About C2RCC Processor

ABraun · January 20, 2020, 6:36am

the chlorophyll and suspended matter concentrations are in the folder Bands > conc

lisa.f · January 21, 2020, 6:22am

I found conc_tsm and conc_ chl, but no image is displayed.

Is there not enough RAM?

ABraun · January 21, 2020, 6:30am

you can check with the statistics operator if the band has values (then it is s problem of RAM) or not (then it was not computed successuflly)

lisa.f · January 21, 2020, 7:05am

So that means that I’m

running out of RAM.

ABraun · January 21, 2020, 1:38pm

I wonder why the band is not stored under “conc”
grafik

To see if there are values in your raster, you can use the statistics tool

It should show values in the table and the histogram:

while an empty raster will not have any data here.

lisa.f · January 22, 2020, 5:32am

Until now, any data has never been divided into files.

When I started the statistics tool, nothing was displayed.

lisa.f · January 23, 2020, 6:38am

There seems to be some processing that cannot be done with L2A. Is there any processing that cannot be done with L1C?

ABraun · January 23, 2020, 11:27am

as I said, the C2RCC needs L1C as an input, not L2A.
You can apply any processing to L1C products (as long as they are resampled to a common resolution)

Ceren · May 22, 2025, 5:35am

hi, i want to ask about C2RCC processor parameters how can i find ozone and air pressure at sea level values for Turkey?

Marco_EOM · May 25, 2025, 12:37pm

There are several sources. You could try the following

rosgo · May 25, 2025, 7:32pm

Trying to run C2X-Complex-Nets from Snappy, but it keeps throwing "RuntimeError: org.esa.snap.core.gpf.OperatorException: Operator ‘C2rccMsiOperator’: Value for ‘Set of neuronal nets’ is invalid: ‘C2X-COMPLEX-Nets’. Any idea why?

C2X-Nets and C2RCC-Nets do work.

Marco_EOM · May 26, 2025, 6:23am

That’s strange. In the code this is the expected name.

github.com

senbox-org/optical-toolbox/blob/6f50b6fdf92d1c4899cf25199829737e593cf991/opttbx-c2rcc/src/main/java/eu/esa/opt/c2rcc/msi/C2rccMsiOperator.java#L160


      
          static final String RASTER_NAME_SUN_ZENITH = "sun_zenith";
          static final String RASTER_NAME_SUN_AZIMUTH = "sun_azimuth";
          static final String RASTER_NAME_VIEW_ZENITH = "view_zenith_mean";
          static final String RASTER_NAME_VIEW_AZIMUTH = "view_azimuth_mean";
          static final String RASTER_NAME_AIR_PRESSURE = "msl";
          static final String RASTER_NAME_OZONE = "tco3";
          static final String RASTER_NAME_WATER_VAPOUR = "tcwv";
          
          private static final String STANDARD_NETS = "C2RCC-Nets";
          private static final String EXTREME_NETS = "C2X-Nets";
          private static final String COMPLEX_NETS = "C2X-COMPLEX-Nets";
          static final String DEFAULT_DEM_NAME = "Copernicus 90m Global DEM";
          static final String DEM_NAME_COPERNICUS30 = "Copernicus 30m Global DEM";
          static final String DEM_NAME_GETASSE30 = "GETASSE30";
          private static final Map<String, String[]> c2rccNetSetMap = new HashMap<>();
          
          static {
              c2rccNetSetMap.put(STANDARD_NETS, NnPaths.getStandard());
              c2rccNetSetMap.put(EXTREME_NETS, NnPaths.getExtreme());
              c2rccNetSetMap.put(COMPLEX_NETS, NnPaths.getComplex());
          }

Can you show your related code?

rosgo · May 26, 2025, 8:35am

I know, right? I got the parameter from both the source code and gpt -h. Here is my code, the problem is at the end:

import sys
import os

sys.path.append('/home/sergio/.snap/snap-python')

from esa_snappy import ProductIO, GPF, HashMap, WKTReader

# Directorio con archivos de entrada y de salida
input_directory = '/home/sergio/PycharmProjects/tfg/downloaded_SAFE'
output_directory = '/home/sergio/PycharmProjects/tfg/ProcesadosC2X-COMPLEX'

