The same Sentinel 3A OLCI l1b data were processed with SNAP reprojection tool and ENVI 5.4.1 "Reproject GLT with Bowtie Correction"tool respectively, but the results have slight shifts geographically. The ENVI result seems to have quite good geographic correction by overlaying an independent shape file to verfiy the corrction accuracy. the SNAP result seems to shift slightly towards southward. I wander how the difference is initiated?
The SNAP reprojection parameter is UTM Zone 51 for this case.
2)The ENVI “Reproject GLT with Bowtie Correction” tools parameters are : Projection Type=Geographic WGS-84，Interpolation Method=Distance Weighted。
By comparasion, the ENVI result is more precise than SNAP result. Maybe it contributes to more detailed implementation of the reprojection algorithm in ENVI. I also overlaid the ENVI result of Sentinel 3A OLCI image with other data sources such as Sentinel MSI , Landsat OLI, the spatial registration is quite satisfactory ,but the SNAP result still showed a discernable shift towards the south (also including a minor westward shift by careful check). I guess the reprojection algorithm implemented in SNAP might have take more situations into consideration. I hope in the later releases can find how the difference occurs and can work through this.
I consulted Macro’s advice and set the OLCI reprojection to pixel based, but the rusult seems the same as the tie-point based case. I alse examined the spatial shifts in UTM 51 projection in meters: the east-west shift is about 50 ,the north-south shift is roughly 355~360. The SNAP result seems to ‘push’ southward mainly and westward very slightly (by visual check on the screen) by comparison to ENVI result.
I don’t know the SNAP implementation of reprojection algorithm is a little bit different from the ENVI implementation. ENVI reprojection implementation shows more precise and reasionable results.Maybe the SNAP reprojection algorithm should be appraised .