diurnal cycle of qsr from daily values

[smasson]

commit this long time existing feature...
from daily forcing of qsr, redistribute it at sbc time step (kt*nn_fsbc) with a temporal interpolation following the diurnal cycle.

• Originally developed by Berbie, already used on the ES and by DRAKKAR

ref, see:

Bernie, DJ, Guilyardi, E, Madec, G, Slingo, JM and Woolnough, SJ
Impact of resolving the diurnal cycle in an ocean--atmosphere GCM. Part 1: a diurnally forced OGCM
Climate Dynamics 29:6, 575-590 (2007)

• input forcing must be daily
• it is recommended to have a least 8 sbc time-setp per day => 86400/(rdt*nn_fsbc) > 8
• works only for blk_core and flx

[smasson]

done, see changeset:2187

[smasson]

reconstruction of diurnal cycle

[smasson]

diff between daily mean and mean of diurnal cycle (see Berbie paper...)

[smasson]

I attached 2 plots:

• 12 reconstructed diurnal cycle every 2h (from 1am to 11pm)
• the difference between the original daily mean and the daily mean recomputed from the diurnal cycle. the difference is small (max 5/12 = ~0.4 W/m2) and explained in the paper of Bernie et al. fig 10). Note that the diff is smaller if there is more time step per day (better fit the analytical solution)

[smasson]

code review done by Gurvan, see changeset:2188

[smasson]

bugfix on the definition of the limits of the time interpolation as sbc_dcy is called every rn_rdt * nn_fsbc.

=> now we have the perfect conservation of the diurnal cycle. see attached plots

see changeset:2210

[smasson]

new reconstruction of the diurnal cycle

[smasson]

NEW diff between daily mean and mean of diurnal cycle

[smasson]

adapt for coupled mode, see changeset:2216

[nicolasmartin]

Milestone 2010 Stream 3: Developments & Implementation of new features deleted