 High-resolution Martian soil thickness derived from THEMIS thermal measurements.  S. N. Heath1, J. Bell1, and P. R. Christensen1. 1Arizona State University School of Earth and Space Exploration (snheath1@asu.edu). Introduction:  Thermal inertia (TI) is an intrinsic property  of  materials,  describing  its  resistance  to changes in temperature.  Because Mars essentially has only a  thin  atmosphere  and  very little  water  vapor, grain size and induration are the dominant effects on thermal inertia [1].  Thermal inertia observations have been  made  on  a  variety  of  scales  using  the  Mars Global  Surveyor  Thermal  Emission  Spectrometer (TES) [2] and Mars Odyssey Thermal Imaging System (THEMIS) [3],  and  are a  useful  proxy for  inferring grain size and general rockiness of a surface.  Most maps and models of thermal inertia assume a homogeneous surface of uniform-sized particles.  The Martian  surface is better  approximated by a layer of low-thermal-inertia  soil  over  a  high-thermal-inertia  foundation of bedrock or permafrost [4],  though this still ignores the effect of mixtures of grain size, such as boulders  strewn in  a  sandy field.   This  two-layer model  enables  unique  determination  of the  physical thickness of the soil layer as well as its thermal inertia  by fitting a computer thermal transport model against  multiple observations.  This layered model, while still an  approximation,  provide  insight  into subsurface permafrost and aeolean processes. A global-scale map  of soil  thickness  based on  a two-layer  thermal  model  was  previously produced using temperature measurements from TES data [4] at a  resolution  of  4x2  degrees  per  pixel.   This  map determined the  extent  of  Martian  permafrost, producing results  broadly agreeing  with results  from the  Mars  Odyssey  Gamma  Ray  Spectrometer experiment.   However,  the  Mars  Odyssey Thermal Emission  Spectrometer  (THEMIS)  instrument  can produce  surface  temperature  images  with  a  thermal resolution nearly as good as TES  yet at much higher spatial  resolution  (100  m/pixel  vs.  5x8  km/pixel), enabling the investigation of the thermal  structure of much smaller surface features [5].  We have produced software  using  THEMIS  thermal  infrared  images along with a layered thermal model to produce maps of  soil  thickness  with  a  spatial  resolution  of  100 meters  per  pixel  and  thickness  resolution  of several centimeters  up  to two  meters.   This  opens  the possibility  of  more  detailed  study  of  the  nature  of small-scale thermal inertia features. Background: The  depth  to  which  the  subsurface  thermal properties  of  a  planetary  body affects  the  observed surface temperature  can be described by the thermal skin depth, δ defined as: δ= I ( ρc ) √ ( P/ π ) where I is the thermal inertia  of the surface,  ρ is the density, c is the specific heat at constant pressure, and  P is the  period of a  cyclic temperature  change. The thermal effect of materials deeper than a few skin depths  is  rapidly  washed  out  by  the  effect  of  the overlying  material.   Thus  the  observed  surface temperature  over  the  length  of a  day/night  cycle is strongly  affected  by  only  the  top  few  mm  of  the surface.  However, by observing over a longer period, we  can  look  "deeper"  into  a  surface's  thermal properties.   Seasonal temperature variations over the period of a Martian year have a thermal skin depth of decimeters  to  meters,  providing  the  opportunity  to observe the structure of the near surface. Methods: We made a general-purpose program  to produce soil thickness maps from THEMIS thermal images.  A spreadsheet of target locations is assembled using the JMars  GIS  software,  with  approximate  values  of surface  albedo,  thermal  inertia  and  average atmospheric opacity from TES maps [6] and elevation from MOLA maps.  For each target location the KRC thermal  transfer  model  [7]  is  used  to  predict the seasons during  which the  effect of sub-surface layer thickness on surface temperature will be greatest.  The THEMIS  image  archive  is  then  searched  for high-quality  images  of  the  target,  taken  at  times within 20 degrees solar longitude of the ideal season. Two observations are necessary to solve for the two free variables of the problem: soil layer thickness and thermal  inertia.   Ideally  they  are  taken  during different  seasons,  generally  mid-summer  and mid-autumn, when variation in soil thickness has the largest  effect  on  surface  temperature;  otherwise  the high  thermal  inertia  bedrock layer  and  low thermal inertia soil layer tend to have opposing effects.  Many images are culled at this stage due to low temperature contrast,  atmospheric dust,  and the presence of CO2 frost  which  obscures  the   surface  beneath.   