Error reading user created solar data file

[andrew_kren@colorado_edu]

I created a solar data file called solar_data_file_3.nc for my WACCM simulations. I followed the generic format of the lean netcdf solar spectral irradiance file and modified it to have the last 4-5 solar cycles repeated. I changed my dates and times to match with my model starting at year 0 and ending at year 199. I put my data as daily data and followed the date and time format as shown in the generic netcdf file. When I ran WACCM, I got this error:/projects/ankr3992/inputdata/atm/cam/solar/solar_data_file_3.nc      196608
solar_data_advance: not able to find data for : 0000-01-01-00000
(shr_sys_abort) ERROR: solar_data_advance: failed to read data from /projects/ankr3992/inputdata/atm/cam/solar/solar_data_file_3.nc
(shr_sys_abort) WARNING: calling shr_mpi_abort() and stopping
solar_data_advance: not able to find data for : 0000-01-01-00000

I cannot figure out why it says it can't find data for the first year and day, I checked my netcdf file and it starts at year 0 with date format of '101' and time of '0'. Does anyone know what may be going on here. I assume its something simple, but not sure. Any help would be much appreciated.Thanks,

 

[santos]

For vanilla CAM, runs typically start at year 1 regardless of the real world year they correspond to. But WACCM often uses the real world year (e.g. 2000). I'm not sure what the reason for the difference is; it may just be historical, or to deal with specific transient data sets. Regardless, the start/end dates on your files need to cover the full time that your run will be over. In this case, that may mean that your file dates should start at 00000101 and go to, say, 01991231. ("101" really is 0000-01-01, which is a concatenation of year/month/day).

 

[andrew_kren@colorado_edu]

Yes, I am starting at year 0 as in my ocean data and model time. So are you saying it should not be 101 or something else?

 

[andrew_kren@colorado_edu]

Hi Sean,I verified that my start date is 101 and my end date is 1991231, so that is all correct. But my time variable starts at 0. Do you think that should be a 1 to start out? My ocean data file has the same dates and it was able to read it at year 0, meaning it started at 101 and ended in year 199.

 

[santos]

Sorry; I meant to post this earlier, but I think there was a momentary forum glitch and it didn't go through.I seem to have misread the original post. I'm still in the dark about what's going on here.One thing that's not clear to me is this line:/projects/ankr3992/inputdata/atm/cam/solar/solar_data_file_3.nc      196608I think that this is a continuation of a line that isn't being printed here. Can you post a few things?1) Model version.2) More lines from the log file above this error.3) atm_in, or at least the values of anything in the solar_inparm section. Actually, that might be your issue; I forgot about the namelist options "solar_data_ymd" and "solar_data_tod", which might get defaults from build-namelist that are wrong in this case.

 

[andrew_kren@colorado_edu]

You are right. My solar_data_ymd is set to 18500101. Should I set that to 101, my start date?

 

[santos]

Yes, I believe that that will fix it. If it doesn't, I suppose you'll find out pretty quickly. Let me know!

 

[andrew_kren@colorado_edu]

Sean,changing that solar_data_ymd = 101, that fixed my problem. Thanks for your input on that. I did have one other question though. When I created my solar_data_file, I did not change tsi_ref and ssi_ref variables. I used the last four solar cycles. The variable definition for these variables say that the reference spectrum is the average over ref_time_bound inclusive. Do I need to change the reference variables to be the average over my specified period (200 yrs)? Or can I just leave those variables as is? Also, what is the ref_time_bound variable anyway? When I do an ncdump on it, it prints this: "669591, 681636. Thanks.

 

[santos]

Hmm. To be honest, I don't know what these references are for. ref_time_bound is not used by the model, so its presence in the file must be purely for documentation. Furthermore, these reference variables are not used at all when RRTMG is on.But for a standard WACCM 4 case, I'm not sure; maybe someone else can answer.

 

[andrew_kren@colorado_edu]

The reference variables say they are a function of ref_time_bound. So I assumed I needed it. I took the reference variables out of the file but waccm said it needed those so I put them back in. Thanks. I hope someone can answer.

 

[mmills]

Andrew,The values for tsi_ref and ssi_ref should be the same across all solar_data_files used with WACCM4. The model was tuned to use these values to scale heating in the CAMRT bins relative to a reference spectrum.You should copy tsi_ref and ssi_ref from one of our files in inputdata/atm/cam/solar, such as:

• spectral_irradiance_Lean_1850_cntl_c100407.nc
• spectral_irradiance_Lean_1950-2140_daily_c100804.nc

The model does not read in ref_time_bound, but you can keep that the values we use consistently in those 2 files.See also this forum post: https://bb.cgd.ucar.edu/node/1001616#comment-1004357