search

daliagachc / wrf_management

package created to manage the run of wrf at uhel supercomputer (taito)
0 stars 0 forks source link

sst_update #2

Open daliagachc opened 5 years ago

daliagachc commented 5 years ago

What I see right now: in our short run we have fixed SST ( sst_update = 0 real's namelist.input ) so no wrflowinp files were created. Would it be hard to run real.exe with sst_update = 1 or do you have wrflowinp file example to play around? It seems the easiest way to handle lakes' temperatures would be to run real.exe as is and then modify wrfinput and wrflowinp files.

daliagachc commented 5 years ago

@Rulock I'll try to make the next run with sst_update=1

daliagachc commented 5 years ago

i created this new run at /proj/atm/saltena/runs/run_2019_02_28_2 with sst_update (therefore the wrflowinp files). its also ready for nudging @Rulock

Rulock commented 5 years ago

i created this new run at /proj/atm/saltena/runs/run_2019_02_28_2 with sst_update (therefore the wrflowinp files). its also ready for nudging @Rulock

I have found the way to modify the .nc files right on the taito-server using NCO so we could modify met files as easy as wrf input. The only problem I'm still struggling with is how to pass shell's variable to *.nco script. @daliagachc

Rulock commented 5 years ago

i created this new run at /proj/atm/saltena/runs/run_2019_02_28_2 with sst_update (therefore the wrflowinp files). its also ready for nudging @Rulock

i've updated wrf inputs files and uploaded it to /wrk/kadantse/ws2019/wrf_inits_mod Still fighting with NCO, probably should do that in Python @daliagachc

daliagachc commented 5 years ago

ok. ill copy those and start the run :). python could help (as long as it does not do any funny stuff with the format )

daliagachc commented 5 years ago

ok... wrf is running!

daliagachc commented 5 years ago

@Rulock: lets keep ourselves updated here to what we find out about the SST

Rulock commented 5 years ago

@Rulock: lets keep ourselves updated here to what we find out about the SST

@daliagachc: I have found this: sst_update = 1 real.exe will create wrflowinp_d01 file at the same time interval as the available input data. To use it in wrf.exe, add auxinput5_inname = "wrflowinp_d01", auxinput5_interval, and auxinput5_end_h in namelist section &time_control https://esrl.noaa.gov/gsd/wrfportal/namelist_input_options.html but it should be working as you set it: http://forum.wrfforum.com/viewtopic.php?f=6&t=2826

daliagachc commented 5 years ago

ok! so i updated all the values but auxinput5_end_h which by default is 0! so I think thats the problem. I'll try a new run with auxinput5_end_h set to 500 hours or so.

Rulock commented 5 years ago

or auxinput4_end_h in our case

VictoriaSinclair commented 5 years ago

if the number (4) applies to the domain number, we should make sure that the SSTs (and the land surface) are updated in all four domains, not just the innermost domain

Rulock commented 5 years ago

if the number (4) applies to the domain number, we should make sure that the SSTs (and the land surface) are updated in all four domains, not just the innermost domain

Oh, you are absolutely right, I've just checked all the domains and it turned out that SST was updated but only in d01. There are some reports on the same issue but with no clear solution: http://forum.wrfforum.com/viewtopic.php?f=47&t=10260 http://forum.wrfforum.com/viewtopic.php?f=6&t=341

daliagachc commented 5 years ago

Ok!
so if it worked for d01 it means that default auxinput4_end_h=0 will be active throughout the run. (@Rulock can you confirm this ?) However, the problem is that i did not set the interval for the other domains so that they are disabled by default. I think that changing: auxinput4_interval_s = 21600 to auxinput4_interval_s = 21600, 21600,21600,21600 should solve the problem. Im creating a new run so that we can check. I'll give a link to the input file so that we can have a look later on. cheers

daliagachc commented 5 years ago

so i set up a new run with suggested changes at /proj/atm/saltena/runs/run_2019_04_03_1

Rulock commented 5 years ago

so i set up a new run with suggested changes at /proj/atm/saltena/runs/run_2019_04_03_1

Bad news: it seems wrf didn't eat SST fields at all, the Lake temperature is ~5 C at the very beginning in all the domains and slowly changes with time. Could it be because wrflowinp linking didn't work? I have checked wrflowinp's, the files contain the right SST.

SST new run SST wrflowinp_d04

daliagachc commented 5 years ago

Good catch. My first guess would be bad linking but the wrflowinps look ok. i will double check.

daliagachc commented 5 years ago

check here: /proj/atm/saltena/runs/run_2019_04_03_1/wrf2 should be working now

Rulock commented 5 years ago

check here: /proj/atm/saltena/runs/run_2019_04_03_1/wrf2 should be working now

no rights :(

daliagachc commented 5 years ago

ok. i fix that. anyways, now is working (i think) check it here: https://github.com/daliagachc/wrf_management/blob/master/wrf_management/plot/sst_check/checkSST_plot.ipynb

VictoriaSinclair commented 5 years ago

I also re-read the documentation: "The WRF model physics does not predict sea-surface temperature, vegetation fraction, albedo or sea ice. For long simulations, the model provides an alternative to read-in the time-varying data and to update these fields. In order to use this option, one must have access to time-varying SST and sea ice fields. Twelve monthly values of vegetation fraction and albedo are available from the geogrid program. Once these fields are processed via WPS, one may activate the following options in the namelist record &time_control before running the program real.exe and wrf.exe:

io_form_auxinput4 =2 auxinput4_inname = “wrflowinp_d” (created by real.exe) auxinput4_interval = 360, 360, 360,

and in &physics sst_update =1

Therefore, inname and interval need to be set for each domain. real and wrf.exe both need to be re-run, and turning sst_update on should also result in the albedo and vegetation fraction being updated.

daliagachc commented 5 years ago

Yes! for the new run (/proj/atm/saltena/runs/run_2019_04_03_1/wrf2) i set up the values as indicated (https://github.com/daliagachc/wrf_management/blob/master/wrf_management/config_dir/run_2019_04_03_1/namelist.input).

Rulock commented 5 years ago

Input files for the big run are ready at (/wrk/kadantse/wrfimp_longrun). There is one day when the lakes are frozen (31.03.2018, four time steps: SEAICE flag set to 1 over the lakes). Since I've modified only Lake Titicaca I kept the ice over the rest: theta_surf_d03_frozen seaice_surf_d03_frozen Is it the right thing to do?

daliagachc commented 5 years ago

Thanks!! Regarding the ice flag, im a little bit concerned. The Amazon should not have any ice at all. Does this happen only for one day? For the Uyuni salt dessert and Poppo lake (nearby) it makes sense to be frozen during the winter but March is not winter. @VictoriaSinclair what do you think?

Rulock commented 5 years ago

Yes, the flag is on only one day (four time steps: 31.03.2018 00:00, 06:00, 12:00, 18:00). SSTs right before the event and right after it are following: theta_surf_d03_before_frozen theta_surf_d03_after_frozen So the temperatures at night seem to be pretty low.

VictoriaSinclair commented 5 years ago

Ice is clearly not correct. Is there an easy way to present ice occurring without modify the temperatures of all of these salt lakes? One way might be to set a minimum temperature allowed for a lake? They are far from the area of interest but I think we should attempt to fix this.

Rulock commented 5 years ago

Two more plots: d02 domain, one time step before all lakes (amazon basin region included) get frozen and first time step with all lakes frozen t_surf_d02_before_frozen t_surf_d02_1st_frozen