initial results of a new ice particle size parameterization in racmo
DESCRIPTION
Initial results of a new ice particle size parameterization in Racmo. Gerd-Jan van Zadelhoff. CloudNet : April 2005. Co-authors: Dave Donovan, Erik van Meijgaard. Observed R’ eff distributions. Cabauw. ARM. R ’ eff (T) not the same for the ARM and Cabauw sites. - PowerPoint PPT PresentationTRANSCRIPT
Initial results of a new ice particle size parameterization in Racmo.
CloudNet : April 2005
Gerd-Jan van Zadelhoff
Co-authors: Dave Donovan, Erik van Meijgaard
ARMCabauw
ObservedObserved R’R’effeff distributionsdistributions
-R’eff (T) not the same for the ARM and Cabauw sites.
- R’eff (Z) similar for Z < 3.5 at the ARM and Cabauw site.
Comparing mean profiles to normalized cloud thicknessComparing mean profiles to normalized cloud thickness
ARM
Cabauw
- Reff(Z, complex polycrystals) ARM-SGP & Cabauw results are the same within the errors.
Reff=A+B(Z/H)+C(Z/H)2
ARM
ARM Cabauw
C(H
)
B(H
)
A
(H)
Parameterization of Reff versus cloud thickness
- Reff(Z, complex polycrystals): Use parabolic fits to the observed values
Current Reff(T) relationship in Racmo vs. ObsCurrent Reff(T) relationship in Racmo vs. Obs
Mean Reff-T relationship observed, using Complex polycrystals as habit.
Current Racmo parameterization
Version 1: Reff parameterization & distribution.Version 1: Reff parameterization & distribution.
First version of the Reff parameterizations based on cloud-thickness. Above a threshold cloud-fraction of 0.1 maximum overlap is assumed.
Differences due to changes in parameterizationDifferences due to changes in parameterization
Differences in the mean net short wave flux at the surface and cloud cover.
Comparing the new parameterization with the current operational one.
Version 2: Reff parameterization & distribution.Version 2: Reff parameterization & distribution.
Assume random-maximum overlap and add fractional cloud-thickness according to the minimum cloud fraction.
Resulting RResulting Reffeff-T distribution due to Parameterization-T distribution due to Parameterization
The Reff-T distribution changes. Note however that the mean stays the same.
Work in progress: Work in progress:
Feed observed IWC profiles and the Reff parameterization to a single column radiation module. To check if the Cabauw radiation observations are similar to the calculated ones.
Current RACMO Reff > Observed Reff and misses a distribution.
As radiation of the models is reasonable tuned, the optical depth should be reasonable as well IWC/Reff IWC should be too large as well.
IWC(Z,T)
Example of radiation Example of radiation profiles and surface profiles and surface values for an values for an observed IWC profile observed IWC profile and a parameterized and a parameterized RReffeff..
There are large differences between the different calculated fluxes. Changing from Reff(T) to Reff(z) is the first step but getting the cloud thickness right is as important.
25 April 2004
In case of an optically thin ice cloud with low IWP there is not much difference between the different parameterizations, but in all cases the Reff(Z) seems to go in the right direction.
IWC(Z,T)
9 June 2004
19 July 2004
Conclusions…
The Reff has been parameterized according
to depth into cloud and total cloud thickness
resulting in large differences in SW-flux,
Cloud cover etc.
Using this parameterization the resulting
Reff-T distribution looks very similar to the
one observed.
The cloud-thickness calculated using the
Racmo vertical resolution results in too thick
clouds.
Using observed IWC profiles and Reff(z)
gives far better result to the observed SW-
flux compared to the original Reff(T)
parameterization.