I’ve got a big problem subsetting a S3-OLCI product.
Using this product, S3B_OL_1_EFR____20201011T151615_20201011T151915_20201011T165920_0179_044_239_2520_LN1_O_NR_002.SEN3 (and others in the same location, but with different dates), I’m attempted to subset it to a specific set of Geo Coordinates:
** There is also an issue when inputting the full coordinates to 13 decimal points precision - the rounding that occurs is incorrect. I have manually input the numbers above, rounding to the nearst 2nd or 3rd decimal).**
These coordinates correspond with the bounding box for the following polygon:
After subsetting, however, the resulting product is incorrect:
What ought to be a more portrait oriented result is being created in a landscape orientation. Furthermore, the north and south are are cut off, and the east and west are extended.
This does not occur with similarly located S2 imagery and is impacting the analysis we do
I understand your confusion but I think I can explain.
The round you observe is not really a rounding. The pixel closest to the entered location is looked up and set as pixel-location on the other tab. The the geo-location of this pixel is updated in the “Geo Coordinates” tab.
Regarding the cut at the edges.
If you have a look at the 4 corners of you subset you should still see the values you have entered. At least roughly.
If you have a look the blue top left geometry and the yellow bottom left.
They are the same before and after reprojection (to WGS84).
You can see the Pin_4 is not anymore the most eastern point of my area. and the shape changed completely.
That’s one of the reason the edges are cut.
If you first reproject and then do a subset, the subset should be okay.
The cut off you can still see in the lower right image at the bottom of the area is probably caused by numerical accuracy issues.
It is good to add always a bit of a margin.
Thanks so much, Marco. This solves the skewing issue once and for all! The margins created are another story, but I can use imagemagick to trim the excess and fit into our polygons.