WRF-Chem restart query

[Lin Lu]

Dear all,

I wish ask some general questions about running WRF-chem with restart files, since I could not find much documentation around my queries. 

> I am running a year of continuous simulation using chme opt 201 and using MOZART for IC BC. I generate restart files at the end of each month.

>>> My question is - do I need to provide the IC BC  for all the  months in addition to using the restart file generated from the previous month?? I mean if the restart file contains the  information from the previous month, are IC BC needed to be provided or they are not not necessary in that case and somehow produce inaccurate model results?

 

 >>>> I am also wondering if a sensitivity run can be started from any point in the simulation if I have the restart files, given the sensitivity run would have changes in emissions and physics schemes etc. As I'd like to consider only last 3 months from  one year base simulation and run all sensitivity experiments in parallel using restart files from the base run. 

Please advice! I'd be grateful. 

 

Lin,

[Mary Barth]

Lin,

Perhaps others could chime in on this conversation -- especially question 2.

1. >>> My question is - do I need to provide the IC BC for all the 

months in addition to using the restart file generated from the previous
month?? I mean if the restart file contains the  information from the
previous month, are IC BC needed to be provided or they are not not
necessary in that case and somehow produce inaccurate model results?

--> You need to have BC files for the entire year, but IC file is only
for the very beginning.

2. >>>> I am also wondering if a sensitivity run can be started from any

point in the simulation if I have the restart files, given the
sensitivity run would have changes in emissions and physics schemes etc.
As I'd like to consider only last 3 months from  one year base
simulation and run all sensitivity experiments in parallel using restart
files from the base run.

--> You should be able to do this, but I recommend checking the WRF
tutorial or User Guide or WRF forum to be sure it's done correctly.

-- Mary

[Lin Lu]

  Dear Mary, 

Many thanks for your reply.

I used IC and BC both for running the entire year and each time using restart files from the previous month for each successive month.

Could you advise if that would lead to any errors in my results and do I need to re-run all year again?  I was not aware I had to switch off IC from the following months.

I appreciate any other inputs from other members also.

Thanks ,

Lin

[Mary Barth]

Lin,

I brought this up at our WRF-Chem group meeting. WRF (meteorology only) should give bit-for-bit reproduceability from a restart. For example, comparing results from a simulation that is run from 0 to 3 days and a simulation that is run from 0 to 2 days then restarted for day 2-3 should have the same meteorology results.

However, WRF-Chem does not give bit-for-bit results with a restart file. You can test this yourself for a similar test case. Nevertheless, for long simulations, WRF-Chem is often run with restarts because there's just not enough computer time to run an entire year at once. Hopefully there's some effort to explore this further by WRF and/or WRF-Chem support team (which is small). 

For running sensitivity simulations, it depends what you plan to test sensitivity to. If it's meteorology parameterizations, then the winds, temperatures, clouds will be different and all that affects the predictions of constituent concentrations. You may want to re-run the entire simulation for this case. If the sensitivity is to emissions and there are NO aerosol effects* on meteorology turned on, then you should theoretically be able to start the sensitivity simulations from a restart time. However, it would be good to check that the meteorology doesn't change.

*aerosol-radiation feedback = OFF, and aerosol-cloud effects are not included.

Again, it would be good to hear of other people's experiences.

Mary

[Gabriele Pfister]

Dear Lin

I would like to second the suggestions Mary provided and also that the considerations for sensitivity simulations will strongly depend on the objective.

In your previous email you had mentioned testing different physics. Also here I would like to second Mary's suggestion to reach out to the WRF community. Some physics settings such as changing the land model or vertical resolution will not be possible from a restart file but would require starting a new initial simulation. 

Gabi

--

================================
Gabriele Pfister, Interim NCAR Associated Director

Atmospheric Chemistry Observations & Modeling

National Center for Atmospheric Research (NCAR)

Email: pfi...@ucar.edu

Temporary Home Office Phone: +1 303 241 1329

Work Phone: +1 303 497 2915

Web: https://staff.ucar.edu/users/pfister

[Lin Lu]

Dear Mary, Gabi,

Many Thanks for the helpful suggestion.

I intend to modify the anthro emissions and fire emissions (40% decrease)  to look at changes in aerosols as well as the radiation flux changes using the 201 scheme. This means I do have aerosol-radiation feedback = ON in all my base runs and the two months that I intend to run the sensitivity experiments for. 

From what Mary says, this means I should ideally not be using the restart files from my base run for my objective as the feedback was  ON. So in this case I'd need to run new simulations for the 

two months of the sensitivity run (wherein I'd modify the anthro  & fire emissions) ?

I'd also imagine looking at with and without nudging scenarios would mean running new simulations without previous restart files for the time period I am interested in. 

I would really appreciate it if you could confirm I got this right or not.

Thanks, lin

[Gabriele Pfister]

Lin, 

again, it depends. You could start from a restart file if you are only changing emissions but you would have to choose the analysis period such that you allow for sufficient time to spin up and have the simulation adjust to the changed inputs. 

Also, to reiterate Mary's point. With aerosol-radiation feedback turned on, you will also have changes in meteorology and transport hence differences in concentrations will be reflective of the changes in emissions and meteorology/transport.   

Gabi

[Lin Lu]

Dear Mary, 

I wish to pick up a point from your answer and seek another suggestion based on your answer about using the restart files for sensitivity runs.

1.  I wish to test the sensitivity of emissions on the meteorology and hence modify emission files to see corresponding changes in the meteorology as well as concentrations. In this case can I utilize the restart files from a base run for the sensitivity runs?

2. Secondly,  Could you help me know how only IC is provided in start and BC in all the runs ? I mean to ask - I have have_bcs_chem = .true.. and I assume each does not need to specify anywhere in the model to input initial conditions only at the start of the run. It automatically does so. Is this correct?

With many Thanks, 

Lin

[Alvin Christopher Galang Varquez]

Dear everyone,

When we do a restart run with wrf-chem, does the wrfrst* file contain the information of wrfchemi?

I only have the information of one time information for the wrfchemi but seems to ask for this file during the restart run.

Is it alright to remove this as an input in the namelist when running a restar run?

Thanks a lot,

Alvin