Iterative and constrained algorithms to generate cloud fields with measured properties Victor Venema Clemens Simmer Susanne Crewell Bonn University R eff Surrogate
Problem Radiative transfer through clouds –Validation, closure experiment –Retrievals and parameterisations Use measured cloud fields Use measured cloud properties
Perfectly fractal clouds Clouds are well described by fractal mathematics Scale free description Full power spectrum –Scale breaks –Waves Exact distribution
Amplitude distribution Amplitude (LWP, LWC, ) alone is already good: See Independent Pixel Approximation (IPA) Especially very important are the cloud free portions Together with power spectrum it also ‘defines’ the structure
Iterative algorithm (Schreiber and Schmitz)
Iterative algorithm Spectral adaptation –Calculate spectrum iterate time series –Replace magnitudes by those from the original time series –The phases are kept unaltered Amplitudes adaptation –By ranking –Replace values by the original values with same ranking –E.g. largest iterate value is replace by largest values of template
1D Iterative LWP surrogates
3D surrogate clouds Made surrogates routinely for the BBC campaign 2 3D-examples 2D LWP fields Time [hr] UT Height [km] LWC template [kg/m 3 ] LWC Surrogate Time [hr] UT Height [km] LWC template [kg/m 3 ] LWC Surrogate
Nonlinear cells – template (Schroeter and Raasch)
Nonlinear cells - surrogate
Nonlinear cells surrogate template
Nonlinearity testing Cells stratocumulus Fall streaks –Also in low LWP sections –Less clear in LWC fields Cloud top and base structure –Convergence Phase space of LWC (in situ)
Validation surrogate clouds 3D LWC fields from LES modelling –Cumulus: Brown et al., ARM –Stratocumulus: Duynkerke et al., FIRE Make surrogates from their statistics Calculate radiative properties Compare all pairs
Validation RadianceReflectance Stratocumulus Cumulus
Constrained surrogates Arbitrary constraints Evolutionary search algorithm Better convergence Try new statistics Fractal geometry for cloud boundaries
Evolutionary search algorithm
Constrained surrogates height profiles –cloud base –cloud top –cloud cover –average LWC Histograms –LWP –LWC –number of layers Power spectra & length –LWP –Highest cloud top –Lowest cloud base
Conclusions and outlook Iterative surrogate clouds have good radiative properties Generate 3D LWC field from a measurement Investigate which statistics are needed to describe structure Iterative wavelet surrogates Constrained surrogates to try different statistical properties –‘Fractal’ cloud boundaries –‘Multifractal’ liquid water –No periodic boundary conditions
Outlook Go from scanning measurement to Cartesian grid: fractal interpolation Anisotropic power spectrum More samples Better decorrelation
More information - Webpage Iterative algorithms (Matlab) Examples –Measurements –Theoretical conditions Articles in PDF victor/themes/surrogates/ Google: surrogate cloud fields