The remaining  images  are  projected  using  the  ISIS3 software package and potential image pairs are chosen and  cropped,  allowing  a  pixel-by-pixel  comparison between  the  two images  of  each  potential  solution. The  KRC  thermal  model  is  then  used  to  generate surface  temperature  predictions during  the  observed seasons  for  a  variety of  TI+thickness  combinations, assuming  a  lower  layer  of  bedrock  or  permafrost  (which  have  similar  thermal  inertias).   The combination of soil TI and thickness that results in the observed  surface  temperature  during  both  seasons provides a unique solution to those properties. Results: The yield of successful model solutions was very low;  even  after  tightening  the  selection  of  source images,  <10%  of the  targeted  areas  produced  maps with  recognizable  features  not  caused  by  terrain,  image artifacts or interference from atmospheric dust or surface frost.  However, due to automatic selection and  processing  of images,  a  large  number  of  areas could be investigated,  with the main  difficulty being interpretation  of  the  output.   Four  surveys  were undertaken:  One of the  landing  sites  of  the  Viking, Mars  Pathfinder,  Mars  Exploration  Rover,  Phoenix and  Mars  Science  Laboratory  spacecraft to  test  the technique  against  ground  observations,  one  of  a catalog of glacier-like forms produced by [8], one of a grid  of arbitrary points selected to cross a  variety of terrains  and  latitudes,  and  one  of  locations hand-picked to contain high-contrast thermal features. The  best  results  came  from  the  survey  of hand-picked  locations,  targeted  at   thermal  features visible  in  minimally-processed  THEMIS  images. These  produced  a  high  proportion  of  good  results, including  several  where  different  image  pairs produced  very  similar  output.   In  line  with expectations,  at  low  latitudes  there  is  usually  not enough  seasonal  temperature  variation  to  produce a strong  signal;  however,  in  some  areas  such  as Cerberus Fossae and Athabasca Vallis,  features such as dust  streaks were very apparent  even with  minor  seasonal temperature changes.  In areas with very low thermal  inertia  (such  as  Tharsis) a  unique  solution could not be found due to the thick layer of low-TI soil obscuring  any  sign  of  underlying  bedrock,  even  at high  latitudes  with  large  seasonal  temperature variations. Discussion and Conclusion: It  is  possible  to  use  surface  temperature measurements  from  spacecraft  combined  with  a detailed  thermal  model  to  investigate  the thermo-physical  structure  of  a  planet's  surface  in detail,  and  the  THEMIS  instrument  is  capable  of providing  this  data  at  relatively  high  spatial resolution,  allowing investigation of specific features on a local scale.  An automated analysis program was developed to automatically perform this analysis on all suitable THEMIS data for a target location, capable of running through a large number of potential locations.  It  has  successfully  produced  maps  of soil  thickness assuming  a  uniform  soil  over  a  layer  of bedrock or permafrost,  which matches well with observations of Mars's surface. However, without direct surface observations it  is very  difficult  to  separate  valid  data  from contamination  and  noise.   The  analysis  is  very sensitive  to  instrument  noise,  environmental interference,  and  model  assumptions.   The  original  intent was to look at a large number of specific classes of features such as glacier-like forms, but the difficulty of interpretation makes it difficult to draw quantitative conclusions from this.  The best results were obtained when analyzing very strong signals, such as the strong thermal  differences  apparent  in  wind  streaks  in  the Cerberus Fossae region.  Thus this technique may still  prove to be useful for studying the nature of thermal  features, wind processes, and local geomorphology. References: [1] Presley M. A., and Christensen P. R. (1997b) JGR, 102,  9221-9229.  [2] Putzig N. E. et al (2005) Icarus,  173, 325-341.   [3]  Fergason  R.  L., Christensen  P.  R.,  Kieffer  H. H.  (2006)  JGR,  111, E12004.   [4]  Bandfield  J.  L.  and  Feldman  W.  C. (2008) JGR, 113, E08001.  [5] Bandfield J. L. (2007) Nature,  447, 64-68.   [6]  Christensen  P.  R. personal communication.   [7]  Kieffer,  H. H. Thermal  models for analysis of Mars infrared mapping, manuscript in preparation.  [8] Souness C. et al (2012) Icarus, 217,  243-255. Figure  1.  Left:  a  THEMIS  night  temperature  image  of Athabasca Vallis,  156°E 10°N,  scale  bar  is  10 km, arrow points north.  Right: Soil thickness map for the same image. 
