Set-up for a new simulation to focus only on quasi-night-time conditions:

We decided to start the simulation at 1000 UTC. At this time, the low-level jet has not developed yet and the turbulence is already much smaller.
As shown in the figure, evol_ws_0_40.gif, 1000 UTC (which corresponds at the 10 hour of run) corresponds to a minimum in the mean wind speed averaged over the first 40 m. It corresponds also to a minimum in the turbulent kinetic energy as shown in the figure evol_tke_0_40m.gif

The idea is to explore the 12 hours after this time.

As the turbulence are going to be very shallow, this permits us to use a much thinner domain (500m wide, 150m thick). Some preliminary work (notably by B Maronga) seems to indicate that it is
better to use an isotropic grid. So we propose for the LES a resolution as *Dx=Dy=Dz=1m*.

As no radiosounding data is available at this time, we propose to use the average of  3 LES using a zo=10-3 (the Meso-NH, the PALM and the JPL simulations).
You can find the initial profiles in the following netcdf file.

Please still use the Stage 3 set-up information for the rest (roughness lengths, forcings...).

Total information:
- Domain:  500mx500mx150m
- Resolution : 1mx1mx1m
- Initial profiles (th, u, v) : GABLS4_SCM_LES_STAGE3_10h.nc
- roughness length: zom=10-3; zoh=10-4
- surface conditions: prescribed surface temperature (but be careful that the
simulation will start at 1000 UTC, you can use the time evolution of Ts in GABLS4_SCM_LES_STAGE3_10h.nc)
- boundary conditions: cyclic
- top: you can add a dumping layer above 120m
- large-scale forcing: geostrophic wind constant in time, no advection in T

To be able to disentangle the impact of flux formulation and turbulence scheme
in the large spread of results of sensible heat fluxes, we propose that each
center runs its simulation with the default formulation of flux computation
and another simulation that can be seen as a sensitivity test where we ask to
use the following formulation for flux computation:

Proposed flux formulation inspired by the flux formulation of GABLS1:
Z_Z0_O_Z0H = 10.0 ! the ratio between roughness length for momentum and heat
ZBM=4.8 ! constant used in the formulation
ZBH=7.8 ! constant used in the formulation
ZL(:)=-9999! initialization of the Monin-Obukhov lengthscale
XKARMAN=0.4 ! Von-Karman constant
XP00=1.E5 ! reference pressure
XCPD=7.*XRD/2.
XRD/XCPD=2./7.

XTG= surface temperature
PPS: surface pressure
PTA: temperature of the 1st level
PPA: pressure of the 1st level
PU: zonal wind of the 1st level
PV: meridional wind of the 1st level
PRHOA= density of the 1st level

ZVMOD= WIND_THRESHOLD(SQRT(PU(:)**2. + PV(:)**2.),ZREF)
!routine to avoid a null modulus of wind (cf below)

DO ZI=1,50 ! iteration over 50 times

! computation of u*
ZUSTAR(:) = ZVMOD(:)/(LOG(ZREF(:)/Z0)/XKARMAN &
+ZBM/(ZL(:)*XKARMAN)*(ZREF(:)-Z0))

! computation of T*
ZTSTAR(:) = (PTA(:)*((XP00/PPA)**(XRD/XCPD))-XTG(:,1,1)*((XP00/PPS)**(XRD/XCPD))) &
& /(LOG(PZREF(:)/Z0/Z_Z0_O_Z0H)/XKARMAN+ZBH/(ZL(:)*XKARMAN)* &
& (ZREF(:)-Z0/Z_Z0_O_Z0H))

WHERE(ABS(ZTSTAR)<1.E-10)
ZTSTAR(:) = 1.E-10
END WHERE

WHERE(ABS(ZUSTAR)<1.E-10)
ZUSTAR(:) = 1.E-10
END WHERE

ZL(:)=(ZUSTAR(:)**2)/(XG/(PTA(:)*((XP00/PPA)**(XRD/XCPD)))*XKARMAN*ZTSTAR(:))

END DO

PSFU(:)  = -PU(:)/ZVMOD(:)*(ZUSTAR(:)**2)*PRHOA
PSFV(:)  = -PV(:)/ZVMOD(:)*(ZUSTAR(:)**2)*PRHOA
PSFTH(:) = -ZTSTAR(:)*ZUSTAR(:)*PRHOA*XCPD
-

Wind_threshold:
= = = = = =  = =
;determine a threshold for the wind modulus
_------------------------------------------
Wind_threshold(Pwind,Zref) result(Pwind_new)
Pwind_new=max(Pwind,0.1*min(10.,Zref))