Trouble running ungrib and metgrid with long-term data with scattered years

[Tam Nguyen]

Dear all,

Please excuse myself since I'm a totally beginner with WRF. I've tried to search for the answer thoroughly but couldn't find any possible solution.

My current aim is I'm trying to run WPS for a few years including 1887, 1889, 1930, 1945, 1954, 1971, 1972 with data from 20CR Reanalysis Datasets of NOAA with timestep is daily (a.k.a. 24 hours).
Domain: I have the intention with 3 domain (100km, 30km, 10km).
Time in &share: As I know, this should cover all the time domain so I put:
start_date: 1887-01-01_00:00:00
end_date: 1973-01-01_00:00:00

Therefore, as instructed, I've set the start_date and end_date is 1887 and 1973 as in the file below. However, every interval_second I've tried is not possible (24 hours, 12 hours, 6 hours, 3 hours). Unless I set the interval_second as 1 seconds, the error when running ungrib will always be:
ERROR STOP IN READ_NAMELIST
INTERVAL must fit evenly between start time and end time:

If I do indeed set the interval_second as 1, running metgrid will give the error:
end_date is earlier than start_date.

I'll attach the namelist.wps below and longing for any suggestions. Thank you

 

[kwerner]

Hi,
Do you have data consecutively (i.e., every 24-hour interval) between 1887 and 1973? I think, based on the title of this inquiry, you only have it for certain years, and not the years in between. Is that correct?

You can only run for the times you have available consecutively. So if you have the full 1887 data, every 24 hours (i.e., 365 24-hour files), you will need to just run that year by itself. Then if you have the full 1889 data (365 24-hour files), you will then run 1889 separately. Unfortunately if you are missing the data between those years, you won't be able to run this is a single continuous simulation.

 

[Tam Nguyen]

Hi,
Do you have data consecutively (i.e., every 24-hour interval) between 1887 and 1973? I think, based on the title of this inquiry, you only have it for certain years, and not the years in between. Is that correct?

You can only run for the times you have available consecutively. So if you have the full 1887 data, every 24 hours (i.e., 365 24-hour files), you will need to just run that year by itself. Then if you have the full 1889 data (365 24-hour files), you will then run 1889 separately. Unfortunately if you are missing the data between those years, you won't be able to run this is a single continuous simulation.

Thank you so much for your reply. I've indeed only collected data from those specific years and been desperate to find a solution. Your recommendation has saved my current miserable life.

Again, thanks a lot!