I am exploring the Collocation operation tool located under the Raster\Geometric Operations\ tab.
I see that one has to resample the S2 product before running the operation. In my case, I have chosen 10 m resampling resolution post which the Collocation operation runs seamlessly. I would ideally like to bring both the datasets to the same pixel size such that each pixel in S1 corresponds to exactly the same area in S2 at a mutually feasible resolution.
Following are few queries regarding the output of the operation and further processing of the data:
Since the S1 footprint is larger than S2, I noticed that the overlapping portion of the S1 scene (with S2) has been cropped and the cropped segment appropriately aligned with the S2 scene.
The S2 scene is still in the L1C format and has not yet been converted to L2A. The S2 scene also needs to be reprojected in order to be able to get exported as KMZ, say post a Band Math or Classification operation. Similarly, the S1 has not yet been speckle filtered and RC terrain correction not yet applied.
I am curious to know if the Range-Doppler terrain correction is an inherent part of the Collocation operation or is it just a mere rotation of the S1 image to match both the scenes.
By looking at the headers of the exported Collocated scenes in the PolSARPro format (Export\SAR Formats\PolSARPro) I see that both the scenes have been dimensionally equalized (same rows & columns) with same GCS (WGS84) and PCS (UTM) but different data types (in ENVI format) for S1 (data type=3) & S2 (data type=12) scenes. I generally export the scenes in this format to process the scenes in Matlab using my own scripts targeted towards a particular application.
I would also like to know the individual processing steps one needs to apply on the S1 & S2 before combining them with the Collocation operation and then exporting them for further processing in Matlab. The final aim is fusing S1 & S2 datasets to derive complementary insights from both these sensors.