The evaluation of the effect of cloud inhomogeneities on their radiative properties is of substantial importance for a better understanding of clouds' critical contribution to the Earth's radiation budget. The solar radiative transfer in warm clouds is a complex function of the distribution of cloud liquid water and of the drop size. Even in stratiform clouds, cloud structure is typically remarkably inhomogeneous, and recent studies have shown that these inhomogeneities can significantly affect the bulk radiative properties of the clouds.

The objectives of this work are to make spectral measurements of short-wave radiances above stratocumulus cloud fields, together with detailed cloud microphysical observations, and subsequently to compare measured radiances to radiances modelled by using the microphysical and cloud structure information as input to sophisticated three-dimensional radiative transfer calculations.

The measurements were carried out onboard the Met Office C-130 Hercules aircraft over sea areas near the British Isles in February 2001. The clouds' microphysical characteristics (drop size distribution, LWC) and their structure (variations of LWC) were observed using a Fast FSSP and a Nevzorov probe. Spectral radiance measurements were done with a nadir looking Shortwave Spectrometer (manufactured by Zeiss/Germany) measuring in two wavelength ranges between 300 and 950 nm (resolution of 3 nm), and 950 and 1700 nm (resolution of 6 nm). The instrument allows a full spectral measurement with frequencies from 1 up to 10 Hz.

The 3D Spherical Harmonic Discrete Ordinate Method radiative transport code is used to calculate the reflected radiances from inhomogeneous stratiform cloud fields. The clouds, used as input to these calculations, are built up with (1) a 2D bounded cascade model based on the measured cloud microphysical statistics and (2) by using large-eddy simulated cloud fields.

The statistics of the measured and modelled radiances for horizontal flights above cloud are compared. It is shown that describing the cloud by using 2D-cascade modelling is not sufficient to reproduce the measured radiance variability above cloud, whereas using LES modelled clouds gives significantly improved results. Hereby, to bring modelled radiance variability into reasonable agreement with measured ones, an adequate representation of the cloud top structure is exposed to be of essential importance.