Surrogate cloud fields generated with the iterative amplitude adapted Fourier transform algorithm

Abstract

A new method of generating two‐dimensional and three‐dimensional cloud fields is presented, which share several important statistical properties with real measured cloud fields. Well‐known algorithms such as the Fourier method and the Bounded Cascade method generate fields with a specified Fourier spectrum. The new iterative method allows for the specification of both the power spectrum and the amplitude distribution of the parameter of interest, e.g. the liquid water content or liquid water path. As such, the method is well suited to generate cloud fields based on measured data, and it is able to generate broken cloud fields. Important applications of such cloud fields are e.g. closure studies. The algorithm can be supplied with additional spatial constraints which can reduce the number of measured cases needed for such studies. In this study the suitability of the algorithm for radiative questions is evaluated by comparing the radiative properties of cloud fields from cloud resolving models of cumulus and stratocumulus with their surrogate fields at nadir, and for a solar zenith angle of 0° and 60°. The cumulus surrogate clouds ended up to be identical to the large eddy simulation (LES) clouds on which they are based, except for translations and reflections. The root mean square differences of the stratocumulus transmittance and reflectance fields are less than 0.03% of the radiative budget. The radiances and mean actinic fluxes fit better than 2%. These results demonstrate that these LES clouds are well described from a radiative point of view, using only a power spectrum together with an amplitude distribution.

A new algorithm for the downscaling of cloud fields

“…However, for the sparse cumulus clouds, this upper limit is likely not reachable: The Fourier spectrum of the original cloud contains information on the relative positions of the cloud clusters with respect to each other. For sparse cumulus clouds with many zero-valued pixels, this information is unambiguous enough to allow a very accurate reconstruction of the original cloud field; see also Venema et al (2006a). Because the exact position of the clouds is a property of a certain realisation, but the positions will vary for every ensemble member, no extrapolation algorithm based on ensemble statistics will be able to estimate the spectrum with this accuracy.…”

“…The 51 cumulus fields represent a diurnal cycle and were generated in the framework of the Atmospheric Radiation Measurement (ARM) project (Brown et al, 2002) and are also employed and described in more detail in Venema et al (2006a). The fields have a resolution of 100 m in the horizontal and 40 m in the vertical.…”

“…Coarse-resolution clouds will tend to have longer correlation lengths than higher-resolution clouds. Both horizontally uncorrelated fields (Venema et al, 2006a) and fields with a correlation length much larger than the cloud depth (Chambers et al, 1997;Davis et al, 1997) have a higher reflectance than clouds with a correlation length in the order of the cloud depth. In the latter case, photons will scatter preferentially towards regions with lower extinction, which increases the transparency of clouds.…”

“…The algorithm is able to handle any extrapolated power spectrum. The code is similarly organized as the Iterative Amplitude Adjusted Fourier Transform (IAAFT) algorithm used for the stochastic generation of surrogate fields (Venema et al, 2006a(Venema et al, , 2006cSchmidt et al, 2007) and can easily be combined with it. The downscaling algorithm iteratively adjusts the spectrum and the coarse fields.…”

Characteristics of cloud liquid water path from SEVIRI onboard the Meteosat Second Generation 2 satellite for several cloud types

“…With these measures, studies for a better description of LWP distribution in models under varying conditions, as for example performed by de Roode and Los (2008), are facilitated. Other possible applications are process studies or input data for cloud generators (Venema et al, 2006) and radiative transfer studies on a wide range of spatial scales. They can also provide verification in microphysical measurement experiments such as the airborne probing of clouds.…”