thank you for providing the data
I just tested it and it technically works:
- import the shapefile into the S1DEM (it is called points) then
- Open the raster
- Open the correlative plot
- Select “Use ROI mask”
- Select “points” as point source and “Elevation” as the attribute
- Select “Show regression line”.
The green line is the regression line, the blue one is the bisector.
As you see, the R² is very low because there is no clear correlation between the actual elevation (from the attribute) and the InSAR elevations from the raster. This has two reasons:
- there are not enough points for a stable regression
- your InSAR DEM is of quite bad quality, probably because of decorrelation (too much time between the images). It doesn’t reflect the topograpy at all and is basically just a ramp.
The regression works if there is a general under or overestimation of elevations or trend in your InSAR DEM. But I’m afraid no regression can fix an InSAR DEM of this quality:
You can select another image pair with shorter temporal baseline. But if there is much vegetation in your study area, you cannot do much about decorrelation, especially with 6 or 12 day repeat pass pairs.