Different size for VV band in v1 and ogc


#1

Hello, when I use
http://services.eocloud.sentinel-hub.com/v1/ and try to get results for 1_VV___ORTORECTIFIED layer or 2_VV_NON_ORTHO. I have different dimension of Sentinel-1 results in comparison with results for this layers from https://services.sentinel-hub.com/ogc/. same BBOX in both cases.

If first case I got (154, 125, 1) for single record. In second (183, 125, 1) (VV band)

Is it normal situation?


#2

Should not be.
Could you send all URL requests, so that we investigate the details? Do mask out major part of the Instance ID (except of first 5 chars)


#3

Hi!
This is the code:
for v1

from eolearn.io.sentinelhub_service import S1IWWCSInput


from eolearn.io.sentinelhub_service import S2L1CWCSInput
from sentinelhub import SHConfig

sh_config = SHConfig()  # Collects configuration from config.json file
sh_config.instance_id = '1f5e2e04...'
sh_config.ogc_base_url = 'http://services.eocloud.sentinel-hub.com/v1/'
sh_config.save() 

key = 'VV'
input_task = S1IWWCSInput(key, resx='10m', resy='10m')

workflow = LinearWorkflow(input_task)
time_interval = ['2016-01-01','2018-12-31']
from sentinelhub import WebFeatureService, BBox, CRS, DataSource

bbox = BBox(bbox=[ 150.6450033,-32.4219569,150.6583071,-32.4383657], crs=CRS.WGS84)

results = workflow.execute({input_task:{'bbox':bbox, 'time_interval':time_interval}})
patch = list(results.values())[-1]
print(patch.data[key][idx][:,:,0].shape)
print (len(patch.timestamp))

for ogc:

from eolearn.io.sentinelhub_service import S1IWWCSInput
from sentinelhub import SHConfig

sh_config = SHConfig()  
sh_config.instance_id = '4877a04...'
sh_config.ogc_base_url = "https://services.sentinel-hub.com/ogc/"
sh_config.save() 

key = 'BANDS-S1-IW'
input_task = S1IWWCSInput(key, resx='10m', resy='10m', maxcc=10.0)

workflow = LinearWorkflow(input_task)
time_interval = ['2016-01-01','2018-12-31']
from sentinelhub import WebFeatureService, BBox, CRS, DataSource
idx = 1
bbox = BBox(bbox=[ 150.6450033,-32.4219569,150.6583071,-32.4383657], crs=CRS.WGS84)

results = workflow.execute({input_task:{'bbox':bbox, 'time_interval':time_interval}})
patchs12bands = list(results.values())[-1]
print(patchs12bands.data[key][idx][:,:,0].shape)
print (len(patchs12bands.timestamp))

#4

Is there any way to make the pictures into same resolution?
Thank you.


#5

The code above is unfortunately not telling. Are you able to output the actual request, that is done to the service?
Something like what I posted below?

Some notes though at this point:

  • Based on your output values it seems you are looking at visualised values, not actual values. Do configure your layer so that you only get one value out. See this post.
  • There is no orthorectified option on “services.sentinel-hub.com” yet, so you should not compare this.
  • Sentinel-1 natural resolution is 20 meters, not 10 meters so you might be looking into interpolated values. Not sure how large differences you are actually noticing (if you create actual value layer as above, this will be more clear), it might be that it is simply approximation differences, e.g. due to different interpolation options being used

https://services.sentinel-hub.com/ogc/wcs/4877a04a-MASKED?coverage=NDVI&request=GetCoverage&crs=EPSG%3A4326&bbox=51.104000000000006%2C-0.908%2C51.204%2C-0.8079999999999999&format=image%2Ftiff%3Bdepth%3D32f&maxcc=10000.0&resy=10m&version=1.3.0&resx=10m&transparent=True&showlogo=False&service=wcs&time=2016-05-07T11%3A06%3A56%2F2016-05-07T11%3A06%3A56


#6

Analyzing this a bit further it seems that pixel size in the output file is not exactly the same. However, if you plot both in QGIS, you will see that they overlap, because georeference file compensates it. Therefore the sie of pixels should not obstruct your work.
Eocloud system is an older version of the “main” one, which might be a reason for the difference. We are working on an update, which will merge the code base. It should be available in a couple of weeks.