Opened 2 years ago
Last modified 23 months ago
#836 new enhancement
New and old drivers don't have the same default for SPREAD_PREC
Reported by: | aducharne | Owned by: | somebody |
---|---|---|---|
Priority: | minor | Milestone: | ORCHIDEE 4.3 |
Component: | Driver files | Version: | trunc |
Keywords: | Cc: |
Description
The default values differs:
- In the old driver, the default SPREAD_PREC is half the forcing time step if the latter is GE 3h, else it equals the forcing time step;
- In the new driver, the default SPREAD_PREC is always half the forcing time step. There is also a possibility to "connect" the rainfall from two consecutive forcing time steps SPREAD_PREC_CONT, but the default is FALSE.
I think we should enforce a similar configuration of the two drivers as much as possible.
Personnally, I prefer the choices of the old drive, as intense rainfall rates reduce infiltration, SM and transpiration/photosynthesis. But I will follow the majority.
Change History (4)
comment:1 Changed 2 years ago by jpolcher
comment:2 Changed 23 months ago by jgipsl
Decided at today's ORCH meeting : the trunk will change the default behaviour for SPRED_PREC according to what is done for the old driver.
comment:3 Changed 23 months ago by xnwang
Forcing_tools.f90 in the new driver in r7611 has the following functionality :
1) if SPREAD_PREC_SEC is defined by user in run.def: use this value for the simulation
2) if SPREAD_PREC_SEC is not defined by user in run.def: use the same default behavoir as in old driver. That is, precipitation is spread over a duration of half of the forcing time step if the latter GE 3h, otherwise the same duration as the forcing time step
This means a bit more flexibility to users about how to spread the precipitation, though it is different from the default behavoir in the old driver.
comment:4 Changed 23 months ago by aducharne
Thanks a lot Xiaoni. Now, we have to ensure that SPREAD_PREC_SEC does not override the default behavior in the FGxnd configs, by commenting out this keyword.
In branch 2.2, this is done, but the code of forcing_tools.f90 does not include any equivalent of r7611 (and the default for SPREAD_PREC_SEC = 3600) => NEEDS TO BE CHANGED.
I chose to set SPREAD_PREC independent of the forcing time step. So have it as a fixes values in seconds. We now have hourly rainfall forcing. It would not be realistic to have 1/2h rain events throughout the year.
Many years ago I showed that in mid-latitudes the "conneted" rainfall events was quite realistic in particular in winter when rain events last many hours. The results were shown to the ORCHIDEE group.