Vorticity budget for 0.1deg POP

Quick start

1. Edit environment variables in makefile accordingly
2. Compile and get zeta.x
3. Generate an input file input.nml from the sample sample_input.nml
4. Run zeta.x

Example

The testing example is using a small region in the Arabian Sea from Jan 1, 2009 in a 0.1deg POP/CICE simulation. The testing is perform on OLCF Rhea.
Input: ./example/momentum_2009-01-01_as.nc
Output: ./example/vorticity_2009-01-01_as.nc In addition, it uses grid files stored at /ccs/home/hewang/pop0_1/ on Rhea To test:

1. copy ./example/input.nml to ./
2. run ./zeta.x > log.out
3. compare sampled outputs in log.out with ./example/log.out

Required file

• Global grid and constants files

  • "ocn_static_grid.nc": 2-D fields ULAT, ULONG, UAREA, TLAT, TLONG, TAREA, DXU, DYU, z_t, z_w, HTN, HTE, HUS, HUW
  • "ocn_static_dz": 3-D fields DZT, DZU, TMASK, UMASK
  • "ocn_constant.nc": constants omega and grav

• Input files with momentum terms
  Depending on the calculation mode, variables needed include

  • UVEL, VVEL, WVEL
  • SSH, ADVU, ADVV, GRADX, GRADY, HDIFFU, HDIFFV, VDIFFU, VDIFFV,
  • UEU, UEV, VNU, VNV, WTU, WTV

Input files (containing momentum fields) and namelist (input.nml) specifications:

• Input files are supposed to be netCDF files with time record length = 1. The fields can be of daily, monthly or annually frequency.

• It is recommended to use at least daily averaged fields for nonlinear decomposition. Technically, the decomposition can be done for any frequency, but large errors are expected with monthly and annual fields.

• sec controls the period of mean/eddy decomposition and overall output frequency. It can be set by a list of options (see comments in sample_input.nml for details), for an arbitrary average length (annually, monthly, fixed number of days etc) within a year.

• Mean/Eddy decomposition section frequency meanfreq can only be "a" for annual fields and "a" or "m" for monthly fields

Climatology (yrnm_clm /= "")

If the daily, monthly and annual inpupt files are also averged across multiple years, the input files are considered to be "climatology".

• yrnm_clm is required. It is a wildcard option to replace the numerics for "year" in the filename. it can be set to be anything, though it is suggested to follow the form of "YYYY-YYYY".
• When mean/eddy decomposition is turned on (ifmeaneddy = T), ifmeanclm is automatically set to T.
• When nonlinear decomposition is turned on (ifdecomp = T), it is calculated only for the mean fields. In this case, ifmeaneddy is equivalent of being turned on.
  • In other words, when ifdecomp = F and ifmeaneddy = T, the outputs are curladv (mean), curlmet (mean), errsub (eddy); when ifdecomp = T, ifmeaneddy is ignored and the outputs are adv(u,v,w) (mean), twi(u,v,w) (mean), curlmet (mean), errsub (eddy)

For daily averaged files

1. Curl of momentum terms (output at the same frequency as input)
   ifcurl = T; ifdecomp = F;
   ifmeaneddy = F
   • func = c, func_m = NULL, func_me = NULL
2. Curl + nonlinear decomposition (output at the same frequency as input)
   ifcurl = T; ifdecomp = T; [flxtwi = T for flux twisting, and F for nonflux]
   ifmeaneddy = F
   • func = cdme, func_m = NULL, func_me = NULL
3. Curl + mean/eddy decomposition (output at section frequency)
   ifcurl = T; ifdecomp = F;
   ifmeaneddy = F; ifdecomposed = F; [ifmeanclm = T for clim mean and F for ann mean]
   • func = c, func_m = am, func_me = func
4. Curl + nonlinear decomposition + mean/eddy decomposition (output at section frequency)
   ifcurl = T; ifdecomp = T; [flxtwi = T for flux twisting, and F for nonflux]
   ifmeaneddy = T; ifdecomposed = T; [ifmeanclm = T for clim mean and F for ann mean]
   • func = cdme, func_m = dm, func_me = func

For monthly averaged files

• dalist_in should be set to any single negative value, which sets doylist to the negative month number
• Nonlinear decomposition not recommened
• meanfreq = 'm' or 'a'

For annually averaged files

• dalist_in and mnlist_in should be set to any single negative value, which sets doylist to 0
• Nonlinear decomposition not recommened
• meanfreq = 'a'

Other Tips

• For mean/eddy decomposiion, the mean nonlinear fields are calculated first and saved, which means the files can be reused for re-calculations.

• For fully-decomposed nonlinear term, output frequency is also the same as the input. Mean/eddy decomposion then reads and averages these files.

• Resume calculation when using daily-averaged input with fully-decomposed nonlinear term:

  • Streamlines:
    1. Generate files containing mean nonlinear field
    2. Files containing curl of each term and decomposed nonlinear term of each day
    3. Load and average files from 2 and output mean (from 1) and eddy (difference averged 2 and 1)
    • If the calculation is interrupted during 2:
      1. Turn off ifmeaneddy and reset lists of days, month and years and resume
      2. Turn off ifcurl and ifdecomp and turn on ifmeaneddy

Algorithm

See the appendix of 10.1175/JPO-D-20-0223.1