Retrieving optical depth from shadows in orbiter images of Mars

The difference in brightness between shadowed and sunlit regions in space images of Mars is a measure of the optical depth of the atmosphere. The translation of this difference into optical depth is what we name the “shadow method”. Our analysis of two HRSC data-sets and a HiRISE data-set indicates that it is possible to estimate the optical depth with the shadow method. In colors between yellow and red the accuracy may be around ±15%, and in some cases ±8-10%. In other colors we found larger errors.We came to these results in two steps. First, we investigated in how far shadow method retrievals are proportional to the true optical depth. To this end we analyzed about 150 locations in Valles Marineris that were imaged by HRSC. Whereas the studied region spans about 8 km in altitude we were able to study the relation between altitude and shadow-method retrievals. Retrievals from five HRSC panchromatic (675 ± 90 nm) stereo images yielded scale-heights with an average of 12.2 ± 0.7 km, which is very close to the expected local pressure scale height. Many studies have shown that the scale-height of optical depth and pressure commonly are similar. This indicates that the shadow method retrievals are on average close to proportional to the optical depth, because otherwise these would probably not yield a correct scale-height. HRSC’s red image yielded very similar results, but the blue, green, and NIR images did not.Next, we compared optical depth measurements by the two MER rovers with shadow method retrievals from orbiter images of the rover exploration sites. Retrievals with the shadow method appear systematically smaller than the rover measurements; dividing the retrievals by a “correction factor” yields an estimate of the real optical depth. Retrievals from three HRSC panchromatic stereo images of a region near the Spirit rover yielded a correction factor of 0.63 ± 0.09 when the sunlit comparison regions were at varying and more or less arbitrary distances from the shadows and 0.71 ± 0.06 when these were close together. Twenty retrievals from a HiRISE red (650 ± 100 nm) image of the Opportunity exploration site similarly yielded 0.68 ± 0.09. The results from these two case studies suggest that the shadow method has an accuracy of about ±15% or around ±8-10% in the best cases.