parameterization of cloud in CAM3

[tao_zhang@noaa_gov]

Dear all,

I have a question about parameterization of cloud in CAM3.

I read the description of CAM3 and know that 3 types of cloud
are diagnosed: low-level marine stratus (Cst), convective cloud (Ccir)
and layered cloud (Cc).
It is said that the marine stratus cloud is diagnosed based on Klein
and Hartmann(1993) using an empirical relationship between the marine stratocumulus cloud fraction (Cst) and the potential temperature difference between 700mb and surface.

Then I use the 700mb and surface temperature output from CAM3 to
calculate the diagnosed cloud based on Klein and Hartmann(1993) scheme, I find the diagnosed cloud cover is very close to the low cloud cover which is directly output from CAM3.

Now my question is that I am not sure if the low cloud cover is perdicted
exactly by Klein and Hartmann(1993) scheme? If it is not, how the low cloud cover is calculated? What's relationship between the low cloud cover output from CAM3 and the diagnosed cloud cover based on Klein and Hartmann(1993) scheme? Does other types of cloud, such as convective cloud and layered cloud also have significant contributions to the low cloud cover?

Any suggestions are highly appreciated.

tz0616

 

[pjr]

Low cloud is not the same as stratocumulus for most parts of the world.

The klein hartman relationship is defined only over ocean, and it is only
non-zero in the presence of an inversion of the appropriate amplitude and
depth.

This cloud amount is merged with two other kinds of clouds:
1) relative humidity cloud
2) convective cloud
to make a variable that I will call CLOUD which represents the
total amount of cloud in any cell.

LOWCLD is an estimate of the amount of cloud that exists between the
surface and a particular pressure level. As such it is supposed to
represent a quantity that is represents more than one layer. You should
read the model description for the particulars.

Phil

 

[tao_zhang@noaa_gov]

Thanks for your nice reply.
The reason why I ask this question is that I want to figure out the
discrepancy between the low cloud from CAM3 and observation. I found that the simulated low cloud cover from CAM3 is not capable of simulating
the observed low cloud center in the eastern Pacific. I am not sure if this
discrepancy is mainly caused by the deficiency of Klein and Hartmann (1993) scheme, because the diagnosed cloud cover using this scheme is
very close to the low cloud cover output directly from CAM3 over the ocean. I don't know how to examine the contribution of two other clouds:
relative humidity cloud and convective cloud to the low cloud cover in the
model.
Do you have a better idea to examine the discrepancy between the low cloud from CAM3 and observation?
Thanks very much.

 

[pjr]

There are definite meteorlogical reasons for the problems with the
stratus in the eastern subtropical oceans. These have been document
in studies (which never made print) by Bjorn Stevens, Chris Bretherton,
and a variety of people here at NCAR. Some of the problems are
associated with the vertical resolution in the PBL (too low), and with the
rate of subsidence (too high), and with the formulation for the boundary layer itself (doesnt use moisture in the formulation), and with the
formulation for the clouds. It is a very difficult problem, and the more
people looking at it the better. We are hoping that some of these
issues can be significantly improved in the next generation model (post-cam3) which will have substantial changes to all the items mentioned above.

In answer to your question, you can turn on a couple of fields that will help you understand the issues. These are a variable called CLDST (which contains the clouds predicted by the K/L relationship) and CONCLD or CNVCLD (or both) which contain information about the convective clouds. The difference between the field called CLOUD and these two fields tell you what the relative humidity contribution is.

Phil

 

[tao_zhang@noaa_gov]

Dear phil,

Thanks your valuable suggestions. I learn a lot from you.
I still have 3 small questions.

1) I am not sure if in each model vertical layer, the field called CLOUD
is equal to max (Cc,Ccir,Cst).
Here Cc, Ccir, Cst is the relative humidity cloud, convective cloud and
stratus cloud.

2) Why the variable called CLDST (which contains the clouds predicted by the K/L relationship) has 26 layers in the vertical? Becuase according to
the formulation of Klein and Hartmann (1993),
that is, Cst=min{1.,max[0.,(theta700-theta_sfc)*0.057-0.5573]}
I only can get 2-D variable Cst, which is independent of the vertical layers?

3) I look at the output from CAM3 /JMCCAA/csm/eul64x128_d50amip/
but I can not find the variable CLDST (which contains the clouds predicted by the K/L relationship) and CNVCLD.
I don't where to get these variables, or I have to run the model by myself to turn on the output for these variables.

Thanks very much,
tao

 

[pjr]

Dear phil,

Thanks your valuable suggestions. I learn a lot from you.
I still have 3 small questions.

1) I am not sure if in each model vertical layer, the field called CLOUD
is equal to max (Cc,Ccir,Cst).
Here Cc, Ccir, Cst is the relative humidity cloud, convective cloud and
stratus cloud.

This should be documented in the manual. Otherwise you will need
to look at the code itself. My memory is that your expression above is correct.

2) Why the variable called CLDST (which contains the clouds predicted by the K/L relationship) has 26 layers in the vertical? Becuase according to
the formulation of Klein and Hartmann (1993),
that is, Cst=min{1.,max[0.,(theta700-theta_sfc)*0.057-0.5573]}
I only can get 2-D variable Cst, which is independent of the vertical layers?

The expression tells you the cloud fraction, but not the layer that the
cloud resides in. We choose the cloud to reside in the two layers below the inversion.

3) I look at the output from CAM3 /JMCCAA/csm/eul64x128_d50amip/
but I can not find the variable CLDST (which contains the clouds predicted by the K/L relationship) and CNVCLD.
I don't where to get these variables, or I have to run the model by myself to turn on the output for these variables.

You will need to run the model by yourself to turn these variable on.

Good luck

Phil