dem.py download dem

[wenxuanm42-ctrl]

If my study area is near the coast and has null values, causing issues with final mosaicking, how should I resolve this?

[fdelgadodelapuente]

Hi,

It seems that the USGS link is broken. You can get the SRTM from the ESA website with the following code

dem.py -a stitch -b -1 3 -80 -76 -r -s 1 -c -u https://step.esa.int/auxdata/dem/SRTMGL1/ -f

[falkamelung]

Or use sardem from Scott Staniewicz. It has an option for the Copernicus DEM which is significantly better than the SRTM DEM:
https://github.com/scottstanie/sardem

[Stephen0255]

Or use sardem from Scott Staniewicz. It has an option for the Copernicus DEM which is significantly better than the SRTM DEM: https://github.com/scottstanie/sardem

What is the difference between the Copernicus DEM and the SRTM DEM? Any improvements for the ISCE output interferograms?

[falkamelung]

My experience is that if you take many looks it does not really affect the phase. But when you do high-resolution in hilly terrain the geolocation of your pixels gets better and you might be able to skip geolocation correction using the estimated DEM error.

[fdelgadodelapuente]

Both are InSAR-derived DEMs, but they differ in resolution and accuracy.

Copernicus is the TanDEM-X 30 m DEM acquired during 2011-2014, but with holes masked and/or interpolated and with a much better resolution because the native data from the TanDEM-X mission has a resolution up to 2 m/pixel. The Copernicus DEM has a vertical accuracy of ~2 m (same as TanDEM-X DEM) with respect to the 7-10 m for SRTM. The Copernicus DEM is updated from time to time when DLR releases an update of their TanDEM-X 30 m DEM.

The elevation datum is also different. SRTM is EGM96 geoid , while Copernicus is EGM2008 geoid. ISCE automatically converts EGM96 to WGS84 ellipsoid. For Copernicus you can do that with Gdal.

As Falk mentioned, for multi looked data it might not matter as along as the Bperp is short, such as for ALOS-2 ScanSAR data. But if you work with X-band data with long baselines such as COSMO-SkyMed, or L-band with long baselines such as SAOCOM-1, processing with the Copernicus DEM results in much better interferograms.

Best,

[falkamelung]

For sardem the default is also WGS84. But you may want to double check that it works as there is a possibility for silent errors as your system PROJ may not be used and or your GDAL is not properly installed. I had an issue in the sardem repo on this (which is probably closed by now).