) and the blue filter (largely attributed to clouds) to see where a majority of the atmospheric scattering phase function is coming from at various points in a Martian year.
An additional data set that provides values and inputs for calculations in the pipeline that were originally being approximated has been recently discovered. Next steps on this project involve re-vamping the pipeline to accommodate these new variable inputs, and then running the approximately 9000 images within the PSP through the pipeline to produce the desired plots mentioned above. From there we can begin the data analysis to see what we can learn!

For more information on this work, check out my LPSC abstract (, and if you’ll be attending the conference, please feel free to stop by my poster on Tuesday March 21 at 6pm in the Atmospheres and Plasmas session!

Thank-you to Rachel Modestino, Christina Smith, and John Moores for their contributions and guidance on this project.

No comments:

Post a Comment