Simone Juul Jakobsen

A thesis for the degree of Doctor of Philosophy defended March 2019.

The PhD School of Science, Faculty of Science, Astrophysics and Planetary Science, Niels Bohr Institute, University of Copenhagen

Academic advisors:
Associate Professors Morten Bo Madsen and Kjartan Münster Kinch

Assessment Committee:
Associate Professor Christine Schøtt Hvidberg, NBI (chairman)

Professor Hans Kjeldsen, Institut for Fysik og Astronomi, AU

Senior Scientist Dr. Antoine Pommerol, Universität Bern, Switzerland

_{Modelling and Correction of an Artifact in the Mars Exploration Rover Panoramic Camera}

In 2004 two identical rovers touched down on the surface of Mars on opposing sides of the planet. The two rovers Spirit and Opportunity were both part of the joined mission Mars Exploration Rovers (MER), and their mission was to characterise their surroundings and identify possible signatures of past liquid water. From the collected data, information about the climate of the planet can be derived and used to shed light on whether life could have existed on the planet at earlier times, or whether life could still exist to this day, just below the red Martian dust.

One of the main rover instruments is the panoramic camera called Pancam. It's aim is to obtain high resolution images of Mars that can be used both for navigation purposes as well as scientific analyses. The importance of being able to trust the collected observations cannot be underlined enough and it was therefore important that Pancam could maintain it's calibration throughout the full mission. In order to calibrate Pancam during the mission a so called calibration target, mounted on the deck of each rover, was utilized. The calibration method consists of having characterised the calibration target to a high degree prior to launch, by measuring the precise reflectance of it's different coloured and grey scale areas, and from that derive a model. Later, during the mission, it was possible to obtain Pancam images of the cal target and extract spectra from the images of the same coloured and grey scale areas, and scale them to the models. This way the Pancam images, obtained at the same time and place as the cal target images, could be calibrated.

Prior to launch, the models were compared to extracted reflectances from Pancam images of the cal target in order to pre-calibrate Pancam, and during these activities, a discrepancy was found at the longest wavelengths let in by the camera, in the filter called the R7 filter. The discrepancy consisted of a decrease in contrast in the R7 filter, that is the brightest area of the cal target was darker in the image than what the model predicted, and the darkest area of the cal target was brighter in the image than what the model predicted.

The work presented in this thesis focus on describing the effect, both in images of the cal target, but also in images of the Martian surface. That the R7 filter artifact is observed in all types of images points towards it being an instrumental effect, and as the present work will show, an effect that most likely arises in the CCD after the light has passed through the camera optics. From this assumption, a hypothesis is formulated saying that the effect arises as a consequence of light being reflected off the backside of the camera CCD, which leads to a signal registered in other pixels than the intended one. Based on this, a mathematical model is developed that describes the effect. The developed model is used to simulate the effect, and it is shown that this can be done to a very high degree of precision. Since the primary aim is to remove the effect from affected R7 images, the model is invoked on the inverse problem through an iterative deconvolution algorithm, where the output is designed to converge towards the corrected version of the image. The work presented here shows that the developed correction removes the effect to a high degree, both from simple, symmetric images, and complex images of the Martian surface. Furthermore, it is shown how the effect could have influenced previously published work, where the analysis was based on R7 filter images.