SSEC forums

Hello,

I'm now researching about the cloud masking techniques applied to Himawari-8 satellite images. I want to ask if there is a feature in Geo2Grid for cloud masking for Himawari-8 or anything related to it.

Thank you.

Hi,

Sorry for the late reply. This message slipped through the cracks.

Geo2Grid does not currently have any cloud masking functionality. In the future we may read level 2 products that could include a cloud mask, but do not currently have anything like this.

Dave

Hello, thanks for the reply.

My situation now is that if I want to apply algorithms in order to produce cloud mask, I want to apply it to the pixel values with the "right unit." Of course reflectance values does not have a unit, what I mean is the result from the ratio itself with decimals values, not the converted int8 (0-255) value.

Is there anyway to convert the image with the resulting pixel values not in the int8 format?

Thanks again.

Hi there,

You have two good options from what I can see:

1. Use the `--dtype float32` flag and the `--no-enhance` flag when generating geotiffs to produce floating point geotiffs with the real observed values. You should then be able to load these in to other programming languages as long as they support floating point geotiffs.
2. Use the Satpy library directly to load the data in Python and work with it there.

In the future Geo2Grid may have writers for data file formats (NetCDF4, HDF5, etc) but these are not implemented yet. These would be the best non-Python solution if they were available.

Dave

Oh, that's great, thank you. It seems that you already answered this from the last discussion, and I just remembered it, my bad.

My question now is would converting it to float32 affect the correction process for 'sunz_corrected' and 'rayleigh_scattering' we tried from the previous discussion or it applies the correction before converting it to int8 or float32?

Changing the saved data type should not affect any corrections applied. The general data flow is:

read -> resample -> corrections -> composite (ex. RGB) -> enhance/scale to fit "dtype" -> save to disk

Resampling may move after corrections or after composites depending what is being done. Compositing is only done if you are making a composite. Corrections are only done if they are asked for or are part of the composite being asked for. BUT enhancements and saving to disk are always at the end of this process. By doing `--no-enhance` you are turning off the scaling step and going straight to "save to disk'.

I hope this helps explain what's going on. Unless I'm misremembering what we worked on before, you should be fine saving as 32-bit floats.

Hello,

Is there a system for Geo2grip to hide the clouds for Himawari-8, I can't find the solution.
Thanks

Hello,

Geo2Grid does not currently have any cloud masking functionality. In the future we may read level 2 products that could include a cloud mask, but do not currently have anything like this.

Kathy