Products » Cloud Properties


OCRA (Optical Cloud Recognition Algorithm) and ROCINN (Retrieval of Cloud Information using Neural Networks) [Loyola et al. (2007) and Loyola et al. (2010)] are used for retrieving geophysical cloud properties from GOME-type sensors. OCRA uses the PMD devices on GOME to deliver the cloud fractions of the measured ground pixel. ROCINN takes the OCRA cloud fraction as input and uses a neural network scheme to invert GOME-type reflectivities in and around the O2 A band. These two algorithms work in tandem as showed in the following figure:

The cloud properties retrieved with OCRA and ROCINN from GOME-type sensors are:
• Cloud Fraction (Cloud Cover)
• Cloud-Top Pressure (Cloud-Top Height)
• Cloud Optical Thickness (Cloud-Top Albedo)

VLIDORT [Spurr (2006)] templates of reflectances based on full polarization scattering of light are used to train the neural network. ROCINN retrieves cloud-top pressure and cloud-top albedo.
The cloud-top pressure for GOME scenes is derived from the cloud-top height provided by ROCINN and an appropriated pressure profile. The cloud optical thickness is computed using libRadtran [Mayer and Kylling (2005)] radiative transfer simulations taking as input the cloud-top albedo retrieved with ROCINN.
A description of the OCRA and ROCINN algorithms and their validation is given in the peer-review papers listed below. For more details see the reference documents listed in the GOME documentation and GOME-2 documentation pages.
Remark: In the region of the Southern Atlantic Anomaly (SAA), the GOME-2 satellite instrument is subject to an increased particle flux which creates spurious signals in individual detectors resulting in a reduced quality of the retrieved GOME-2 trace gas columns, i.e. increased scatter in trace gas columns over South America and the Southern Atlantic.

Scientific Papers


Thanks to Rob Spurr (RTS) for the generation of VLIDORT templates of O2 A band reflectances used in ROCINN, to Werner Thomas (DWD) for the validation of OCRA/ROCINN with MSG data and to Bernhard Mayer (LMU) for the libRadtran simulations for the conversion of cloud-top albedo to cloud optical thickness.

For further information please contact:

Diego Loyola, DLR-IMF
E-Mail: Diego.Loyola [at]

DLR IMF AP Copyright 2010 | Imprint