# El directorio existe
os.makedirs(output_directory, exist_ok=True)

# Coordenadas de la región de interés
wkt = 'POLYGON ((-0.915298 37.610152, -0.915298 37.835819, -0.671883 37.835819, -0.671883 37.610152, -0.915298 37.610152))'
geom = WKTReader().read(wkt)

# Obtener lista de archivos en el directorio de entrada
for file_name in os.listdir(input_directory):
    if file_name.endswith('.SAFE'):
        input_file = os.path.join(input_directory, file_name)
        output_file = os.path.join(output_directory, f'processed_{file_name}')

        print(f'Procesando: {input_file}')

        # Leer el producto de entrada
        source_product = ProductIO.readProduct(input_file)

        # Resampling
        parameters = HashMap()
        parameters.put('targetResolution', '20')        #TODO: si lo bajo a 10 pasa a tardar 1h/producto, pero podría mejorar
        parameters.put('geoRegion', geom)
        resampled_product = GPF.createProduct('Resample', parameters, source_product)

        # Recorte (que en principio ya está hecho, pero no sé si puede aplicar el resample a la vez que el ajuste de la zona)
        parameters = HashMap()
        parameters.put('geoRegion', geom)
        subset_product = GPF.createProduct('Subset', parameters, resampled_product)

        # Parámetros C2RCC
        parameters = HashMap()
        parameters.put('salinity', '35.0')   
        parameters.put('temperature', '15.0')
        parameters.put('ozone', '330')
        parameters.put('press', '1000')
        parameters.put('elevation', '0')
        parameters.put('TSMfac', '1.06')
        parameters.put('TSMexp', '0.942')
        parameters.put('CHLexp', '1.04')
        parameters.put('CHLfac', '21.0')
        parameters.put('outputAsRrs', 'false')
        parameters.put('outputRtoa', 'true')
        parameters.put('outputAcReflectance', 'true')
        parameters.put('outputRhown', 'true')
        parameters.put('outputKd', 'true')
        parameters.put('outputUncertainties', 'true')
        parameters.put('netSet', 'C2X-COMPLEX-Nets')        # Cambiando esto se aplican distintas redes neuronales: C2X-Nets, C2RCC-Nets

        # Aplicar el procesador C2RCC
        c2rcc_product = GPF.createProduct('c2rcc.msi', parameters, subset_product)

        # Guardar el producto
        ProductIO.writeProduct(c2rcc_product, output_file, 'BEAM-DIMAP')
        print(f'Archivo procesado guardado en: {output_file}\n')

print('Procesamiento completado.')```

Marco_EOM · May 26, 2025, 11:19am

From the GUI the parameter works. I currently have no snappy setup. So, I can’t reproduce this.
Have you tried str('C2X-COMPLEX-Nets')?
It should work without it, but who knows?

rosgo · May 26, 2025, 11:26am

Didn’t work, it outputs the same error. Do you know anyone else with a snappy setup that could try and reproduce the error? I really need to get this to work.

Marco_EOM · May 26, 2025, 11:32am

Maybe @dolaf or @TomBlock can try.

If as also made a pull request to enhance the error message.
If accepted, the error message will include the accepted values of the value set, like “Value for ‘X’ is invalid: ‘d’. Allowed values are: [a, b, c].”

Marco_EOM · May 30, 2025, 7:53am

It seems the devs do not have the time or don’t want to reply.
I have now set up a snappy environment for SNAP 12. I’ve used a miniconda environmment and Python 3.12.9.

The script worked. I’ve only changed the region and removed the geom from the resampling as it is not used by this operator.
Also, you should use the S2Resampling. It improves the results of the C2RCC. And if you not really need the uncertainties, disable them. They are computational intense.

