CGILS (CFMIP-GCSS Intercomparison of Large-Eddy and Single-Column Models) 

 

 

 

Simulation Results For Participants

(Password Required 06/26/2011)

 

 

 

Update 5: results from 2nd round revised simulations  (06/26/2011)

 

 

Update 4: results from the revised simulations  (10/06/2010)

 

 

Update 3: forcing and standardized setup (7/20/2010)

 

 

Update 2 (8/14/2009)

(LES Specifications, LES Common Radiation Interface to the RRTM Radiation Code)

 

 

Update 1 (6/15/2009)

(Vancouver Meeting Summary and Action Items 6/15/2009)

 

 

Stony Brook Meeting Summary and Plan (3/11/2010)

Presentations (Password Required)

CGILS Meeting, March 1-2, 2010 at Stony Brook, Long Island, New York

 

 

 

 

 

 

Participating Groups as of October 6, 2010

Model (SCM, LES)

NCAR CAM3

NCAR CAM4

NCAR CAM5

CCC

CSIRO

ECHAM5-ETH

ECHAM6-MPI

UCLA/MPI-LES

ECMWF

GFDL

GFDL-LES

GISS

GSFC

JMA

KNMI-RACMO

KNMI-LES

LARC/UCLA-LES

LMD

SAM-LES

SNU

UKMO

UKMO-LES

SAM/UW LES

U. Wisconsin Madison

 

 

 

 

------------------------------------------------------------------------------------------------------------------------

 

Cass Specification for the CFMIP-GCSS Study of Cloud Feedback Mechanisms

by Using SCM/CRM/LES Models

 

Minghua Zhang, Chris Bretherton, Adrian Lock, Mark Webb,

Marat Khairoutdinov, and Anning Cheng

 

February 2008

Updated: March 25, 2009

 

Contact Information:

Minghua Zhang (mzhang@notes.cc.sunysb.edu), Address: 125 Endeavor, Stony Brook University, NY 11794-5000, Tel: 631-632-8318, Fax: 631-632-6251

 

Table of Contents

 

 

1.     Introduction and objective

2.     Large-scale forcing data

3.     Simulation instructions

4.     Output requirement for SCMs

a.      Set A for single-level fields

b.     Set B for multi-level fields

c.      Set C from parent GCMs if available

5.     Output requirement for LES and CRM

a.      Set A for single-level fields

b.     Set B for multi-level fields

6.     Model Description

7.     Time Table

8.     Submission Information

Appendix A: Header of the large-scale forcing netcdf files

Appendix B: Tables and plots of vertical profiles

Appendix C: Preliminary results

Appendix D: Description of the large-scale forcing construction

a.      Surface fields and atmospheric state variables

·       SST and SLP

·       Atmospheric temperature

·       Atmospheric water vapor

·       Atmospheric winds

b.     Vertical velocity and advective tendencies

·       Atmospheric radiative cooling

·       Vertical velocity

·       Horizontal advective tendencies in the free troposphere

·       Horizontal advective tendencies in the surface layer

c.      Summary of the forcing fields

References and Acknowledgements
 

1. Introduction and Objectives

 

Despite progresses in recent years to evaluate clouds in GCMs and to diagnose their climate feedbacks, there is still a general lack of knowledge on the physical mechanisms of cloud feedbacks in climate models and the causes of discrepancies among them.   Understanding the physical mechanism is necessary to improve our confidence on the climate sensitivity of the GCMs and thus their projection of future climate change on global and regional scales.

 

The difficulties of understanding cloud feedbacks in GCMs arise from several factors.  First, the transient and spatial variabilities of clouds are typically much larger than the small signal of cloud feedbacks.  Second, clouds are highly interactive with the atmospheric dynamical circulations which tend to hide the true physical mechanism of cloud feedbacks.  Third, clouds are intimately connected with all components of physical parameterizations that are highly interactive among themselves.

 

In view of these difficulties, CFMIP and the GCSS WG1 initiated a project to use idealized large-scale dynamical conditions to evaluate cloud feedback processes in GCMs.  The approach has several advantages.  It isolates the model physics from dynamics, thus dramatically simplifying the problem to a few selected locations.  Moreover, it allows the use of CRMs and LESs, which contain considerable more realistic description of subgrid scale processes in the GCMs, to study the same problem.   Furthermore, it can prescribe large-scale dynamical conditions to represent future climate which is not available in observations. 

 

The disadvantage of using idealized simulations is that the results cannot be compared with observations.  This disadvantage is alleviated if the idealized forcing for the control climate can capture the essential features of the observed large-scale condition.  This is aimed for in this CFMIP-GCSS project.  Wherever appropriate, the ECMWF analysis for July 2003 (courtesy of Martin Kohler) and GCM simulations from the NSF CPT project led by Chris Bretherton from the CAM3.1 and AM2 as well as simulations from CAM3.5 are consulted. The design is guided by emulating the large-scale forcing in the control and warmer climate in the GCMs that is independent of any physical parameterizations.

 

The objectives of the projects are:

 

(1)  To understand the physical mechanism of cloud feedbacks in GCMs through SCMs by using appropriately designed idealized large-scale forcing conditions.

(2)  To understand and evaluate low cloud processes in the SCMs and GCMs by using cloud-resolving and large-eddy models.

 

(go to top)


 

2. Large-scale forcing data

(http://atmgcm.msrc.sunysb.edu/cfmip/)

 

Large-scale SCM/CRM/LES forcing data have been constructed with the following two questions in mind:

 

(1)  How representative are the idealized forcing conditions to those in GCMs? 

(2)  Are cloud feedback processes in the SCMs using the idealized forcing the same as those in their parent GCMs? 

 

This project will focus on low clouds over the eastern oceans since these clouds have the largest impact on the radiative forcing and recent studies have suggested low clouds to be the main cause of model differences of cloud feedbacks (Bony and Dufresne, 2005).   In GCMs, even over the subsidence region of the atmospheric circulation, the large-scale forcing fields vary transiently and interactively with clouds.  To simplify the problem, however, we assume that the vertical velocity and horizontal advective tendencies of temperature and water vapor do not change with time.  Our preliminary experimentation has indicated that the constant forcing can be used to obtain similar cloud responses as those from transient forcing within a GCM over the eastern Pacific.

   

Responses of clouds to climate forcing from any GCM typically vary spatially even for their signs.  It is thus unlikely that the dynamical and thermodynamic condition at any single grid point can yield cloud feedbacks the same as in all GCMs.  We therefore constructed forcing conditions along the GPCI (GEWEX Pacific Cross Section Intercomparison) cross section (Siebesma et al. 2004), for July condition, over the northeast subtropical Pacific to sample different cloud regimes.   The purpose is not to reproduce the quantitative value of the global cloud feedbacks in the GCMs.  Instead, it is to use the cloud responses in the selected region to interpret the cloud feedback in the global models.  

