What is the difference between GRD and GRD + Terrain correction?
Or how to georeference my SLC SAR images
A SAR radar image is acquired in an inclined frame of reference. The SLC image thus has two axes: range and azimuth.
The question then arises as to how we can overlay it with an optical image of the same area. In this post, I will only take into account the geometric effect of the terrain, and not the so-called ‘layover’, i.e. the 3D elements placed on the terrain (trees, buildings, bridges). Here, we are only interested in the relief of the area.
Level-1 data can be processed into either Single Look Complex (SLC) and/or Ground Range Detected (GRD) products. SLC products preserved phase information and are processed at the natural pixel spacing, whereas GRD products consist of SAR images that has been multi-looked and projected to ground range using the Earth ellipsoid model WGS84.
One might think that the “GRD” product level corresponds to an image that can be superimposed on the optics. This is not quite the case. It is necessary to apply an additional field correction. Indeed, the GRD only takes into account the ellisoidal model of the earth, without the description of the relief!
Terrain Correction is a kind of orthorectification: it converts data from ground range geometry (GRD), which does not take terrain into account, to σ° using the SRTM 30 meter DEM or the ASTER DEM for high latitudes (greater than 60° or less than -60°).
See instead an example of why image sampling differs, even without changing the backscatter value as a function of local incidence angle:
To superimpose a radar image on an optical image, it is therefore necessary to perform this Terrain Correction step from a GRD image. A subsidiary question arises on how to perform this terrain correction from SLC products? This may be the case if I am working with interferometry, DinSAR, or polarimetric time stacks, etc.
A trick can be to calculate the displacement existing between my reference image SLC of departure, and its corresponding image “Terrain correted” TC. Then, it will be enough to apply this displacement function to the product here of my coherent processing, in order to obtain the orthorectified result.
Note now that some platforms (EObrowser, Google Earth Engine) offer the availability of data where this field correction has been applied.
Bottom Line
● GRD only takes into account the ellisoidal model of the earth, without the description of the relief!
●Terrain Correction is a kind of orthorectification, necessary to combine the radar images with optical ones
●First do the processing in SLC level when possible, then export the result to an orthorectified reference image.
