Blank images when openning the image with unknown values units

Apps / Services Python SDK
1

Hello,

I have created request for sentinelhub to to download image. before saving the image to the disk I display the image with plot_image, and it looks good, but then when I reopen the image with rasterio I get blank image. When I print the image as array it has unexcpected values, with values between 7-8906. That’s een though Ihave used normalizatio nfactor for it to be reflectance.

This is what I have done:

#step 1: evaluate script + dusplay the image, then save it to disk):

evalscript_all_bands = """
    //VERSION=3
    function setup() {
        return {
            input: [{
                bands: ["B01","B02","B03","B04","B05","B06","B07","B08","B8A","B09","B10","B11","B12"],
                units: "DN"
            }],
            output: {
                bands: 13,
                sampleType: "INT16"
            }
        };

    }

    function updateOutputMetadata(scenes, inputMetadata, outputMetadata) {
        outputMetadata.userData = { "norm_factor":  inputMetadata.normalizationFactor }
    }

    function evaluatePixel(sample) {
        return [sample.B01,
                sample.B02,
                sample.B03,
                sample.B04,
                sample.B05,
                sample.B06,
                sample.B07,
                sample.B08,
                sample.B8A,
                sample.B09,
                sample.B10,
                sample.B11,
                sample.B12];
    }
"""


request_all_bands = SentinelHubRequest(
    data_folder='imgs',
    evalscript=evalscript_all_bands,
    input_data=[
        SentinelHubRequest.input_data(
            data_collection=DataCollection.SENTINEL2_L1C,
            time_interval=time_interval,
            mosaicking_order='leastCC'
    )],
    responses=[
        SentinelHubRequest.output_response('default', MimeType.TIFF)
    ],
    bbox=bbox,
    size=bbox_size,
    config=config
)


all_bands_img = request_all_bands.get_data(save_data=True)

plot_image(all_bands_img[0][:, :, 12], factor=3.5/1e4, vmax=1)

print(f'The output directory has been created and a tiff file with all 13 bands was saved into ' \
      'the following structure:\n')

for folder, _, filenames in os.walk(request_all_bands.data_folder):
    for filename in filenames:
        print(os.path.join(folder, filename))
        
        
# try to re-download the data
all_bands_img_from_disk = request_all_bands.get_data()

# force the redownload
all_bands_img_redownload = request_all_bands.get_data(redownload=True)

image
image893×776 394 KB

step 2 : loas with rasterio and display as rgb:

mport rasterio
from rasterio.plot import show
img = rasterio.open('imgs/efd033b10e3ba9e660bf1d97dd111fdb/response.tiff')


#image information
print('  ')
#number of bands
print('number of bands',img.count)

#number of raster columns
print('number of columns:',img.width)
#number of raster rows
print('number of rows:',img.height)

plt.figure(figsize=(20,10))
show(img.read([4,3,2]),transform=img.transform,title='Image- RGB ',vmin=0,vmax=8000)
array=img.read()


>>>
  
number of bands 13
number of columns: 968
number of rows: 879

and the image is blank:

image
image772×649 6.28 KB

when I print the array I get the values which I am not sure in which unit they are:

image

My question is why do I get the image blank? what are the units of this image? can I get it in surface reflectance (for example to acess the sentinle2a which gives me error right now)?

2

The reason you get a blank image is because you retrieve data as digital numbers (DN) - see the “units” in the input of your evalscript setup function, and sampleType INT16 in the output. The normalisation factor “norm_factor” is - in the evalscript - stored to the output metadata, but you never set up the evalscript to output it; you only request the tiff image.

See for instance sh docs or an example with sh-py

When you plot with plot_image, you then apply the factor yourself: you convert DNs to top-of-atmosphere reflectance (you divide by 10k), and you multiply this by 3.5 (to get brighter image), so your first part of the code produces image as expected.

Hope that helps!

4

Sure, just change the collection/dataset to S2 L2A;

1 Like