 

The GPCI cross section is described by 13 grid boxes starting from (1oS, 173oW), with increments of 3 degrees in latitude and 4 degrees in longitude in the northeastward to the California coast at (35oN, 125oW).  For the first phase of this inter-comparison study, we will use three grid points:

 

(1)  S6 — at (17oN, 149oW), the middle point of the GPCI cross section, to represent the shallow cumulus regime;

(2)  S11 — at (32oN, 129oW), near the California coast, to represent the stratocumulus regime;

(3)  S12 — at (35oN, 125oW), immediately off the California coast, to represent the stratus regime

 

The development of the forcing data follows Zhang and Bretherton (2008) with modifications and detailed descriptions in the Appendix D. 

 

The following table lists some of the single-level fields in the control climate. In the perturbed climate, SST and surface air temperature are increased by 2 degrees.

 

 

Table 1: Specifications of variables at the three locations for the control climate

           

 

S6

Shallow Cu

S11

Stratocumulus

S12

Stratus

Latitude (Degrees North)

17oN

32oN

35oN

Longitude (Degrees)

149oW

129oW

125oW

SLP (mb)

1014.1

1020.8

1018.6

SST (oC)

25.6

19.3

17.8

Tair_surface (oC)

24.1

17.8

16.3

U_surface (m/s)

-7.4

-1.8

2.1

V_surface (m/s)

-2.7

-6.5

-8.0

RH_surface (m/s)

80%

80%

80%

Mean TOA insolation (w/m2)

448.1

471.5

473.1

Mean daytime solar zenith angle

51.0

52.0

52.7

Daytime fraction on July 15

0.539

0.580

0.590

Eccentricity on July 15

0.967

0.967

0.967

Surface Albedo

0.07

0.07

0.07

 

The forcing data (in netcdf format) can be downloaded from http://atmgcm.msrc.sunysb.edu/cfmip/. There are six data files in the folder: ctl represents control climate; p2k represents perturbed climate.  Each data file contains 4 identical time steps; the four steps are to facilitate the use in some SCMs.  Data from only one time step is needed.

 

Each data file contains the following variables:

 

Atmospheric state variables:          T, q, u, v, omega,p(lev)

Surface variables:                                            lat, lon, SST(Tg), Tsair, Ps, u_srf, v_srf, rh_srf

Horizontal advective tendencies:  divT, divq

Optional fields:                                : Vertical and total advective tendencies

: (vertdivT, vertdivq, divT3d, divq3d).

                                                             ( The total and vertical temp. tendency has included the adiabatic term)

                                                            : Diurnally averaged TOA insolation(solin)

                                                            : Daytime mean solar zenith angle, and fraction of daytime

: surface albedo (srf_alb), Ozone mixing ratio (o3mmr), lh, sh          

 

An example of a header file is in Appendix A.  The profile data for each of the six data files are listed in Appendix B, which also contains vertical profile plots of temperature, water vapor, winds, subsidence, and the horizontal advective tendencies at the three locations.

 

Data at other pressure or height levels should be linearly interpolated with pressure. 

 

(go to top)

 

 

3. Simulation Instruction

 

Number of simulations

 

Each model will carry out six simulations, one for control climate, one for perturbed climate, at each of the three grid points (S6, S11, S12).

 

Model setup

 

We ask for perpetual July 15 simulation with no diurnal cycle.  One can use either the latitude, longitude and date information to calculate the diurnally averaged incoming radiation, or use the TOA incoming shortwave radiation given in Table 1 (also in the files of the forcing data). Please choose whatever is convenient for you to calculate the shortwave heating rate without diurnal variation.  Either the daytime mean solar zenith angle and the fraction of daytime (in Table 1 or the forcing files), or the diurnally averaged solar zenith angle, on July 15 at each location, along with eccentricity (in table 1 or the forcing files) and solar constant of 1367 W/m2, can be used.  There are small differences between the two methods, but both are acceptable, since the vertical advective tendency will tend to reduce the impact of the differences on the simulations.

 

Ozone concentration at each grid point is given in the data, but you can use your own O3 concentration.

 

Latent and sensible heat fluxes should be calculated by the model with the given SST and winds, even though they are included in the files of the forcing data. The LES models should use their full radiation-transfer packages rather than the simplified formula-based flux radiation parameterizations in several previous GCSS cases. The domain top for accurate radiation transfer computations should be placed in the upper troposphere or lower stratosphere.

 

Vertical advective tendencies should be calculated by using the given pressure vertical velocity (omega) and the model simulated profiles of temperature and water vapor. The vertical advective tendencies included in the forcing files should not be used. A simple backward difference scheme in the vertical direction should be considered to produce smooth forcing profiles and to avoid numerical instability, e.g., current level subsidence times (state variable at a level above minus that at the current level) divided by the pressure difference.

 

Advective tendencies should be applied to the model at each time step.

 

For single column models, we suggest relaxation of the simulated temperature and water vapor above 400 mb to their initial values with relaxation time linearly decreasing from 1 day at 400 mb to 3 hours at 200 mb.

 

Integration duration

 

We ask that you run your SCMs or LES to quasi-equilibrium states.  Our experiences suggest that LESs may need 20 to 30 days to equilibrate, while SCMs may require more time.

 

We ask that you save hourly averaged data for the entire integration periods.

 

Grid size for LES

 

We leave it to participants to decide which domain size and resolution to use; however, we ask the participants to follow the following guidelines as closely as possible given their computational resources. The guidelines are mostly based on our preliminary LES results (see Appendix C) as well as previous GCSS BL-WG cases. We ask for the horizontal grid spacing of 50 m for the S11 and S12 cases, and 100 m for the S6. The horizontal domain size is to be about 4 km for the S11 and S12, and 8 km for the S6. The high vertical grid resolution in the vicinity of the inversion can be very important for the S11 and S12 cases; however, it could be difficult to predict the equilibrium cloud top height before the actual simulation is done, so a few trial runs may be needed. We ask for the grid spacing of 10 m or better in the vicinity of inversion for the S11 and S12 cases. For the BOMEX-like S6 case, the vertical resolution of 30 m or better below 4000m is requested.  Above the cloud layer, the vertical resolution can be gradually degraded to as low as 500 m. For more accurate radiation-transfer computations, the domain top is to be placed above 20 km. Boussinesq models should use the actual density variation with height when computing the radiative heating rates to maintain realistic free-tropospheric thermodynamic profiles.

 

Since this case study is for low clouds, one may try to set up the LES model top at 4 km with strong relaxation to the initial profiles of T and q at this level.  But we have not tested this yet.

 

(go to top)

 

 

4. Output fields for SCMs

Data should be saved either in netcdf format or ascii in (5E15.7).  Suggested names of the variables are in the bracket <X>, although we will take submissions with different names as long as they are clearly described. 

Some SCMs may not have all the output fields, or have different types of output diagnostics (such as liquid water static energy and total water instead of temperature and water vapor).  We will take whatever is feasible to you, but may come back to you asking for your assistance for converting your diagnostics to the common fields.

Set A: Hourly-averaged single-level fields as a function of time X(n_hours).

