TRUE-COLOR-S2-L1C layer configuration about eo-learn HOT 8 CLOSED

Hi @nilshempelmann

Make sure the layer configuration is set up properly as for these requirements.

Login to the Sentinel Hub configurator and add a new configuration using the Python scripts template. Then update your instance ID and the script should run.

from eo-learn.

Thanks for pointing it out. I've added a README (https://github.com/sentinel-hub/eo-learn/tree/master/examples) where instructions on how to add new configuration to your SH account using a predefined template that has all layers used in the examples are given -- including the TRUE-COLOR-S2-L1C and BANDS-S2CLOUDLESS.

Based on your error I assume you have changed the layer name in the following line

add_clm = AddCloudMaskTask(cloud_classifier, 
                           'BANDS-S2CLOUDLESS', 
                           cm_size_y='160m', 
                           cm_size_x='160m', 
                           cmask_field='CLM',
                           cprobs_field='CLP',
                           instance_id=INSTANCE_ID)

If you did, I suggest you to change it back to BANDS-S2CLOUDLESS, since the cloud detector needs the 10 bands in order to make the cloud classification.

from eo-learn.

Thank you very much, this helped to actually be able to use your notebooks for testing. Unfortunately, now I get the following error from the same cell:

DownloadFailedException: Failed to download with HTTPError:
400 Client Error: Bad Request for url: https://services.sentinel-hub.com/ogc/wcs/0f6c0cd7-002d-4460-8017-4d78502d7d35?SERVICE=wcs&BBOX=370230.5261411405%2C5085303.344972428%2C380225.31836121203%2C5095400.767924464&FORMAT=image%2Ftiff%3Bdepth%3D32f&CRS=EPSG%3A32633&RESX=160m&RESY=160m&COVERAGE=TRUE_COLOR&REQUEST=GetCoverage&TIME=2017-01-01T10%3A04%3A07%2F2017-01-01T10%3A04%3A07&MAXCC=80.0&ShowLogo=False&Transparent=False&EvalScript=cmV0dXJuIFtCMDEsQjAyLEIwNCxCMDUsQjA4LEI4QSxCMDksQjEwLEIxMSxCMTJd
Server response: "Layer TRUE_COLOR not found"

I did not change the notebooks in any way, so I am sure to use the proper version. Now I don't know why it asks for the TRUE_COLOR layer, since the layers specified are TRUE-COLOR-S2-L1C and BANDS-S2CLOUDLESS. The configuration is as explained in the README.

If it is of any help: The error occured after the cell was running for several minutes. Network traffic was on full load, so I guess it was actually downloading some data.

from eo-learn.

OK, I see now where the problem is. The AddCloudMaskTask actually doesn't create a SentinelHub request using the 'BANDS-S2CLOUDLESS' layer but does this by using eval script (simple script that returns the 10 necessary bands). Unfortunately, the OGC standard requires the LAYER parameter always to be specified although it has no impact on what kind of data is returned (10 bands in this case). The AddCloudMaskTask at the moment sets the layer to 'TRUE_COLOR' when it makes the request (the layer name is hardocded). If this layer doesn't exist then the above error is issued.

This clearly needs to be fixed in the AddCloudMaskTask, but until then please use the updated version of example notebook. The issue can be circumvented by downloading the 10 bands in another task. If the data is already available the cloud detector doesn't download the data again.

from eo-learn.

@fvivian the issue should be solved now. Please update your version of eo-learn and run the examples again.

from eo-learn.

Thank you for the explanation and quick fix. The example runs without any issues now.

from eo-learn.

@fvivian Sorry for opening the issue again.
There is still/again a :
Server response: "Layer BANDS-S2-L1C not found"

Was running your notebook

Is therer a way to query the Layer names?

from eo-learn.

Got it. Thanks !

from eo-learn.