Unfortunately, I can’t reproduce your error.

import os
from esa_snappy import ProductIO, GPF, HashMap, WKTReader

# Directorio con archivos de entrada y de salida
input_directory = 'D:\\EOData\\_temp\\c2rcc'
output_directory = 'D:\\EOData\\_temp\\c2rcc\\output'

# El directorio existe
os.makedirs(output_directory, exist_ok=True)

# Coordenadas de la región de interés
# wkt = 'POLYGON ((-0.915298 37.610152, -0.915298 37.835819, -0.671883 37.835819, -0.671883 37.610152, -0.915298 37.610152))'
wkt = 'POLYGON ((-127.46064834768475 49.835330370524886, -127.15343986701647 49.830842294687784,-127.16162293324992 49.614676446975885, -127.46747144562684 49.61913054266424,-127.46064834768475 49.835330370524886))'
geom = WKTReader().read(wkt)

# Obtener lista de archivos en el directorio de entrada
for file_name in os.listdir(input_directory):
    if file_name.endswith('.SAFE'):
        input_file = os.path.join(input_directory, file_name)
        output_file = os.path.join(output_directory, f'processed_{file_name}')

        print(f'Procesando: {input_file}')

        # Leer el producto de entrada
        source_product = ProductIO.readProduct(input_file)

        # Resampling
        parameters = HashMap()
        parameters.put('targetResolution', '20')  # TODO: si lo bajo a 10 pasa a tardar 1h/producto, pero podría mejorar
        # parameters.put('geoRegion', geom) # Does not have this parameter
        # better use S2Resampling - improves C2RCC results
        resampled_product = GPF.createProduct('Resample', parameters, source_product)

        # Recorte (que en principio ya está hecho, pero no sé si puede aplicar el resample a la vez que el ajuste de la zona)
        parameters = HashMap()
        # parameters.put('geoRegion', geom)
        subset_product = GPF.createProduct('Subset', parameters, resampled_product)

        # Parámetros C2RCC
        parameters = HashMap()
        parameters.put('salinity', '35.0')
        parameters.put('temperature', '15.0')
        parameters.put('ozone', '330')
        parameters.put('press', '1000')
        parameters.put('elevation', '0')
        parameters.put('TSMfac', '1.06')
        parameters.put('TSMexp', '0.942')
        parameters.put('CHLexp', '1.04')
        parameters.put('CHLfac', '21.0')
        parameters.put('outputAsRrs', 'false')
        parameters.put('outputRtoa', 'true')
        parameters.put('outputAcReflectance', 'true')
        parameters.put('outputRhown', 'true')
        parameters.put('outputKd', 'true')
        parameters.put('outputUncertainties', 'true')
        parameters.put('netSet',
                       'C2X-COMPLEX-Nets')  # Cambiando esto se aplican distintas redes neuronales: C2X-Nets, C2RCC-Nets

        # Aplicar el procesador C2RCC
        c2rcc_product = GPF.createProduct('c2rcc.msi', parameters, subset_product)

        # Guardar el producto
        ProductIO.writeProduct(c2rcc_product, output_file, 'BEAM-DIMAP')
        print(f'Archivo procesado guardado en: {output_file}\n')

print('Procesamiento completado.')

rosgo · May 30, 2025, 3:12pm

Dear Marco, thank you very much for the advice.

I just solved the problem by reinstalling the whole Sentinel 3 toolbox. I believe the error comes from installing an older version of the project. By following the instructions in s3tbx-c2rcc, I got misleaded by the “How to build” section, which holds a link to the deprecated C2RCC processor github page. This version of the processor happens to be older than the release of C2X-COMPLEX nets, thus not recognizing the appropriate parameter.

Again, thank you for your efforts and I apologize for making you lose time over this.

Marco_EOM · May 30, 2025, 4:08pm

No worries. Great that it now works for you.

Ceren · June 30, 2025, 6:05am

Thank you, i also wonder, is C2RCC working with S2-L2A (BOA)