1.     Time <time> (in hour)

2.     Total cloud cover <cldtot>  

3.     Low-level cloud amount <cldlow> (if you don’t have low cloud amount, skip this).

4.     Vertically integrated liquid water <tglwp> (in kg/m^2)

5.     Precipitable water <precw> (in kg/m^2)

6.     Surface air temperature <tsair> in (K)

7.     Surface pressure <ps> in (mb)

8.     Convective precipitation <precc> in (mm/day)

9.     Stratiform precipitation <precl> in (mm/day)

10.  Total precipitation <prect> in (mm/day)

11.  Surface Latent heat flux <lh> in (W/m^2)

12.  Surface sensible heat flux <sh> in (W/m^2)

13.  PBL height <pblh> in (m)

14.  TOA SW net downward clear-sky radiation <fsntc> in (W/m^2)

15.  TOA SW net downward total-sky radiation <fsnt> in (W/m^2)

16.  TOA LW clear-sky upward radiation <flntc> in (W/m^2)

17.  TOA LW total-sky upward radiation <flnt> in (W/m^2)

18.  Surface SW net downward clear-sky radiation <fsnsc> in (W/m^2)

19.  Surface SW net downward total-sky radiation <fsns> in (W/m^2)

20.  Surface LW net upward clear-sky radiation <flnsc> in (W/m^2)

21.  Surface LW net upward total-sky radiation <flns> in (W/m^2)

Set B: Hourly-averaged vertical profiles of multi-level fields as a function of time, X(n_levels, n_hours).

22.  pressure <p> (in mb)

23.  temperature <T> (in K)

24.  water vapour mixing ratio <qv> (g/kg)

25.  liquid water mixing ratio <ql> (g/kg)

26.  cloud fraction <cloud>  

27.  updraft convective mass flux <mu> (kg m^-2 s^-1)

28.  downdraft convective mass flux <md> (kg m^-2 s^-1)

29.  dT/dt due to PBL-scheme <tdt_turb> (K/day)

30.  dT/dt due to large-scale condensation scheme <tdt_cond> (K/day)

31.  dT/dt due to shallow (or total if not separated) convection scheme <tdt_shal> (K/day)

32.  dT/dt due to deep (or total if not separated) convection scheme <tdt_deep> (K/day)

33.  dT/dt due to LW radiation <tdt_lw> (K/day)

34.  dT/dt due to SW radiation <tdt_sw> (K/day)

35.  dT/dt due to large-scale forcing <tdt_ls> (g/kg)/day) (for sanity check since vertical advective forcing is derived from the model)

36.  dqv/dt due to PBL-scheme <qdt_turb> ((g/kg)/day)

37.  dqv/dt due to large-scale condensation scheme <qdt_cond> ((g/kg)/day)

38.  dqv/dt due to shallow convection scheme <qdt_shal> ((g/kg)/day)

39.  dqv/dt due to deep convection scheme <qdt_deep> ((g/kg)/day)

40.  dqv/dt due to large-scale forcing <qdt_ls> ((g/kg)/day)

41.  dql/dt due to PBL-scheme <wdt_turb> ((g/kg)/day)

42.  dql/dt due to large-scale condensation scheme (c minus e) <wdt_cond> ((g/kg)/day)

43.  dql/dt due to shallow convection scheme <wdt_shal> ((g/kg)/day)

44.  dql/dt due to deep convection scheme <wdt_deep> ((g/kg)/day)

45.  dql/dt due to conversion to precipitation <wdt_prec> ((g/kg)/day)

46.  dql/dt due to cloud sedimentation <wdt_sedi> ((g/kg)/day)

Set C: Optional, information of the cloud feedbacks in your parent GCM. 

47.  Global cloud feedback from Cess experiment <dcrf> (W/m2/K)

48.  Global cloud feedback from Cess experiment for JJA <dcrf7> (W/m2/K)

49.  Horizonal distribution of LWCRF for control and perturbed climate in JJA <dlwcf7>

50.  Horizonal distribution of SWCRF for control and perturbed climate in JJA <dswcf7>

51.  Horizonal distribution of low cloud amount for control and perturbed climate in JJA <cldlow7c><cldlow7p>

52.  Horizonal distribution of total cloud amount for control and perturbed climate in JJA <cldtot7c><cldtot7p>

53.  Horizonal distribution of cloud liquid water path for control and perturbed climate  <cldlwp7c><cldlwp7p>

(go to top)

 

5. Output fields for LES and CRM

 

Data should be saved either in netcdf format or ascii in (5E15.7).  

 

The required output is in many ways similar to the BOMEX GCSS case study (Siebesma et al. 2003, http://www.convection.info/blclouds/) and the RICO case study (http://www.knmi.nl/samenw/rico/output3d.html), except for some variables to more directly evaluate SCMs and cloud feedbacks.   

Suggested names of the variables are in the bracket <X>, although we will take submissions with different names as long as they are clearly described. 

Some LES/CRMs may not have all the output fields, or have different types of output diagnostics (such as liquid water static energy and total water versus temperature and water vapor).  We will take whatever is feasible to you, but may come back to you asking for your assistance for converting your diagnostics to the common fields.

A cloudy grid cell is defined as one which contains cloud water qc>0.01 g/kg. Liquid water includes all (cloud and rain) condensed water (ql = qc + qr). A cloud core grid cell is defined as one which is cloudy and positively buoyant, i.e. &thetav is larger than the slab averaged &thetav.

Set A: Hourly-averaged single-level fields as a function of time X(n_hours).

1.     Simulation time <time> (hours)

2.     Fractional cloud cover <cldtot> (in percent), defined as the fraction of grid columns that contain cloud water.

3.     Vertical Integrated liquid water <tglwp> (in kg/m^2)

4.     Vertical Integrated water vapor <precw> (in kg/m^2)

5.     Vertical Integrated turbulent kinetic energy <tke> (in kg/s^2)

6.     Surface air temperature <tsair> in (K)

7.     Surface pressure <ps> in (mb)

8.     Total precipitation <prect> in (mm/day)

9.     Surface Latent heat flux <lh> in (W/m^2)

10.  Surface sensible heat flux <sh> in (W/m^2)

11.  TOA SW net downward clear-sky radiation <fsntc> in (W/m^2)

12.  TOA SW net downward total-sky radiation <fsnt> in (W/m^2)

13.  TOA LW clear-sky upward radiation <flntc> in (W/m^2)

14.  TOA LW total-sky upward radiation <flnt> in (W/m^2)

15.  Surface SW net downward clear-sky radiation <fsnsc> in (W/m^2)

16.  Surface SW net downward total-sky radiation <fsns> in (W/m^2)

17.  Surface LW net upward clear-sky radiation <flnsc> in (W/m^2)

18.  Surface LW net upward total-sky radiation <flns> in (W/m^2)

Set B: Hourly-averaged vertical profiles of multi-level fields as a function of time, X(n_levels, n_hours) if in ASCII format.

