RegIPSL MED Wiki

All the simulations compared here are run and compared for the month of July 2014. We start from all restart states of the MEDCORDEX-VEG simulation.

In order to be more efficient, all corrections are applied sequentially. So to understand the impact of the various corrections the runs need to be compared pairwise.

This is our reference simulation for the state of the model as it was during MEDCORDEX-A. The major difference to MEDCORDEX6A is that vegetation phenology is activated and thus LAI becomes a prognostic variable.

Based on this simulation we can see the impact of the various errors corrected.

In this simulation, the average of surface temperature within the radiation driver is now corrected. This average is built so that all surface temperatures computed by the rest of the model between radiation time-steps are used for computing upward long wave flux. This is only performed over the continents because of its faster fluctuations.

In MEDCORDEX-A the average temperature wax divided by one less time-step than should have been. So less upward long-wave radiation has to be expected for this simulation.

WRF is designed to work with exchange coefficients for the surface turbulent fluxes, while ORCHIDEE works with surface drag coefficients. This has lead the exchange coefficients instead of the drag to be passed to ORCHIDEE. The different is the module of the wind which is not part of the drag coefficient.

To correct this, module_sf_mynn now also outputs the drag coefficient. In the WRFORCH-CDRAG simulation that variable is passed to ORCHIDEE.

WRF makes the distinction between the roughness for momentum and scalar fluxes (T and q). In MEDCORDEX-A only the roughness length for momentum computes by ORCHIDEE was passed to WRF. Thus WRF computes the turbulent fluxes with a surface drag which was not consistent with the surface states described by ORCHIDEE. It also feedback to ORCHIDEE as the drag coefficient received from WRF did not take into account the surface roughness for scalars which it computed.

In order to clear this module_sf_mynn now also receives the roughness for T and q from ORCHIDEE. This variable is then used to estimate the surface drag within the module and which given back to ORCHIDEE.

This should be the configuration for MEDCORDEX-B. After the correction of the various coupling error, we have reduced the time step of WRF to 90s. It is hoped that this is possible because the model seems much more stable than in previous versions.

The comparison are centered on 2m Temperatures (average, maximum and minimum), precipitation, surface latent and sensible heat, and net longwave and shortwave radiations.

This comparison measures the impact of the error on the time averaging of surface temperature for the radiation calculations.

Figures : http://www.lmd.jussieu.fr/~polcher/CliMAF/MEDCORDEX-B/index_WRFORCH-RADCORR-WRFORCH-VEG.html

Interpretation :

• The surface warms-up considerably as the long-wave exchanges with the atmosphere have now the right magnitude.
• Precipitation increases.
• Evaporation also increases thus leading to a more active water cycle over the continents.
• The impact on the flux exchanges of the atmosphere with the oceans are not negligible.
• There is even a small but non negligible impact on SWnet over the ocean thus indicating that cloudiness might be affected.

This comparison evaluates the combined effect of moving to an exchange of surface drag and taking into account the surface roughness for T ans q computed in ORCHIDEE.

Figures : http://www.lmd.jussieu.fr/~polcher/CliMAF/MEDCORDEX-B/index_WRFORCH-Z0QT-WRFORCH-RADCORR.html

Interpretation :

• The surface warms further over the continents.
• The diurnal amplitude of the 2m temperature increases.
• Evaporation is reduced over the continents. An effect which could be lower in a longer simulation as this also depends on the amount of water in the soil and the restart comes from a run which too little rainfall.
• Sensible heat is generally increased (by about 2-5 W/m²).
• Precipitation and net surface radiation fluxes are not significantly affected.

Here we examine the impact of the longer time step on the surface fluxes of the model.

Figures : http://www.lmd.jussieu.fr/~polcher/CliMAF/MEDCORDEX-B/index_WRFORCH-90s-WRFORCH-Z0QT.html

Interpretation :

• The precipitation is slightly increased by the longer time step.
• This seems to lead to higher evaporation and sensible heat flux.
• There is no systematic increase in net radiation.

Daily analysis on the point lon=-4.53, lat=42.28 to see if the variability of the surface variables increases.

Variable	Daily values	3 hourly values
2m temperature
Latent heat flux
U* : turbulent wind
Precipitation

Interpretation :

• The impact of the time step on turbulent fluxes and surface temperature is relatively weak.
• There is also a slight increase in precipitation when it rains.

Comparing WRFORCH-VEG : http://www.lmd.jussieu.fr/~polcher/CliMAF/MEDCORDEX-B/index_WRFORCH-VEG-E-OBS.html

Comparing WRFORCH-RADCORR : http://www.lmd.jussieu.fr/~polcher/CliMAF/MEDCORDEX-B/index_WRFORCH-RADCORR-E-OBS.html

Comparing WRFORCH-Z0QT : http://www.lmd.jussieu.fr/~polcher/CliMAF/MEDCORDEX-B/index_WRFORCH-Z0QT-E-OBS.html

Comparing WRFORCH-90s : http://www.lmd.jussieu.fr/~polcher/CliMAF/MEDCORDEX-B/index_WRFORCH-90s-E-OBS.html

Some interpretations :

• WRFORCH-VEG was clearly too dry and cold when compared to E-OBS.
• The correction of the radiation brought more rainfall (positive when compared to E-OBS) but also a hotter surface than observed.
• That bug-correction also leads to an amplitude of the diurnal cycle which is larger than the one proposed by E-OBS.
• The 90s time step increases rainfall a little and the difference to E-OBS is similar to the one of WRFORCH-Z0QT.

One essential question is how much the restart state on 1st on July influences these results. The initial conditions come from the MEDCORDEX-VEG simulations and thus have a very dry soil and probably sparse vegetation. In a simulation starting 1st on January there could be more water in the soils by July and thus leading to more evaporation and cooler soils.