Developement of exact radiative transfer methods Andreas Macke, Lüder von Bremen, Mario Schewski Institut für Meereskunde, Uni Kiel
Introduction ❑ Monte Carlo method exactly simulates radiative transfer prosesses in arbitrary complex scattering and absorbing media (clouds) ❑ Numerical application feasible for ❑ local and domain average ❑ fluxes and radiance fields at specified layers ❑ Full 3D radiance fields SHDOM (not exact)
Complete MC package (GRIMALDI) ❑ forward scheme to solve radiative transfer in 3d scattering and absorbing atmosphere with directed (solar) illumination ❑ preprocessor for absorption properties of atmospheric gases ❑ monochromatic and spectral band fluxes and radiances for finite sized angular bins ❑ photon path length pdf for finite sized angular bins ❑ data base for scattering phase functions and single scattering albedos for spherical and non-spherical cloud particles
Forward Local Estimate MC (MC-UNIK) ❑ forward scheme to solve radiative transfer in 3d scattering and absorbing atmosphere with directed (solar) illumination ❑ monochromatic fluxes and radiances for discrete directions (Local Estimate scheme) ❑ data base for scattering phase functions and single scattering albedos for spherical and non-spherical cloud particles
Backward Local Estimate MC (MC-UNIK-BW) ❑ backward scheme to solve radiative transfer in 3d scattering and absorbing atmosphere with directed (solar) illumination ❑ monochromatic radiances for discrete directions (Local Estimate scheme) ❑ photon pathlength pdf for predefined viewing geometries and viewing locations ❑ data base for scattering phase functions and single scattering albedos for spherical and non-spherical cloud particles
Backward MC (3RAD-UNIK) ❑ backward scheme to solve radiative transfer in 3d scattering, absorbing and emitting atmosphere ❑ monochromatic radiances for discrete directions ❑ preprocessor for gas absorption from the thermal to the microwave ❑ data base for scattering phase functions and single scattering albedos for spherical cloud particles
Internet Programs & Tools
Remote Sensing of inhomogeneous clouds - vertical structure, photon path length - Steffen Meyer, Mario Schewski, Andreas Macke Institut für Meereskunde, Uni Kiel
Problem - cloud variability breaks unique cloud-radiance relation -
MC-Local Estimate - discrete directions, spatial weighting function - ❑ Initialisation: ❑ scattering phase function ❑ single scattering albedo 0 ❑ vol. extinction coefficient x ❑ detector position ( ) ❑ output: ❑ reflected radiances L ❑ weighting function of L (x,y,z)
artificial vertical cloud profiles ❑ local variations of cloud extinction coefficient with preseved optical thickness
settings ❑ = 10, 20, 40 ❑ = 0.625; 0.850; 1.000; 1.600; m ❑ r eff = 16 m; variable ❑ surface albedo = 0.0 ❑ geometries: ❑ sun: = 60 0 ; 0 = 0 0 ❑ satellite: =0 0 ; =0 0 (nadir)
constant profile ( =10; r eff =16 m)
linear profile ( =10; r eff =16 m)
constant profile ( =10; r eff =var)
constant profile ( =40; r eff =16 m)
linear profile ( =40; r eff =16 m)
dependency of penetration depth on absorption single scattering albedo 0 wavelength [ m] single scattering albdo
spectral variation of scattering phase function and single scattering albedo cloud profile spectral weighting function
spectral variation of extinction coefficient percent of incoming radiation absorbed [ m] m m m m
settings ❑ = [0.9, 1.0, 0.025] m (b1) [1.4, 1.5, 0.025] m (b2) [1.95, 2.2, 0.025] m (b3) [2.8, 3.0, 0.025] m (b4) ❑ = 10.8 ❑ r eff = proportional to extinction coefficient ❑ surface albedo = 0.0 ❑ geometries: ❑ sun: = 60 0 ; = 0 0 ❑ satellite: =0 0 ; =0 0 (nadir)
Weighting function for gaussian profile of ext. coeff. and particle size ( =10.8; Q ext = 2.0)
extinction efficiency - size and wavelength dependency -
Weighting function for gaussian profile of ext. coeff. and particle size ( =10.8; Q ext = f(, r eff ))
Simulation of photon pathlength distributions - backward Monte Carlo for solar radiative transfer - ❑ Initialisation: ❑ scattering phase function ❑ single scattering albedo 0 ❑ vol. extinction coefficient x ❑ detector position ( ) ❑ detector location (x,y) ❑ output: ❑ transmitted radiances L ❑ photon pathlength distributions
example cloud
photon pathlength pdf
Summary and outlook ❑ MC methods developed, validated and made public available ❑ profiling of vertical cloud structure at solar wavelengths not feasible ❑ backward MC at solar wavelengths developed and validated ❑ photon pathlength pdf in progress ❑ simulation of radiance power spectra as a function of vertical and horizontal cloud variability ❑ more detailed consideration of cloud microphysical properties
The Future ❑ Proposal for a new research activity in support of 4DCLOUDS (year 4 & 5) ❑ cloud resolving models (dynamical, statistical) with explicit cloud microphysics as data base for cloud-radiation correlation
Correlation between domain averaged fluxes and cloud fields (parameterisation?)
Radiative transfer calculations ❑ Radiative fluxes calculated with MC-forward modell GRIMALDI ❑ Scattering and absorption at atmospheric gases and cloud particles ❑ 756 GESIMA clouds ❑ 52 x 52 x 26 grid boxes ❑ 2 km horizontal resolution ❑ 100 m – 1 km vertical resolution 10 km domain height ❑ 9 solar zenith angles (0°, 10°... 80°) ❑ 13 spectral bands
Spectral intervals
Scattering phase functions
Parameterization ❑ Multiple regression ❑ Nonlinear dependencies between cloud-parameters and radiative fluxes
Best fitting cloud parameters Cloud parameters used to obtain the individual radiative properties at solar zenith angle 0 = 50°:
Dependency on solar zenith angle
The End!
vertikale Profile GESIMA Wolke 'Issig' Wolke idealisierte Wolke
homogene Wolke ( =67.21)
Gauss-Wolke ( =1.0; =100.0 ) r eff =1.0
Gauss-Wolke ( =1.0; ) r eff =10.0
GESIMA ( =19.25) = 0.0 = 0.06
'Issig' Wolke
Gauss-Wolken ( =1.0; 100.0)(detector=bottom)
GESIMA ( =19.25) (detector=bottom)
Zusammenfassung I ❑ reflektierte Strahldichte am Satelliten stammt aus oberen Wolkenschichten ❑ Berücksichtigung versch. Wellenlängen führt zu keinem Gewinn an Information ❑ auch mit bodengebundener Fernerkundung keine Information über Profil
effektiver Radius Extinktionskoeffizient optische Dicke
Gauss-Issig ( =272) = 0.06 = 0.0 = 0.8
Gauss-Wolken (detector=bottom)
reste = Grass ?
'Issig' Wolke (detector=bottom)
Doppelschichtwolke