NOTICE: This version of the NSF Unidata web site (archive.unidata.ucar.edu) is no longer being updated.
Current content can be found at unidata.ucar.edu.
To learn about what's going on, see About the Archive Site.
NOTICE: This version of the NSF Unidata web site (archive.unidata.ucar.edu) is no longer being updated.
Current content can be found at unidata.ucar.edu.
To learn about what's going on, see About the Archive Site.
Hi,I have discovered what I believe is a very curious bug in the FMRC. We have a ROMS ocean model including 4D variational data assimilation (4D-Var) that we are serving both averages and history output for. There are no issues with the averages files but history files trigger a strange bug in which the "bathymetry at RHO-points" variable (named h) is removed by the FMRC.
As part of the 4D-Var system, additional variables are written to the history files and 4 of them seem to be triggering this bug:
float DU_avg1(ocean_time, eta_u, xi_u) ;
DU_avg1:long_name = "time averaged u-flux for 2D equations" ;
DU_avg1:units = "meter3 second-1" ;
DU_avg1:time = "ocean_time" ;
DU_avg1:coordinates = "lon_u lat_u ocean_time" ;
DU_avg1:field = "DU_avg1, scalar, series" ;
DU_avg1:_FillValue = 1.e+37f ;
float DU_avg2(ocean_time, eta_u, xi_u) ;
DU_avg2:long_name = "time averaged u-flux for 3D equations
coupling" ;
DU_avg2:units = "meter3 second-1" ;
DU_avg2:time = "ocean_time" ;
DU_avg2:coordinates = "lon_u lat_u ocean_time" ;
DU_avg2:field = "DU_avg2, scalar, series" ;
DU_avg2:_FillValue = 1.e+37f ;
float DV_avg1(ocean_time, eta_v, xi_v) ;
DV_avg1:long_name = "time averaged v-flux for 2D equations" ;
DV_avg1:units = "meter3 second-1" ;
DV_avg1:time = "ocean_time" ;
DV_avg1:coordinates = "lon_v lat_v ocean_time" ;
DV_avg1:field = "DV_avg1, scalar, series" ;
DV_avg1:_FillValue = 1.e+37f ;
float DV_avg2(ocean_time, eta_v, xi_v) ;
DV_avg2:long_name = "time averaged v-flux for 3D equations
coupling" ;
DV_avg2:units = "meter3 second-1" ;
DV_avg2:time = "ocean_time" ;
DV_avg2:coordinates = "lon_v lat_v ocean_time" ;
DV_avg2:field = "DV_avg2, scalar, series" ;
DV_avg2:_FillValue = 1.e+37f ;
I don't know if it's significant but the long_name attribute does contain "time averaged". If I
remove these variables with ncks, "h" magically reappears in the FMRC. I tried removing with ncml
but while it did remove the variables it did not bring back "h". I believe this is because the ncml
is applied after at least part of the FMRC processing so "h" has already been discarded.
Cheers, David Robertson
thredds archives: