mask_fillvalues usage

[pepcos]

Hi,
I am working with observational monthly data, and I'd like to check if there is more than 9 unmasked months per year before computing the annual means. If 9 months or less are present in a given year, I'd like the 12 months to be masked before computing the annual mean. Is this something that could be accomplished with the mask_fillvalues preprocessor? I've been trying to understand how it works and I'm still not quite sure if it can provide what I need.

Thanks!

[bouweandela]

@ESMValGroup/scientific-lead-development-team Any suggestions anyone?

[zklaus]

I am also not sure, but could you try the following and let us know how it goes:

mask_fillvalues:
  threshold_fraction: 0.8
  time_window: 12

NB: Make sure your data starts in a January...

[pepcos]

I've created this .nc (starting in the J month) to test the function and the FM months from year 1, the FMA months from year 2 and the FMAM months from year 3 have been masked:

dimensions:
        time = 48 ;
        lat = 1 ;
        lon = 1 ;
        bnds = 2 ;
variables:
        double tos(time, lat, lon) ;
                tos:_FillValue = 1.e+20 ;
                tos:standard_name = "sea_surface_temperature" ;
                tos:long_name = "Sea Surface Temperature" ;
                tos:units = "degC" ;
                tos:cell_methods = "time: mean" ;
        double time(time) ;
                time:axis = "T" ;
                time:bounds = "time_bnds" ;
                time:units = "days since 1850-1-1 00:00:00" ;
                time:standard_name = "time" ;
                time:long_name = "time" ;
                time:calendar = "gregorian" ;
        double time_bnds(time, bnds) ;
        float lat(lat) ;
                lat:axis = "Y" ;
                lat:bounds = "lat_bnds" ;
                lat:units = "degrees_north" ;
                lat:standard_name = "latitude" ;
                lat:long_name = "latitude" ;
        float lat_bnds(lat, bnds) ;
        float lon(lon) ;
                lon:axis = "X" ;
                lon:bounds = "lon_bnds" ;
                lon:units = "degrees_east" ;
                lon:standard_name = "longitude" ;
                lon:long_name = "longitude" ;
        double lon_bnds(lon, bnds) ;
data:
 tos =
  27.717640057806,
  _,
  _,
  27.7431312627977,
  26.5711427666333,
  25.6364652502723,
  25.0604854041511,
  24.0554036073369,
  24.3704020690196,
  25.0088960607156,
  25.8707415628147,
  26.6312291783994,
  28.0374939871062,
  _,
  _,
  _,
  26.430334205727,
  25.6898753940644,
  24.5925397125184,
  23.2712455871171,
  24.301818588923,
  24.8110357553041,
  25.306293452285,
  26.8521529549938,
  27.6551259122313,
  _,
  _,
  _,
  _,
  26.2452195028113,
  25.4719862847309,
  24.4930026263605,
  24.544591969796,
  25.1187510155606,
  25.9284002407721,
  27.0287705895789,
  28.2025798860998,
  28.452636468399,
  28.5017980780258,
  28.1959036181258,
  27.4299536014715,
  26.2239768319849,
  25.0131445948809,
  24.2702580494095,
  24.3867892722286,
  25.0859766091428,
  25.5454252324449,
  26.6803907880262,

If I run it through the preprocessor you suggest, the output in the preproc directory is unchanged (no masking is performed).