How Parallel Offline Radiative Transfer Model (PORT) continue with restart file

[liren]

Dear all,

I'm using PORT and drive it with offline data for many years. For example, running over input data for years 0001-0010. What I have tried is set the "offline_driver_fileslist" and I have successfully finish year 0001. Then at year 0002, I set

./xmlchange CONTINUE_RUN=TRUE
./xmlchange RUN_STARTDATE=0002-01-01

and replace the offline_driver_fileslist with new input file list (in which the timestamps is 0002)

However PORT finished fastly, seems like PORT did not read in the correct data. How can I restart with my restart file? Or I have to reset the timestamps?

Thanks!

 

[liren]

I have noticed the the this thread: Can Parallel Offline Radiative Transfer Model (PORT) loop over years? but I think I am trying to restart the PORT with previous restart file. I am still not sure how to solve this problem now

 

[jzhan166]

Hi Liren,

I have never tried to restart PORT, but there might be an alternative way to solve the problem.
Do you mean you have 'offline_driver_fileslist' for each year separately? Could you try to combine your offline_driver_fileslist for 0001 and 0002 year together? That way you don't need to restart the run. Each time the run will continue running through all the years if the CONTINUE_RUN is set to TRUE.

Best,
Jun

 

[liren]

Hi Liren,

I have never tried to restart PORT, but there might be an alternative way to solve the problem.
Do you mean you have 'offline_driver_fileslist' for each year separately? Could you try to combine your offline_driver_fileslist for 0001 and 0002 year together? That way you don't need to restart the run. Each time the run will continue running through all the years if the CONTINUE_RUN is set to TRUE.

Best,
Jun

Thanks for your kindly advice. However, I have to run the model for about 10 years and the input data is too large to put them together so this method may not suitable for me. Anyway, thank you for your advice.

 

[rrbuchholz]

Hi Liren,

Just to clarify, do the time stamps in your second file start at 0002-01-01? And both the date and time variable is consistent? The model will be looking for dates 0002-01-01 onwards. I am not familiar with PORT, but with other data, you usually need the information about one timestep or more before the start dates and one timestep or more after the end dates for the model interpolation algorithms to work correctly.

You could also try branching the simulation to read in the second file.

Best,
Rebecca

 

[jzhan166]

Thanks for your kindly advice. However, I have to run the model for about 10 years and the input data is too large to put them together so this method may not suitable for me. Anyway, thank you for your advice.

Hi Liren,

Are you running PORT on Cheyenne? If so, would you like to share the path to your simulation? I'd be happy to take a look at your case. If not, would you like to share your offline_driver_fileslist?

Best,
Jun

 

[liren]

Hi Liren,

Just to clarify, do the time stamps in your second file start at 0002-01-01? And both the date and time variable is consistent? The model will be looking for dates 0002-01-01 onwards. I am not familiar with PORT, but with other data, you usually need the information about one timestep or more before the start dates and one timestep or more after the end dates for the model interpolation algorithms to work correctly.

You could also try branching the simulation to read in the second file.

Best,
Rebecca

Thank you for your reply. I checked the time stamps of restart file and it start at 0002-01-01 and I have a series of dataset for 1 year. I also tried to change the time stamp of input data back to year 0001 using nco but PORT mistaked it for year 0002.

Now I bypassed this problem by changing the source code of offline_driver.F90. This is not the right way, but it seems to work. Anyway, thank you for your kindly advice.

 

[liren]

Hi Liren,

Are you running PORT on Cheyenne? If so, would you like to share the path to your simulation? I'd be happy to take a look at your case. If not, would you like to share your offline_driver_fileslist?

Best,
Jun

Thank you for your kindly reply. I am not running PORT on Cheyenne. My "offline_driver fileslist" only contains a list of files and one of the file seems like this:

$ cdo sinfo test_run_sst_3h.wv.h1.0002-01-01-00000.nc
Warning (cdfCheckVars): Unsupported data type (char/string), skipped variable date_written!
Warning (cdfCheckVars): Unsupported data type (char/string), skipped variable time_written!
...

   Time coordinate :  8 steps
     RefTime =  0001-01-01 00:00:00  Units = days  Calendar = 365_day  Bounds = true
  YYYY-MM-DD hh:mm:ss  YYYY-MM-DD hh:mm:ss  YYYY-MM-DD hh:mm:ss  YYYY-MM-DD hh:mm:ss
  0002-01-01 00:00:00  0002-01-01 03:00:00  0002-01-01 06:00:00  0002-01-01 09:00:00
  0002-01-01 12:00:00  0002-01-01 15:00:00  0002-01-01 18:00:00  0002-01-01 21:00:00

Though I bypassed this problem by changing the source code of offline_driver.F90 now, this is not the right way, but it seems to work.