Skip to main content

To everyone using the s3://sentinel-s1-l1c bucket:



To anyone using Sentinel-1 data:
The entire Sentinel-1 GRD data archive ( Sentinel-1 - Registry of Open Data on AWS) has been reprocessed to fix a GDAL bug affecting overviews (average resampling: “Starting with GDAL 3.1, this is a weighted average taking into account properly the weight of source pixels not contributing fully to the target pixel.” from gdaladdo — GDAL documentation). This affected overview levels whose size did not exactly divide the original image size; by extension the quali…

Hi,


Thanks for your efforts in maintaining these datasets.


Can you please clarify if the GRD data within these buckets are Cloud Optimized Geotiffs or not?


I recently found out about Planetary Computer and that Sentinel 1 GRD COGs are being put up there, and that the data is being managed by Sinergise.


Hence I wanted to know if the AWS Sentinel-1 bucket is also COGs like in Planetary Computer.


Thanks.


Yes, S1 GRD on AWS bucket is in COGs. The structure (and content) is actually very much alike the one in the Planetary Computer.


Oh ! Well… is there any documentation that mentions that it these are COGs?


Sorry but I couldn’t find it anywhere in the registry.aws documentation.


It does indeed not say anywhere, but they are. It should not be too difficult for you to check this by loading a tiff and checking it…


Right, thanks 

I haven’t checked it coz I dont have an AWS account yet. When I use no-sign-request it stops at Access Denied.

But thanks again for confirming. Its really nice to know about this.

Please do consider adding a “COG” mention somewhere for anyone else sifting through the datasets. Thank you!


Hi,


Just tried a rasterio.open on one of the Sentinel-1 GRD files in the Requester-Pays bucket.


GRD/2021/4/23/IW/DV/S1B_IW_GRDH_1SDV_20210423T002140_20210423T002209_026587_032CCC_D1C4/measurement/iw-vh.tiff

with open and “src.meta” gives the following output.


{'driver': 'GTiff', 'dtype': 'uint16', 'nodata': 0.0, 'width': 25983, 'height': 19359, 'count': 1, 'crs': None, 'transform': Affine(1.0, 0.0, 0.0,
0.0, 1.0, 0.0)}

I then tried rioinfo on it which gave a similar answer -


AWS_REQUEST_PAYER=requester GDAL_DISABLE_READDIR_ON_OPEN=YES CPL_VSIL_CURL_ALLOWED_EXTENSIONS=.tiff AWS_ACCESS_KEY_ID=$AWS_ACCESS_KEY_ID AWS_SECRET_ACCESS_KEY=$AWS_SECRET_ACCESS_KEY AWS_REGION=eu-central-1 rio info  /vsis3/sentinel-s1-l1c/GRD/2021/4/21/IW/DV/S1B_IW_GRDH_1SDV_20210421T004331_20210421T004356_026558_032BE8_BD44/measurement/iw-vv.tiff

The output is big, hence please check the public link below -

0d1a4ad2274e00123f146e04e25d300cde546a4b.pngSlack
9cbad1f91e077300dad85387d6c75661bc30f078.png

Untitled



Public file shared from https://slack.com/







then I also tried a simple gdal_translate, which gave the following output-


gdal_translate /vsis3/sentinel-s1-l1c/GRD/2021/4/23/IW/DV/S1B_IW_GRDH_1SDV_20210423T002140_20210423T002209_026587_032CCC_D1C4/measurement/iw-vh.tiff output.tif \
-projwin_srs EPSG:4326 \
-projwin -97.42469787597656 42.83292607575776 -97.39688873291016 42.85407131053984 \
-co COMPRESS=LZW \
--config AWS_ACCESS_KEY_ID xxxx \
--config AWS_SECRET_ACCESS_KEY xxxxx \
--config AWS_REQUEST_PAYER requester

Input file size is 25983, 19359
ERROR 0: -projwin_srs ignored since the dataset has no projection.
ERROR 1: Error: Computed -srcwin -98 42 0 0 has negative width and/or height.


In all cases the CRS seems to be null.


But the “gcps” key in the above “rioinfo” output has EPSG:4326 as CRS.


Oddly when I downloaded the whole COG file and opened it in QGIS, it was correctly projected to EPSG:4326.


Please do let me know your thoughts.


Thanks.


Hi,
note that the community on this forum focuses on access to data via Sentinel Hub. I can therefore not guarantee you will get an answer to your question. Perhaps Stack Overflow or similar would be better.
Best,
 


Just wanted to confirm if its related to the dataset in anyway. That’s all.


Thanks.


The above issue listed by me, does not occur on other requester pays buckets, such as Landsat Collection 2 in AWS.


Please let me know where I can contact folks from Sinergise. Maybe they can answer my question. Because this is the link kept in AWS Sentinel-1 Bucket page - https://shforum.sinergise.com/c/aws-sentinel


I will withdraw the post from here in that case.


Well, we are Sinergise folks… But note that our replication of data to AWS bucket is a voluntary project and we have no resources to provide support to AWS users.

The data, as they are available, work perfectly fine for us and for many other users of AWS.


Sorry for not being able to help you.


And no need to withdraw post from the Forum. It might still happen that someone will answer you. Just wanted to inform you that there is a chance nobody will.


Okay great. Thank you.


After lots of searching I have ended up with the following link , that helps in creating a code for querying Sentinel-1 data from AWS -


github.com/rasterio/rasterio












create transform matrix from GCPS




rasterio:maint-1.0vincentsarago:gcptransform




opened 06:31PM - 20 Aug 19 UTC










ref #1479
👋 @sgillies,
This PR is a first sketch for getting the geotransform from GCPs. The code itself works but the logic might not be perfect. Right now what happens is:
- when opening a files rasterio will try to fetch de transform matrix using `GDALGetGeoTransform`
- if `GDALError.failure` is returned then it fetches the gcps (L308). (need better error handling)
- after `self.meta` call it check if the CRS is defined, and if `_gcps` is set then it get the CRS from it <--- this is not great

### To Do
- rx] better logic (especialy for CRS)
- rx] better error handling
- rx] add tests

I know this issue is set for rasterio 1.1, but someone asked me if it was possible to use rio-tiler on sentinel-1 data so I wanted to give it a try. ref: https://github.com/RemotePixel/remotepixel-tiler/issues/10.

Feel free to reject/close this PR if it's adding noise to the repo 😄







Leaving it here for anyone looking for a similar solution.


Thanks to @gmilcinski for informing me that the datasets are working fine. 🙂


Thanks for your efforts in maintaining these datasets.


Can you please clarify if the GRD data within these buckets are Cloud Optimized Geotiffs or not?


I recently found out about Planetary Computer and that Sentinel 1 GRD COGs are being put up there, and that the data is being managed by Sinergise.


Hence I wanted to know if the AWS Sentinel-1 bucket is also COGs like in Planetary Computer.



Custom Bottled Water


Hi,

AWS Open Data bucket for Sentinel-1 GRD does indeed contain COGs, from the very beginning.

Best,
 


Reply