Set B.1 compares the hourly averaged vertical distribution in the simulation:

19.  Height  <h> (in m)

20.  Pressure <p> (in mb)

21.  potential temperature <theta> (in K)

22.  liquid water potential temperature <thl> (in K)

23.  water vapour mixing ratio <q_v> (g/kg)

24.  cloud liquid water mixing ratio <q_c> (g/kg)

25.  liquid water mixing ratio (cloud plus rain water) <q_l> (g/kg)

26.  fraction of cloudy grid points  (fraction of q_c> 0.01 g/kg) <a_cl> (0-1)

27.  reference density profile rho(z) <rho> (kg/m^3) (use mean density for compressible models)

Set B.2 compares the hourly averaged vertical distribution of turbulent velocity variances, skewness, vertical turbulent fluxes (resolved plus subgrid), and hourly tendencies:

28.  u'^2 + v'^2 <uvke> (in m^2/s^2)

29.  w'^2 <wke> (in m^2/s^2)

30.  Skewness profile <w'^3> / <w'^2>^(3/2) <skew>

31.  x-momentum flux u'w' (resolved plus sgs) <uw> (in 10^-2m^2/s^2)

32.  y-momentum flux v'w' (resolved plus sgs) <vw> (in 10^-2m^2/s^2)

33.  liquid water potential temperature flux (resolved plus sgs) w'thetal' <wthl> (10^-2 K m/s)

34.  total water flux w'qt' (resolved plus sgs) <wqt> (10^-5 m/s)

35.  liquid water flux w'ql' (resolved plus sgs) <wql> (10^-5 m/s))

36.  cloud liquid water flux w'qc' (resolved plus sgs) <wqc> (10^-5 m/s))

37.  virtual potential temperature flux w'theta_v' (resolved plus sgs) <wtv> (10^-2 K m/s))

38.  dtheta/dt due to net condensation <tht_cond> (K/day)

39.  dtheta/dt due to LW radiation <tht_lw> (K/day)

40.  dtheta/dt due to SW radiation <tht_sw> (K/day)

41.  dtheta/dt due to large-scale forcing <tht_ls> ((g/kg)/day)

42.  dqv/dt due to conversion to precipitation <qdt_pr> ((g/kg)/day)

43.  dqv/dt due to net condensation <qdt_conv> ((g/kg)/day)

44.  dqv/dt due to large-scale forcing <qdt_ls> ((g/kg)/day)

45.  dq_c/dt due to conversion to precipitation <wdt_pr> ((g/kg)/day)

46.  dqc/dt due to sedimentation <wdt_conv> ((g/kg)/day)

47.  dqc/dt due to large-scale forcing <wdt_ls> ((g/kg)/day)

48.  dT/dt due to large-scale forcing <tdt_ls> (g/kg)/day)

Set B.3 compares hourly vertical distributions of the resolved turbulent kinetic energy (in m^2/s^2) and its budget terms (in 10^-4 m^2/s^3):

49.  height (in meters), where the following variables locate <h>

50.  resolved turbulent kinetic energy <q'> = <0.5*(u'^2 + v'^2 +w'^2)> where u' = u - <u> etc. <tke>

51.  resolved shear production - (<u'w'>d<u>/dz + <v'w'>d<v>/dz)  <tke_sh>

52.  resolved buoyancy production <w'b'>  <tke_bu>

53.  resolved transport (turbulent plus pressure transport) <tke_tp>
-d/dz<w'(q'+p'/rho(z)) -u'*tau_13 - v'*tau_23 - w'*tau_33> <tke_tr>

54.  dissipation <tau_ij*du_i/dx_j> summed over i, j = 1,2,3.
Here u_i are the velocity components and tau_ij are the components of the subgridscale stress tensor. <tke_di>

55.  storage (i. e. (<resolved TKE>hour)- < resolved TKE>hour-1))/3600s) (in 10^-4 m^2/s^2) <tke_dt>

56.  residual (3) + (4) + (5) - (6) - (7) <tke_re>

The formulas above are correct for a Boussinesq fluid, but should be modified as needed to correspond to the equation set you are using.

Set B.4 compares the hourly vertical distribution of conditionally sampled cloud fields.

Remark: A cloudy grid cell is defined as one which contains cloud water qc>0.01 g/kg. The vertical velocity has to be taken relative to the specified large-scale subsidence so that <w>=0.

57.  height (in meters), where the following variables locate <h>

58.  fraction of cloudy grid points <a_c> (0-1)

59.  average over all cloudy grid points of vertical velocity <w_c> (m/s)

60.  average over all cloudy grid points of liquid water potential temperature <thl_c>(K)

61.  average over all cloudy grid points of total water content <qt_c> (g/kg)

62.  average over all cloudy grid points of total liquid water content <ql_c> (g/kg)

63.  average over all cloudy grid points of cloud liquid water content <qc_c> (g/kg)

64.  average over all cloudy grid points of virtual potential temperature <thv_c> (K)

Set B.5 compares the hourly and horizontally averaged covariances <wx>_cl for w and x=(theta_l, theta_v, q_t, q_l,q_c) for the conditionally sampled cloudy grid points.

65.  height (in meters), where the following variables locate,

66.  a * w_cl (m/s) <aw_c>

67.  a * wtheta_l_cl (K m/s) <awthl_c>

68.  a * wq_t_cl (g/kg m/s) <awqt_c>

69.  a * wq_l_cl (g/kg m/s) <awql_c>

70.  a * wq_c_cl (g/kg m/s) <awqc_c>

71.  a * wtheta_v_cl (K m/s) <awthv_c>

Set B.6 compares the hourly vertical distributions of conditionally sampled cloud core fields.

Remark: A cloud core grid cell is defined as one which is cloudy and positively buoyant, i.e. &thetav is larger than the slab averaged &thetav. The vertical velocity has to be taken relative to the specified large-scale subsidence so that <w>=0.

72.  height (in meters), where the following variables locate <h>

73.  fraction of cloudcore grid points <a_cc> (0-1)

74.  average over all cloudcore grid points of vertical velocity <w_cc> (m/s)

75.  average over all cloudcore grid points of liquid water potential temperature <thl_cc> (K)

76.  average over all cloudcore grid points of total water content <qt_cc> (g/kg)

77.  average over all cloudcore grid points of liquid water content <ql_cc> (g/kg)

78.  average over all cloudcore grid points of cloud liquid water content <qc_cc> (g/kg)

79.  average over all cloudcore grid points of virtual potential temperature <thv_cc> (K)

Set B.7 compares the hourly covariance profiles <wx>_core for w and x=(theta_l, theta_v, q_t, q_l, q_c) for the conditionally sampled cloudcore grid points.

80.  height (in meters), where the following variables locate,

81.  a * w_core (m/s) <aw_cc>

82.  a * wtheta_l_core (K m/s) <awthl_cc>

83.  a * wq_t_core (g/kg m/s) <awqt_cc>

84.  a * wq_l_core (g/kg m/s) <awql_cc>

85.  a * wq_c_core (g/kg m/s) <awqc_cc>

86.  a * wtheta_v_core (K m/s) <awthv_cc>

Set B.8 compares hourly and horizontally averaged "in_cloud" variances of some fields x= { theta_l, theta_v, q_t, q_l , q_c}.

