DEM Co-registration Shifts

Raster GIS menu,  Grid migration (optimal lag), best correlation

El Hiero, Canary Islands

Compare Copernicus and Aster 1" DEMs

Conclusion: horizontal shifts are not a major problem.  This is subjective, and proving it is currently not a major priority.

Initial investigation, detailed results at selected points

Lags/shifts from -10 to 10 in both x and y considered.  The colored regions on the map show the correlations between the two DEMs at each lag, violet being the best and blue the worst; each test region has a different color scale.  In flat regions the difference between the best and worst correlations are small.  The patterns in the correlations relate to the ground; on a ridge, there are high correlations along the strike of the ridge, and much lower perpendicular to it.

Region for the shift ranged from 5 to 100 pixels in each direction from the center point, with region sizes from about 300 m to 6000 m (numbers in the tables below are average; because these are geographic grids, they are rectangular pixels).  For these regions, the smallest regions often gave different lags, while larger regions gave consistent lags.  As the region size increases, so does the relief; the average slope may also change, but could get smaller (if the larger area means a larger proportion of the area is flatter) or larger (if the larger area starts to include steeper areas).

Entire map analysis

TOTAL LAG  Mean 1.0968 Avg Dev 0.1786 Std Dev 0.2988 Skewness 2.4101 Kurtosis 9.0737 n 2832 Minimum 0 Maximum 3.16 Median 1 Lag X direction Y Direction 0 2546 25 1 122 2628 -1 160 175 2 3 175 -2 1   3   4

DEM shifts for El Hiero 1" DEMs, ASTER (should be "worst") and Copernicus (should be "best").  Lags out to 10 considered (max shift is 3.16, so that should be plenty), search radius 25 pixels, and samples every 10 pixels

Other DEM coregistration options

• https://github.com/dshean/demcoreg--might be hard core Python, and setup is more than I want to do at the moment
• https://stereopipeline.readthedocs.io/en/latest/tools/pc_align.html  can do LAS or Geotiff, but Linux and Mac only, so not for me
• https://www.orfeo-toolbox.org/CookBook/Applications/app_FineRegistration.html "works" with some caveats, and the shifts seem smaller than I would have thought
• MA Sutton, WJ Wolters, WH Peters, WF Ranson, SR McNeill, 1983, Determination of displacements using an improved digital correlation method: Image and Vision Computing,Volume 1, Issue 3,Pages 133-139, ISSN 0262-8856,https://doi.org/10.1016/0262-8856(83)90064-1. (https://www.sciencedirect.com/science/article/pii/0262885683900641)
• Nuth, C. and Kääb, A.: Co-registration and bias corrections of satellite elevation data sets for quantifying glacier thickness change, The Cryosphere, 5, 271–290, https://doi.org/10.5194/tc-5-271-2011, 2011.

Last revision 5/29/2022