160 likes | 287 Views
LMDZ Single Column Model + what is it ? + why is it interesting ? + List of 1D cases + how to install and run it ? M-P Lefebvre and LMDZ team. Use of 1D cases. Observation. Test case, field campaign experiment. Evaluation. Explicit simulations, Grid cell, 20-100 m.
E N D
LMDZ Single Column Model + what is it ? + why is it interesting ? + List of 1D cases + how to install and run it ?M-P Lefebvre and LMDZ team
Use of 1D cases Observation Test case, field campaign experiment Evaluation Explicit simulations, Grid cell, 20-100 m « Large scale » conditions imposed Evaluation Climate model, parameterizations, « single-column » mode Courtesy F.Hourdin → Parameterizations are evaluated against other models → Can be done for realistic test cases but also with more idealized forcing (check the response of the parameterization to perturbations)
3D is a collection of many “single column models”, covering earth and interaction with each other through a set of rules known as “large scale dynamics”. In a 1D model, there is no dynamics. We use observations or model output or idealized forcing to impose forcing at the boundary of the column.
LMDZ model in 1D mode - We impose large scale conditions. - Duration of the case varies from few hours to few months - We study parameterizations in a given environment. z~20km Δz = 30m - 1km T (K) Q (g/kg) Large scale forcings (constant or not): Temperature, humidity, Wind advection Surface conditions: Surface fluxes or Surface temperature Δx = 50-300 km Initial conditions: Temperature, humidity, wind profils
Why use SCMs ? + simplicity: technical and understanding, usable on any labtop + it's a useful tool for parameterization development: shallow convection, deep convection, transition from stratocumulus to cumulus, stable boundary layer, radiation... + we can compare results to observations or to explicit simulations (CRM, LES) + then we go back to GCM and test new parameterizations ... + hierarchy of models: SCM, CRM, AGCM, GCM ...
shallow convection • Arm_cu (diurnal cycle of shallow cumulus over land) • Rico (shallow precipitating cumulus over sea) • Ayotte (convective boundary layer ) Arm_cu case - Cloud fraction
Stratocumulus and transition to cumulus • Sandu (transition case with 3 options according to variation of SST) • Fire (diurnal cycle of stratocumulus) Fire case: Cloud fraction (%) Top: standard version Bad representation because not Enough entrainment at the cloud top Bottom: version developped by A.Jam
Deep convection: Over ocean: • Toga • case_e (part of Toga) • TWPICE TWPICE Case Vertical speed (m/s) Precipitation (mm/j)
LES LMDZ_AR4_L39 AMMA case: Thetal tendencies due to all schemes (K/j) Deep convection: Over land: • Hapex • AMMA • Idealized case: eq_rad_conv (RCE) LMDZ_NP_L70
How to proceed ? + install LMDZ 3D with install.sh + install LMDZ 1D with instal1d.sh. * this script adds some routines specific to 1D * creates phy1D directory * compiles * runs 3 cases (toga, twpice, arm_cu) and show cloud cover plot
drwxr-xr-x 13 mpllmd lmdjus 4096 Nov 29 17:18 .drwxr-xr-x 10 mpllmd lmdjus 4096 Nov 29 19:32 ..drwxr-xr-x 3 mpllmd lmdjus 4096 Nov 21 16:30 bibiodrwxr-xr-x 3 mpllmd lmdjus 4096 Nov 21 16:30 cosp-rw-r--r-- 1 mpllmd lmdjus 888775 Nov 29 19:29 create_make_gcm.log-rw-r--r-- 1 mpllmd lmdjus 67587 Nov 29 19:27 create_makeg_gcm.logdrwxr-xr-x 3 mpllmd lmdjus 8192 Nov 21 16:30 dyn3ddrwxr-xr-x 3 mpllmd lmdjus 8192 Nov 21 16:30 dyn3dmemdrwxr-xr-x 3 mpllmd lmdjus 8192 Nov 21 16:30 dyn3dpardrwxr-xr-x 3 mpllmd lmdjus 4096 Nov 21 16:30 filtrezdrwxr-xr-x 4 mpllmd lmdjus 97 Nov 29 19:32 griddrwxr-xr-x 4 mpllmd lmdjus 12288 Nov 29 19:23 phy1ddrwxr-xr-x 3 mpllmd lmdjus 4096 Nov 21 16:30 phydevdrwxr-xr-x 4 mpllmd lmdjus 8192 Nov 21 16:30 phylmddrwxr-xr-x 6 mpllmd lmdjus 92 Nov 21 16:30 .svn Some changes ... in ~/LMDZ20131121.trunk/modipsl/modeles/LMDZ5/libf = phylmd + 1D specific routines In phy1d, you can modify the code
Some changes also ...in ~/LMDZ20131121.trunk/1dcases drwxr-xr-- 8 mpllmd lmdjus 4096 Dec 3 10:28 amma drwxr-xr-x 4 mpllmd lmdjus 4096 Dec 3 10:50 arm_cu drwxr-xr-x 5 mpllmd lmdjus 4096 Dec 3 11:55 ayotte drwxr-xr-x 3 mpllmd lmdjus 4096 Dec 3 12:12 case_e drwxr-xr-x 3 mpllmd lmdjus 4096 Dec 3 12:14 eq_rd_cv drwxr-xr-x 5 mpllmd lmdjus 4096 Nov 29 19:29 fire drwxr-xr-x 3 mpllmd lmdjus 4096 Nov 29 19:29 hapex92_init drwxr-xr-x 4 mpllmd lmdjus 4096 Nov 29 19:29 rico drwxr-xr-- 5 mpllmd lmdjus 4096 Nov 29 19:29 sandufast drwxr-xr-- 5 mpllmd lmdjus 4096 Nov 29 19:29 sanduref drwxr-xr-- 5 mpllmd lmdjus 4096 Nov 29 19:29 sanduslow drwxr-xr-x 5 mpllmd lmdjus 4096 Nov 29 19:30 toga drwxr-xr-x 4 mpllmd lmdjus 4096 Nov 29 19:31 twpice Here, you can compile the model and run your favorite case
What can you do in each case directory ? + compile with compile.x: choose level number + profils and forcings (ascii files or netcdf files) + def files + readme + execute with xqt.x or mr file + get results in netcdf files (histhf.nc) + compare to LES results in resul_LES
In LMD, we used these cases to develop New Physics version. For 2 cases, Arm_cu and Rico, we compare results of « standard physics » (CMIP3), « new physics » (CMIP5) and LES model. Shade= cloud cover Contour= specific humidity (g/kg)
Rico case : Sensitivity Tests to vertical discretization and time step