These are defined as: <sig(x)>_cl = < (x - <x>_cl)^2 >_cl

87.  height (in meters), where the following variables locate <h>

88.  sig(theta_l_cl (K^2) <sig_thl>

89.  sig(q_t_cl (g/kg)^2 <sig_qt>

90.  sig(q_l_cl (g/kg)^2 <sig_ql>

91.  sig(q_c_cl (g/kg)^2 <sig_qc>

92.  sig(theta_v_cl (K^2) <sig_thv>

 

(go to top)

 

 


 

6. Model Description

 

|Scientist Name:
|Model Type (SCM, 2D cloud resolving, or 3D LES):
-----------------------------------------------------------------------
|(1) Integration time
|
|Control s6:

|Perturbed s6:

|Control s11:

|Perturbed s11:

|Control s12:

|Perturbed s12:

-----------------------------------------------------------------------
-----------------------------------------------------------------------
|(2) Numerical Domain
|
|Dimensions (fill in if different than specified in problem description):
|Domain size in x-direction:
|Domain size in y-direction:
|Domain size in z-direction:
|Number of grid points in x-direction:
|Number of grid points in y-direction:
|Number of grid points in z-direction:
|Grid size in x-direction:
|Grid size in y-direction:
|Grid size in z-direction:
|Time step for dynamics:
-----------------------------------------------------------------------
-----------------------------------------------------------------------
|(3) Numerical Technique (LES and CRM only)
|
|Numerical method (finite-difference, spectral, etc.):
|Advection scheme and its order of accuracy:
|Time scheme and its order of accuracy:
|Dynamical equations (elastic, anelastic, etc.):
|Lateral boundary conditions:
|Upper boundary condition:
|How was translation velocity of the reference frame chosen?
-----------------------------------------------------------------------
-----------------------------------------------------------------------
|(4) Physical Parameterizations
|
|Microphysical parameterizations:
|Drizzle/Rain parameterization:
|PBL parameterization (SCM):
|Cloud amount and macro-physical parameterization (SCM):
|Shallow convection (SCM):
|Deep convection parameterization (SCM):
|LW radiation:

|SW radiation:

|Cloud Feedback in parent GCM (LW, SW and Net) if known:
|Other Comments:
-----------------------------------------------------------------------
-----------------------------------------------------------------------
|(5) Turbulence Closure Scheme (LES, if applicable)
|
|Turbulence closure (SGS or ensemble mean):
|Specific turbulence closure type (Mellor-Yamada, etc.):
|Variables predicted by the turbulence closure (eddy diffusivities, turbulent kinetic |energy, etc.):
|Closure for dissipation rate:
|Closure for turbulent dissipation length scale:
|Closure for turbulent diffusion length scale:
-----------------------------------------------------------------------
-----------------------------------------------------------------------
|(6) Documentation
|
|Please provide references that more fully describe your model, if available.

 

(go to top)

 

 

 

7. Time Table

 

 

1.     March 15, 2009:  Comments due on the case specifications

2.     April 30, 2009:     Submission of results

3.     May 15, 2009:      Preliminary analysis completed

4.     May 31, 2009:      Comments on preliminary results due

5.     June 8, 2009:        CFMIP-GCSS meeting

6.     December 30, 2009:         Wrap-up and publication preparations

 

(go to top)

 

8. Submission

 

 

Please email Minghua Zhang (mzhang@notes.cc.sunysb.edu) to

1.     Express your interest to participate

2.     Ask any questions concerning this case study

3.     Arrange submission of results

 

(go to top)

 

 


Appendix A: An example of the header of the netcdf forcing data file

 

netcdf ctl_s6 {

dimensions:

      lat = 1 ;

      lon = 1 ;

      lev = 26 ;

      time = 4 ;

variables:

      int bdate ;

            bdate:long_name = "baseline date" ;

      float phis ;

            phis:long_name = "Surface geopotential" ;

            phis:units = "m2/s2" ;

      float lon(lon) ;

            lon:long_name = "longitude " ;

            lon:units = "degrees E" ;

      float lat(lat) ;

            lat:long_name = "latitude" ;

            lat:units = "degrees N" ;

      float lev(lev) ;

            lev:long_name = " pressure levels" ;

            lev:units = "Pa" ;

      int tsec(time) ;

            tsec:long_name = "time " ;

            tsec:units = "seconds" ;

      float lhflx(time, lat, lon) ;

            lhflx:long_name = "Surface latent heat flux" ;

            lhflx:units = "W/m2" ;

      float shflx(time, lat, lon) ;

            shflx:long_name = "Surface sensible heat flux" ;

            shflx:units = "W/m2" ;

      float Ps(time, lat, lon) ;

            Ps:long_name = "Surface pressure" ;

            Ps:units = "Pa" ;

      float Ptend(time, lat, lon) ;

            Ptend:long_name = "Surface pressure tendency" ;

            Ptend:units = "Pa/s" ;

      float Tsair(time, lat, lon) ;

            Tsair:long_name = "Surface air temperature" ;

            Tsair:units = "K" ;

      float Tg(time, lat, lon) ;

            Tg:long_name = "Surface temperature" ;

            Tg:units = "K" ;

      float solin(time, lat, lon) ;

            solin:long_name = "TOA Insolation" ;

            solin:units = "W/m2" ;

      float zenith(time, lat, lon) ;

            zenith:long_name = "Solar zenith angle" ;

            zenith:units = "Degrees" ;

      float dayfrac(time, lat, lon) ;

            dayfrac:long_name = "Fraction of daytime" ;

            dayfrac:units = "Fraction" ;

      float delta(time, lat, lon) ;

            delta:long_name = "Solar declination angle" ;

            delta:units = "Degrees" ;

      float ecc(time, lat, lon) ;

            ecc:long_name = "Eccentricity" ;

            ecc:units = "dimensionless" ;

      float u_srf(time, lat, lon) ;

            u_srf:long_name = "u-component wind at surface" ;

            u_srf:units = "W/m2" ;

      float v_srf(time, lat, lon) ;

            v_srf:long_name = "v-component wind at surface" ;

            v_srf:units = "m/s" ;

      float rh_srf(time, lat, lon) ;

            rh_srf:long_name = "Surface air relative humidity" ;

            rh_srf:units = " " ;

      float srf_alb(time, lat, lon) ;

            srf_alb:long_name = "Surface albedo" ;

            srf_alb:units = " " ;

      float T(time, lev, lat, lon) ;

            T:long_name = "Temperature" ;

