Hauteur de deplacement dans l'interface surface atmosphere

[jpolcher]

The driver interpolates the wind (usually at 10m) to the level at which T and q are available (usually 2m). This done simply by assuming a logarithmic profile in the surface layer. It has to carried out each time the wind and the scalar variables are not at the same level (i.e. in most cases).

So the interpolation is written as :

u_2m = u_10m * log(zlev_tq/z0)/log(zlev_uv/z0)

This produces non sense as soon as z0 is larger than zlev_tq (i.e. 2m). Furthermore this is not consistent with what is assumed for the vertical wind profile in diffuco_aero when the neutral drag coefficient is computed.

So we need to replace this interpolation by the following formula :

u_2m = u_10m * log((zlev_tq+roughheight/z0)/log((zlev_uv+roughheight)/z0)

This means that roughheight has to come out of the sechiba_main and through intersurf so that it can be used in dim2_driver. This changes the interface !!

The consequence of this change will be that the wind over rough terrain will be higher and thus we are likely to generate more turbulence and higher fluxes. The consequences of this change need to be studied in detail.

[aducharne]

Jan, can you explain the difference between z0 and roughheight ?
(as I understand z0 as a roughness height...)

[jpolcher]

In ORCHIDEE and LMDZ, z is the height above the point where the wind is zero (in log profile assumption). In other words the GCM does not simulate place where the air does not move.

In the case of the a short vegetation or bare surface z is the actual height above the ground. But when one considers a tall vegetation this point where u=0 is well above the surface. Thus in the computation of the neutral drag coefficient we need to take this into account.

z is the height above the 0 wind point, i.e. the height of the atmospheric levels.

roughheight is similar to the displacement height (the displacement height is used when z is the total height above the ground and u=0 at z-displacement_height).

z0 is the roughness length of the surface.

Thus the neutral drag coefficient needed for ORCHIDEE is proportional to :
log((z+roughheight)/z0) This can be verified in diffuco_aero.

The vertical interpolation in dim2_driver (an probably LMDZ) compute the neutral drag as :
log(z/z0). One sees readily that as soon as z approaches z0 or becomes smaller we have a problem in the interpolation.

[fcheruy]

the screen level variables are only diagnostics now (Orchidee uses the value at the first atmospheric level ? Not true for Orchidee_cmip6.
(ok_q2m_t2m should be set to false by default).

[jgipsl]

Changed milestone as decided during ORCH meeting today