I would give a try to look in here:

http://swl10.blogspot.dk/2015/02/accessing-esa-sentinel-mission-data.html

Hi Sara,

I'm personally not familiar with the SAFE format (thanks for bringing it to my attention!). But, if the link below describes it as I think it does, what you want are the .tif files in the SAFE directory:

https://earth.esa.int/web/sentinel/user-guides/sentinel-1-sar/data-formats/sar-formats

I'm pretty sure GDAL with the Python bindings supports .tif natively. I'm sure GDAL through the command line does; what I do when I need something from GDAL that isn't well supported through the Python bindings is simply call a command line tool using Python's "subprocess" method - you just pass the string you would type on the command line.

Hope that helps!

P

Thank you Michal for your link.

It is interesting but more related to the way to get the data and visualize the tif files. It does not speak about the georeferencing.

Perhaps for a next step

Hello Paulo,

GDAL can indeed read the tif files but unfortunately it looks like there's no georeferences in the ESA tif files. This information is somewhere else in one of the numerous files within the .SAFE directory.

I hoped that someone has already make the effort to find and manage this information using python.

However the link you provide would be useful to any one who want to do it. I will try if I have no answer.

Thanks for your inputs,

sara

Ah, I see, the issue is the transform. Well, as you probably know the .tif files may contain that information themselves. If so, you can read those values with GDAL-Python:

datafile = gdal.Open("theFile.tif")

geoinformation = datafile.GetGeoTransform()

... will give you the six numbers that typically go into a worldfile (.tfw), describing the geotransform of the raster. You get a list six items long with that method. See:

https://bundleblogadjustment.wordpress.com/2011/04/17/extracting-raster-infromation/

...and...

http://gis.stackexchange.com/questions/9421/creating-tfw-and-prj-files-for-folder-of-geotiff-files

Otherwise, if they really don't have those values integrated into the .tif and there isn't also a worldfile (.tfw) to go with it, my next guess is that the info might be in the .safe file, which appears to be an .xml metadata file.

Bon courage!

P

Indeed, my problem is that it looks like there's no geo information in the tif files and GDAL does not know how to find them. For sure they are somewhere in the SAFE data.

So my question is: did someone could find out this transformation and use it to project (lon,lat) points in the image ?

I will have a look on your links that seems interesting.

Merci pour votre aide !

sara

Dear Sara,

I had the same problem while using sentinel-1A L1 data, the image file has a few GCP coordinates but i could not find the total Lat, Long information of the image, if you could find the coordinate information do let me know

Hey,

I do not know if my reply is a little bit too late, but I managed to geo-reference the Sentinel-1 tiffs using the gdal_translate and gdal_warp functions (just as you would for .N1 data). The four corner lat/longs are in the manifest.safe file (read as a xml file) and also in the xml files in the annotation folder.

Let me know if you still need help and whether I should elaborate on my answer.

Gopika

 Hello Gopika,

Yes, I am still very interested with any information on this subject!

Do the four corners are suffisant to retrieve precisely enough the lat/lon positions over all th image ? You don't use the GCP ? However, the way you us them with GDAL very interest me. Do you use python also ?

Thank you,

sara

Dear Sara,

Yes I use Python.

What I did was I extracted the corner coordinates from the xml and then assigned them into the gdal_translate command as GCPs (-gcp pixelposx pixelposy lon lat) followed by a gdalwarp. However, there appeared to be a slight shift in the resulting geotiff in my image (58°N). I've used the same command earlier with .N1 data and it worked fine in lower latitude images.

A colleague of mine came up with a better solution using gdal and GCPs. The GCPs of the sentinel images are available in the header (you can see the 100 or more GCPs when you do gdalinfo for an image). Once you open the sentinel-1 image, you can access the GCPs using the .GetGCP command. Save the GCPs in another array. All that you need to do now is initialise a new image (with its driver, projection and with the array from the original image) and set the GCPs to this using the SetGCPs()command.  This method works beautifully for me even at my LAT.

Hope this helps. Feel free to ask again if you need more clarification.

Gopika

Thank you Gopika, I will try your suggestion !

As of version 2.1 GDAL ships with a Sentinel-1 and Sentinel-2 driver so you can open the archives (.zip) or the .SAFE folders directly.

http://www.gdal.org/frmt_safe.html

http://www.gdal.org/frmt_sentinel2.html

Thanks a lot Kersten for the info and the link.

In fact my problem was more to add the georeferences to the image than to open the files but it looks like gdal 2.1 does both. Great !

Dear all,

As indicated by Kersten, I confirm that GDAL 2.1 can read Sentinel-1 .SAFE data.

However I am still not able to get the transformation in order to be able to project lon/lat positions over the image (I dont want to convert the image itself).

I can get the GCP as suggested by Gopika but from there how can I convert my (lon,lat) coordinates to pixel coordinates ?

Thanks, sara

Dear all,

You should use the OSGEO OSR module in order to init a transformation from one system (your geographic location in your SRS) to another (the location of the image).

Such a transformation can be initialized with the GCP provided by the product itself.

Best regards

Alternatively you can use this:

ds = gdal.Open(filename_tiff)

gcps = ds.GetGCPs()

gt = gdal.GCPsToGeoTransform(gcps)

Hi,

did someone find a solution? I am having the same problem. I am trying to Get the GCPs for the Transformation but I am facing soem problems. @ Gopika Suresh could you provide the code?