            T:units = "K" ;

      float q(time, lev, lat, lon) ;

            q:long_name = "Water vapor mixing ratio" ;

            q:units = "kg/kg" ;

      float u(time, lev, lat, lon) ;

            u:long_name = "u wind" ;

            u:units = "m/s" ;

      float v(time, lev, lat, lon) ;

            v:long_name = "v wind" ;

            v:units = "m/s" ;

      float omega(time, lev, lat, lon) ;

            omega:long_name = "Vertical pressure velocity" ;

            omega:units = "Pa/s" ;

      float divT(time, lev, lat, lon) ;

            divT:long_name = "Horizontal large scale temp. forcing" ;

            divT:units = "K/s" ;

      float divq(time, lev, lat, lon) ;

            divq:long_name = "Horizontal large scale water vapor forcing" ;

            divq:units = "kg/kg/s" ;

      float vertdivT(time, lev, lat, lon) ;

            vertdivT:long_name = "Vertcal large scale temp. forcing" ;

            vertdivT:units = "K/s" ;

      float vertdivq(time, lev, lat, lon) ;

            vertdivq:long_name = "Vertical large scale water vapor forcing" ;

            vertdivq:units = "kg/kg/s" ;

      float divT3d(time, lev, lat, lon) ;

            divT3d:long_name = "3d large scale temp. forcing" ;

            divT3d:unTts = "K/s" ;

      float divq3d(time, lev, lat, lon) ;

            divq3d:long_name = "3d large scale water vapor forcing" ;

            divq3d:units = "kg/kg/s" ;

      float div(time, lev, lat, lon) ;

            div:long_name = "Large scale horizontal divergence" ;

            div:units = "1/s" ;

      float o3mmr(time, lev, lat, lon) ;

            o3mmr:long_name = "O3 mixing ratio" ;

            o3mmr:units = "kg/kg" ;

 

// global attributes:

            :Title = "IOP Analysis" ;

            :time_step_length = " " ;

data:

 

 bdate = 20030715 ;

 

 phis = 0 ;

 

 lon = 211 ;

 

 lat = 17 ;

 

 

 

(go to top)

 


Appendix B: Data Tables

 

(use netcdf files at http://atmgcm.msrc.sunysb.edu/cfmip to read)

 

Table B.1 Vertical profiles for CTL_S6

 

 

Table B.2 Vertical profiles for P2K_S6

 

 

Table B.3 Vertical profiles for CTL_S11

 

 

Table B.4 Vertical profiles for P2K_S11

 

 

 

Table B.5 Vertical profiles for CTL_S12

 

 

 

Table B.6 Vertical profiles for P2K_S12

 


Figure B.1.  Vertical profiles of initial temperature and water vapor in the control climate, and the two wind components, at the three locations of S6, S11, and S12.

 

 

 

 


Figure B.2.  Vertical profiles of subsidence in the control climate, subsidence in the control and perturbed climate, and horizontal advective tendencies of temperature and water vapor, at the three locations of S6, S11, and S12.

 

 

 

(go to top)

 

 


Appendix C. Preliminary Results

 

C.1 UCLA LES from Anning Cheng and Kuan-man Xu.

 

The following figure shows cloud fraction in the control (top) and warmer (bottom) climate at the three locations (s6, s11, s12) from the UCLA LES. 

 

Shallow cumulus, stratocumulus and stratus are simulated at the three locations. Negative cloud feedback is simulated at all locations.

 

cld6_11_12

 

The LES domain size is 6.4 km by 6.4 km by 12.8 km in horizontal and vertical directions, with grid spacings of 200 m in horizontal, and 30 m near surface stretching to 90 m at 2 km,  to 156 m at 4 km, and to 350 m at 12 km in vertical directions, respectively. The integration time is 30 days. 

 


 

 

C.2. SAM results from Marat Khairoutdinov

 

The following figure shows cloud fraction in the control (top) and warmer (bottom) climate at the three locations (s6, s11, s12) from the SAM LES.  Negative cloud feedback is simulated at all locations.

 

 

The LES domain size is 6.4 km by 6.4 km by 20 km in horizontal and vertical directions, with grid spacings of 200 m in horizontal, and 30 m near surface stretching to 90 m at 2 km,  to 156 m at 4 km, and to 350 m at 12 km in vertical directions, respectively. The integration time is 30 days.  For S6 and S11, a diurnal cycle was included in the simulation.

 


C.3 NCAR CAM3.5 SCM

 

The following figure shows cloud fraction in the control (top) and warmer (bottom) climate at the three locations (s6, s11, s12) from the CAM3.5 SCM.  Negative cloud feedback is simulated at S11 and S12; small positive cloud feedback is simulated at S6.  The GCM has a negative cloud feedback.

 

 

 

 

 

(go to top)


 

Appendix D: Derivation of the large-scale forcing data

 

As in Zhang and Bretherton (2008), large-scale subsidence and advective forcing of temperature and moisture are calculated based on the heat and moisture budget of the clear-sky atmosphere in both the control and perturbed climate for July conditions.  The present study expands ZB08 in the following two aspects:

 

(1)  Horizontal advective tendencies of temperature and moisture below 950 mb are calculated based on surface winds and SST gradient. In the case of moisture, a surface relative humidity of 80% is assumed.  The calculated values are close to those in GCMs as will be shown later. 

 

(2)  The large-scale subsidence at the first grid point off the California coast is derived by accounting for horizontal advective cooling in the lower troposphere in addition to radiative cooling.  In ZB08, the balance is only between subsidence warming and radiative cooling.  This modification results in more realistic rate of lager-scale subsidence. 

 

There are some other differences between the present design and that in ZB08.  These are described below.  The forcing data are derived for the entire GPCI cross section (Figure 1), but only 3 locations are used in this study (S6, S11 and S12)

 

 

D.1 Surface and atmospheric state variables

 

SST and surface pressure

 

SST in the control climate at the thirteen grid points is specified in Celsius as follow: 

 

 

where x is the latitude.  SST is the only variable to be perturbed to represent a climate change. In the present study, SST is uniformly raised by 2 degrees to represent a warmer climate.  Figure 2a compares the above SST (red) with those from the ECMWF analysis along the cross section.

 

Surface pressure is specified in mb as a function of latitude as

 

 

This fits well to the observed surface pressure along the GPCI cross section for July in the ECMWF analysis as shown in Figure 2b.  SLP is considered as the same for the control and perturbed climate.

 

Atmospheric temperature

 

Surface air at the ITCZ latitude of 8oN is assumed to be 1.5 oC cooler than the SST and with 80% relative humidity, which approximately represent conditions in the ECMWF analysis.  Atmospheric temperature at this latitude is calculated by following a dry adiabat and then a moist adiabat up to the tropopause as in ZB08.  Above the tropopause, the temperature is specified as in the ECMWF analysis.  Since the atmospheric lapse rate in the middle troposphere is typically in between the moist and dry adiabat, the magnitude of the moist lapse rate is increased by 8% to match observations.  In the upper troposphere above 500mb, the temperature lapse rate is set to be larger than -9.5 oC /km to prevent convective instability.  The temperature profile thus constructed is shown in Figure 3a in red along with comparison with the ECMWF analysis.

