Conventions#

Structure of data_cube#

The data_cube (PicassoProc()) provides all the methods to do the processing as well as the data produces. Most importantly there are two variables for the averaged profiles retrievals_profile and for the high resolution time-height data retrievals_highres. Both contain dictionaries with the content outlined below.

data_cube.retrievals_profile#

Each key with retrieved profiles hold a list with all the data_cube.clFreeGrps containing subdictionaries with individual channels. An list of available retrievals is kept in

data_cube.retrievals_profile['avail_optical_profiles']
>>> ['klett', 'raman', 'klett_OC', 'raman_OC']

additionally the inputs for the retrievals (aggregated profiles) are also stored, e.g,

['RCS', 'sigBGCor', 'BG', 'refH', 'sigTCor', 'BGTCor', 'overlap', 'sigOLCor', 'BGOLCor']

data_cube.retrievals_highres#

Holds all the hight temporal-resolution data, including auxiliary variables like range, height, time, time64.

These are the inputs at various levels of processing:

['preproSignal', 'BG', 'sigBGCor', 'SNR', 'RCS', 'sigTCor', ...]

flags, such as:

['lowSNRMask', 'shutterOnMask', 'fogMask', 'mask607Off', 'mask387Off', 'mask407Off', 'mask1064_RROff', ...]

and geophysical retrievals:

['attBsc_532_total_FR', 'voldepol_355_total_FR', 'quasiBscV1_1064_total_FR', 'tcMaskV2']

A full list is available using data_cube.retrievals_highres.keys() The exact content depends on the usage of the collect_debug flag.

Channel naming#

Extensive discussion in Issue 303 of the matlab processing chain came to the conclusion to work with

{wavelength}_{total|cross|parallel|rotraman}_{NR|FR|DFOV}

For example

  • 355_total_FR

  • 532_cross_NR

  • 1064_total_FR

  • 532_cross_DFOV

  • …