Author Topic: Vertical Reference for SOCET SET HiRISE DTMs vs. PDS?  (Read 1380 times)

dmayer

  • Osiris (Active Member)
  • **
  • Posts: 5
    • @GeologyDave
    • Email
Vertical Reference for SOCET SET HiRISE DTMs vs. PDS?
« on: February 25, 2015, 10:33:12 AM »
Hi all,

I'm posting this in "GIS Applications" rather than "Planetary Datasets" because I think it has more to do with processing in SOCET SET than with PDS datasets themselves.

I've extracted a DTM from a HiRISE stereopair over Gale crater using SOCET SET by following the USGS tutorial in http://astrogeology.usgs.gov/search/details/Docs/Photogrammetry/April-29-May-1-2014-Files/zip
The HiRISE images I'm using are
http://www.uahirise.org/PSP_001488_1750
http://www.uahirise.org/PSP_001752_1750

I'd like to compare my DTM to both MOLA and the HiRISE DTM on the PDS.  First, just to get a rough sense of the quality of my DTM, I used map2map to match the projection of a section of the MOLA MEGDR to my HiRISE DTM and then subtracted the MOLA elevations from my DTM using algebra in ISIS.  The resulting difference image shows that my DTM is typically hundreds of meters lower than MOLA, with a median difference of ~-430 meters. Similarly, subtracting the PDS DTM elevations from my DTM yields a median difference of ~-430 meters (so at least the PDS DTM is consistent with MOLA).

The documentation for the HiRISE DTMs on the PDS says that they're referenced to the areoid. It is my (faulty?) understanding that the HiRISE DTMs on the PDS are produced using SOCET SET following a procedure similar to the one I link above, so I figured that both my DTM and its PDS counterpart have the same vertical reference. However, I recall that during ingestion of the HiRISE data into SOCET SET, the USGS procedure instructs the user to set the vertical reference to the Mars2000 ellipsoid, which I believe is different from what MOLA uses.

I'm more than open to the possibility that there's a problem with my DTM, but a deviation of several hundred meters seems too large to be completely explained by user error during DTM generation.  As an additional test, I downloaded HRSC DA4 and DT4 products for the study area and calculated image difference betwen my DTM and each of the HRSC products. As expected, the deviation between my DTM and the DA4 product is ~-430 meters. The median deviation between my DTM and the DT4 product is ~-50 meters. The deviation between my DTM and the DT4 product appears to increase from ~-20 meters to ~-70 meters moving from south to north in the scene. I know that there is an approximately northward-dipping slope in the scene. Perhaps this trend in the deviations is due to the fact that the DT4 product is referenced to a sphere with radius 3396.0 km?

Questions:
1a) What is the vertical reference for HiRISE DTMs produced according to the USGS SOCET SET tutorial? Is it an aeroid, a sphere, something else?
1b) If my DTM is referenced to something other than the areoid, what would I need to do in order to produce DTMs with SOCET SET that are referenced in the same manner as the DTMs on the PDS?
2) In areas where MOLA shot data (PEDR data) are sparse (such as this area over Gale), is there a best practice for evaluating HiRISE DTMs against other published DTMs such as MOLA and HRSC?

Any insight would be greatly appreciated!

--Dave