 

ZB08 used the weak temperature gradient approximation to assume that free tropospheric temperature in the subtropics is the same as near the equator.  In reality, free tropospheric temperature decreases toward the subtropics by about 5 degrees to 32oN at 400 mb.  This is accounted by constructing an adiabat at 32oN by increasing the magnitude of the moist lapse rate from 8% to 18% with the same surface condition as at 8oN.  The constructed temperature profile above 850 mb is shown in Figure 3b as the red line, along with temperature in the ECMWF analysis. 

 

Tropospheric temperature at latitudes higher than 8oN is calculated by linearly interpolating the profile at 8oN and 32oN based on latitude.  South of 8oN, temperature is kept the same as at 8oN.  To improve the representation of radiative cooling in GCMs, atmospheric temperature below 850 mb at each grid is linearly interpolated vertically using temperature at 850 mb and the surface air temperature, which is assumed to be 1.5oC colder than SST at all locations. 

 

Figure 4a shows the temperature field on the latitude-pressure cross section.  The difference with ECMWF analysis is shown in Figure 4b, which is in general less than 1.5 degrees in the free troposphere.  Near the surface, the idealized temperature is higher than the ECMWF analysis, which is expected to be cooled by the development of a boundary layer in the model.  

 

The temperature field in the perturbed climate is calculated by using the same procedure except that the SST is increased by 2 degrees everywhere.  The temperature in the stratosphere for the perturbed climate follows ZB08 in which the temperature increment at the tropopause is linearly interpolated upward with zero change at 50 mb.

 

Figure 5a shows the temperature difference between the perturbed climate and the control climate.  This is compared with those in the CAM3.5 and AM2 from their DSST=2K simulations in Figures 5b and 5c.  It is seen that the temperature change differs among the model, which is the reason why an idealized case is proposed in this study.

 

Atmospheric water vapor

 

As in ZB08, relative humidity (RH) is specified and is assumed to be the same in the control and perturbed climate.  Unlike ZB08 in which a constant RH of 15% is used in the free troposphere, the present study assumes a profile of RH that maximizes at 950 mb and near the tropopause, and minimizes in the middle troposphere.  The profiles at three latitudes in the ECMWF analysis are shown in Figure 6.  We assume 95% of RH at 950 mb at all locations.  Between 950 mb and the surface, it is linearly interpolated to 80% at the surface.  The RH maximum near the tropopause and the minimum in the mid-troposphere are assumed to change with latitude as in Figure 7a.  The pressures of the two extremes change with latitude as in Figure 7b.  The RH at 950 mb and the mid-tropospheric minimum are used to vertically interpolate the RH in between.  Above the mid-tropospheric minimum to 300 mb, it is assumed to be constant.  Above 300 mb, it is linearly interpolated to the tropopause maximum.  Above the tropopause, a fixed mixing ratio of 4 ppmv is used.  The specifications described above are intended to provide a more realistic estimation of the radiative cooling than what can be obtained from one fixed relative humidity for the entire troposphere.

 

Figures 8a to 8c show a comparison of the constructed RH distribution with those in the ECMWF analysis and in CAM3.5.   

 

The relative humidity and temperature fields are then used to obtain the water vapor mixing ratio in both the control and perturbed climate for subsequent radiative calculations.  Figure 9a shows the water vapor distribution in the control climate.  Figures 9b-9c show the corresponding field in the ECMWF analysis and in CAM3.5.  The water vapor concentration is expected to change when clouds develop.

 

Atmospheric winds and surface fluxes

 

The atmospheric wind components u, v are taken from the mean fields in the ECMWF analysis and are assumed to be the same for the control and perturbed climate.  They are shown in Figure 10.  For the northern part of the cross section, the dominant features are the northeasterly in the lower troposphere associated with the subtropical high over the region, and the southwesterly in the upper troposphere associated with the North America summer monsoon high.  For the southern half of the cross section, easterly prevail.   These wind features are also grossly simulated in GCMs.

 

Surface winds are shown in Figure 11.  They are used in the calculation of horizontal advections of temperature and water vapor near the surface.  They are also used to calculate the surface latent and sensible heat fluxes with an aerodynamic coefficient of 0.001 s/m2 and surface relative humidity of 80% when needed.

 

(go to top)

 

D.2 Vertical velocity and forcing fields

 

Atmospheric radiative cooling

 

Insolation is fixed at July 15 with no diurnal variation.  The solar constant is taken as 1367 W/m2.  Figure 12 shows the diurnally averaged TOA insolation as a function of latitude.  Surface albedo is specified as 0.07.  Ozone profiles are interpolated based on latitude by using tropical and summer middle latitude profiles in the standard atmosphere that are assumed to be at the equator and 30oN. 

 

The rates of clear-sky infrared radiative cooling and heating from solar radiation are shown in Figure 13a and 13b for the control climate.  Over the bulk of the troposphere, the radiative cooling is between 1.5 K/day and 2.5 K/day; the shortwave warming is about 1 K/day.  

 

The difference of radiative cooling between the perturbed climate and the control climate is shown in Figure 14.  As pointed in ZB08, the warmer climate is with appreciable larger infrared cooling only at the emission level in the upper troposphere and near the boundary layer.  In the bulk of the free troposphere, the impact of increased temperature on radiation is offset by the increased amount of water vapor greenhouse effect aloft.  This feature is generally consistent with what is in the CAM3.5 (Figure 15).

 

Atmospheric vertical velocity

 

In the subsidence region, the vertical velocity is calculated from the steady-state thermodynamic equation by using the net clear-sky radiative cooling:

 

                                                                           

 

 As in ZB08, the shape of the vertical profile of is prescribed to peak at 800 mb as

 

 

where pmax is equal to 800 mb. For the first grid point at 35oN, where clear-sky condition prevails in July, the amplitude of ω is calculated from:

 

                                                        

 

The integration is from 300 mb to 900 mb.  This choice is to minimize the impact of boundary layer clouds.  The mean horizontal advective cooling rate is set as -2.3 K/day at (35oN,125oW) which is based on values in the ECMWF analysis and in several GCMs.  This horizontal advection is considered as the same for the control and perturbed climate.

 

In the calculation of vertical velocity, the clear-sky radiative cooling assumes no clouds in the boundary layer, which leads to an underestimation of the subsidence rate.  Because only integrated cooling rate is used, and a 95% relative humidity is used at 950 mb, the additional impact of clouds in the PBL on QR above 900 mb is found to be small.  The integrated radiative cooling rates for the clear-sky and total-sky atmosphere in the CAM3.5 differ only by 0.01 K/day.

 

The vertical profile of ω thus calculated for the control climate is shown in Figure 16a.  This is compared with those in the ECMWF analysis in Figure 16b (blue). 

 

For the perturbed climate at the same station, vertical profile of ω corresponding to its own radiative cooling and vertical stratification is shown in red in Figure 17a along with ω in the control climate.  The weaker subsidence in perturbed climate is primarily due to large vertical stability, much less due to increased radiative cooling.  As a comparison, the ω profiles in the CAM3.5 and AM2 form the control and perturbed climate are shown in Figures 17b and 17c.  The response in the vertical velocity differs somewhat in the models which are also associated with difference in their cloud feedbacks.

 

For the other grid points, the ω profiles are scaled against the subsidence profile at the first station with a fixed set of scaling factors.  This choice is more of a necessity than a physical cause since otherwise the horizontal advection tendencies at other stations would be needed, and for regions of upward motion the clear-sky thermodynamic equation is no longer valid.  Our guiding principle is to emulate the conditions in the GCMs.  The set of the scaling factors is based on the ECMWF analysis and is given in Figure 18.  This treatment assumes that circulation conditions at the cross section vary linearly as a single system.

 

Figure 19a shows the distribution of vertical velocity in the control climate.  This is compared with the ECMWF analysis in Figure 19b.  Results from CAM3.5 and AM2 are shown in Figure 19c-19d.  There are two latitudes with maximum subsidence.  The stronger one is near the coast that is associated with the subtropical high.  The weaker one is at around 20oN that is likely associated with compensating subsidence of the rising motion in the ITCZ.

 

The distribution of vertical velocity in the perturbed climate is shown in Figure 20a.  There is no corresponding figure in the reanalysis.  Figures 20b-20c show those in the CAM3.5 and AM2.  It is seen that the idealized field captures the general weakening of both the subsidence and upward motion in the perturbed climate relative to the control climate, but the model circulations differ from each other as expected.

 

Advective tendencies in the free troposphere

 

In the free troposphere for regions of subsidence, the vertical profiles of the horizontal advective tendencies of temperature and water vapor are calculated from the clear-sky thermodynamic and water vapor mass continuity equations as:

 

                                                                          

                                                                                                                         

In the free troposphere for regions of upward motion, the horizontal advective tendencies of temperature and water vapor are set to zero.

 

Advective tendencies in the surface layer below 950 mb

 

In the surface layer, the advective tendency of temperature is calculated from the surface winds and SST gradient as

 

            

 

where l is along the direction of the cross section pointing toward southwest.  Since the direction of the horizontal temperature gradient is unknown in the idealized setup, we assume that the surface wind is along the direction of the cross section.  The surface advective tendency of temperature is plotted in Figure 21a as a function of latitude.  It is compared with the ECMWF analysis (red) in Figure 21b, and in the CAM3.5 (red) as well as AM2 (blue) in Figure 21c.  This surface tendency does not change in the perturbed climate since SST is uniformly increased by 2K in the warmer climate.

 

The horizontal advective tendency of water vapor at the surface is calculated from

 

                                                            

 

where RHs is the surface relative humidity taken as 80% and qs is the saturation mixing ratio at the surface.  The calculated tendency is plotted in Figure 22a.  A comparison with those in the ECMWF analysis (red) is in Figure 22b.  Figure 22c compares it with those in the CAM3.5 (red) and AM2 (blue). Unlike the temperature tendency, the water vapor tendency differs slightly between the control and perturbed climate, which is also the case in the GCMs.

 

The horizontal advective tendencies in the atmosphere below 950 mb are assumed to be the same as those at the surface.  Between 850 mb and 950 mb, they are linearly interpolated vertically based on pressure.

 

 Figures 23a and 23b show the horizontal and vertical advections of temperature (including the adiabatic compression and expansion term) for the control climate in the idealized forcing.  This is compared with the ECMWF analysis in Figures 24a and 24b.

 

The idealized forcing captures the main features of the ECMWF analysis.  The exception is the feature near the top of the PBL in the ECMWF analysis that is a result of the model boundary layer clouds.  This is an undesirable feature to include in the idealized forcing.  An SCM or CRM will generate its own vertical temperature gradient once clouds are formed.

 

Figure 25 and Figure 26 show the same variables in CAM3.5 and AM2.  They all compare qualitatively well with the idealized forcing.

 

The horizontal and vertical advective tendencies of water vapor in the idealized forcing are shown in Figures 27a and 27b.  The corresponding values in the ECMWF analysis are shown in Figure 28.  Those in CAM3.5 and AM2 are in shown Figure 29 and Figure 30.  Again, the idealized forcing has captured the gross features in the ECMWF and in the GCMs.

 

The forcing fields for the perturbed climate differ primarily due to the variation of the vertical velocity shown before. 

 

(go to top)

 

 

D.3 Summary of forcing fields

 

The above procedure completely specified the thermodynamic and dynamic conditions, the vertical velocity and advective tendencies of temperature and water vapor of the atmosphere along the cross section for both the control climate and the perturbed climate.  The procedure is summarized as follow with x representing latitude:

 

SST(x) and RHs --> T(p, x)

RH and T        --> q(p, x)

SST, T, q         --> QR(p, x)

QR(p, x1)         -->omega(p, x1)

omega(p, x1)   --> omega(p, x)

omega, QR, SST, Vs     --> Tadv(p,x), qadv (p,x)

 

We suggest using the SCM/CRM model profiles of temperature and water vapor for the calculation of vertical advection to include a feedback in the simulations.  This is written as

 

                                                    

 

where subscript m is for model fields.  In the absence of clouds, the prescribed temperature and water vapor profiles are steady-state solutions of the model equations in the free atmosphere above 800 mb.  Below 800 mb, there are advective cooling and drying in the atmosphere.  These are necessary to generate clouds in the subsidence region (Adrian Lock, personal communication).  They are expected to be balanced by the surface evaporation and latent heating in the simulations.  An upstream finite differencing in the vertical is desirable to avoid computational instability. To prevent shallow convection near the tropopause, temperature and water vapor above 400 mb can be relaxed to the initially specified profiles in the subsidence region.

 

(go to top)

 

 

References

 

Bony, S., and J.-L. Dufresne, 2005: Marine boundary layer clouds at the heart of tropical cloud feedback uncertainties in climate models. Geophys. Res. Lett., 32, L20806, doi:10.1029/2005GL023851.

Siebesma, A. P., and coauthors, 2003: A large-eddy simulation intercomparison study of shallow cumulus convection. J. Atmos. Sci., 60, 1201-1219. 

 

Siebesma,A.P., C. Jakob, G. Lenderink, R.A.J. Neggers, E. van Meijgaard, J. Teixeira, J. Calvo, A. Chlond, H. Grenier, C. Jones, M. Kohler, H. Kitagawa, P. Marquet, F. Muller, D.Olameda and C. Severijns, 2004: Cloud representation in General Circulation Models over the Northern Pacific Ocean: A EUROCS intercomparison study.  Quart. J. Roy. Meteor. Soc., 130 3245-3268.

 

Zhang M, Bretherton C, 2008: Mechanisms of Low Cloud Climate Feedback in Idealized Single-Column Simulations with the Community Atmospheric Model (CAM3). Journal of Climate, DOI: 10.1175/2008JCLI2237.1

 

Acknowledgements: TBC

 

(go to top)