InitMIP Change Log

From CliC Wiki
Revision as of 09:45, 1 November 2016 by Hgoelzer (Talk | contribs)

Jump to: navigation, search

initMIP Dataset Preparation

This is a log of the changes made on Greenland initMIP submissions during the Summer/Fall of 2016. Datasets were downloaded from the searise ftp server and nco and python scripts were run to regrid, rename attributes, fix errors, etc.


ARC

  • PISM5KM is the latest submission. This supersedes PISM_5km.
  • The files were already in 5km spacing, so they did not need regridding.
  • The most recent submission (PISM5KM) did not have time for any of the variables. However, there was time with the earlier submission (PISM_5km). So, I took the time from the earlier submission and entered it into the new one.
  • I renamed some of the attributes to the standard names.
  • Note that "hfgeoubed" and "licalvf" do not have a time variable
  • Initially, there were no masks for any of the directories. Nick Golledge provided them on Sept 7th 2016
  • The masks did not have any time attribute associated with them. However, the mask variables are 3D and have a time dimension. Since I didn't have an older version of the masks to copy the time from, I copied the time from the "orog" files. They both have 21 time steps, so I think this should be o.k.
  • This experiment is missing a scalar file in the init directory.
  • For some reason, I wasn't able to remove the "_ChunkSize" attribute no matter what I did, so I just left it in there.
  • I tar'd the directory PISM5KM_CF and it is ready for upload.


AWI

  • (hg) added mask sftgif=sftgrf+sftflf


BGC

  • BISICLES1, BISICLES2, and BISICLES3 experiments were submitted.
  • The BISICLES grids were already regridded, so I didn't need to do anything with that. However, there were some attribute names and values that needed altering.

BISICLES1

  • Variable with name "year" was changed to "time". Values were converted to seconds and attributes: "axis", "units", and "calendar" were added.
  • Mapping variable was added, and "grid_mapping" attribute was added to all datasets.
  • For the masks ("sfrgrf", "sftflf", and "sftgif"), the units were changed to "dimensionless". Also, the mask "sftflf" was misspelled as "sftfif". I had to change the filename but I also had to change the name of the variable within the file as well.
  • BISICLES1 has two variables that are not in the initMIP list of variables(according to the wiki): "wvelmean" and "liacabf".
  • Initially, in ctrl, "orog" and "lithk" only had 1 time step. I contacted Vicky Lee, and she said this was a mistake, so she provided me with new files for BISICLES1 ctrl directory. I updated these in my local directory as well as on the ftp server.
  • Lat/lon variables were removed and replaced with X and Y.
  • Time in the "scalar" variable was converted to seconds.
  • Attributes "amr_name" and "coordinates" were removed from each variable.
  • The scalar file has a variable named, "tendacaf" was renamed to "tendacabf".
  • We received the scalar file for the ctrl at a later date. I downloaded that file
  • I tar'd up the directory with the version that is ready for upload: BGC_BISICLES1_5km.tar.gz As of 8/31/2016 I have not uploaded the file.

BISICLES2

  • BISICLES2 had the same issues as BISICLES1 above, so those things were corrected.
  • I tar'd up the directory with the version that is ready for upload: BGC_BISICLES2_5km.tar.gz As of 8/31/2016 I have not uploaded the file.

BISICLES3

  • BISICLES3 had the same issues as BISICLES1 above, so those things were corrected.
  • I tar'd up the directory with the version that is ready for upload: BGC_BISICLES3_5km.tar.gz As of 8/31/2016 I have not uploaded the file.


DMI

  • PISM0, PISM1, PISM2,PISM3, PISM4 experiments were submitted.
  • The PISM grids were already regridded, so I didn't need to do anything with that. However, there were some attribute names and values that needed altering.
  • Tar files for each of the corrected experiments have been created. As of 09.06.2016 I have not uploaded the tar files to the ftp server
  • Corrections made:
   *  "litemppbot" was misspelled and corrected to "litempbot"
   *  files named "sfrgrf..." were renamed to "sftgrf....".  
   *  The variable within the file was also renamed to "sftgrf..."
   *  Remaining changes were to "long_name" and "unit" attributes.
   *(hg) correct sign in libmassbf
   *(hg) dlithkdt in m/s not m/yr  divide by spy
  • Things to note that still remain with all submitted PISM experiments:
  *  missing file for "dlithkdt" in init directory
  *  missing file for "licalvf" in init directory
  *  hfgeoubed does not have a time variable
  *  The scalar variable has a time size of 300
  *(hg) libmassbf magnitude is still not correct!

IMAU

  • (hg) corrected amplitude of licalvf
  • (hg) dlithkdt in m/s not kg/s divide by 910.0

JPL

  • (hg) renamed dim name nTime,nx,ny --> time,x,y


LANL

  • (hg) extracted scalar init from first time step


LGGE

  • ELMER2 experiment was submitted. It needed to be regridded to 5km spacing
  • Time variable has been converted to "solar" seconds. The original data has time in 5 year increments. I copied the 5year time in seconds from an IMAUICE grid and multiplied by increments of 5.
  • Time is still type "double". When I tried to do a simple convert to float, it mysteriously removed increments of 6 seconds from each interval. I couldn't figure out why, so I left it as a double
  • The mask variables (sfrgrf, sftflf,sftgif): For the original data, it looks like it is a pure mask with values either 1 or 0. However, when you re-grid, because of the interpolation you will have fractional values. Therefore I left the data type as float and not short
  • It seems like CF-convention for mask units is "unitless", so I added that to the mask variables in #2. However, are these variables intended to be masks, or percent of area as the name suggests? The original values don't seem to be fractional, so this is a bit confusing.
  • The bedrock variable (topg) had some issues. It doesnt have time as a dimension (it only has it as a separate variable). therefore the script behaved badly and I had to re insert the time variable. The original data does not have time as a dimension in "topg", so this may be by design? On the initMIP wiki, it states that: "No bedrock adjustment in forward experiment" so this may be why?
  • "Tendency" is mis-spelled in standard name of dlithkdt and others. This needs to be fixed!
  • I ran the script, ELMER2_FixScalarTime.bash to fix the three scalar files for ELMER2. One weird thing is that the scalar file in the "init" folder has times of: -4,-3,-2,-1,0. I assume that is by design, so I just converted it to seconds.


LSCE

  • (hg) extracted scalar init from first time step
  • (hg) renamed 2d variables ctrl variable_5285 --> uvelmean
  • (hg) renamed 2d variables asmb variable_4732 --> uvelmean
  • (hg) extracted for each variable init is first time step in ctrl
  • (hg) corrected velocities from m'yr to m/sec
  • (hg) corrected masks sftgrf=sftgif; sftflf=0


MIROC

  • ICIES00 and ICIE01 were submitted. They were both regridded to 5km.
  • Ran MIROC_FixScalarTime.bash to fix the time on the scalar variable. It was in years and needs to be in seconds.
  • Ran MakeFCCompliant.bash to give the proper attribute names.
  • The mask file "sfgrf" has wrong variable name in the filename as well as the variable name itself. It should be sftgrf. I fixed the variable name, and did a manual renaming of the filename in both the new and original files.
  • Note that the variable hfgeoubed DOES NOT have a time variable.
  • The cleaned/regridded versions of the data were uploaded to the searise ftp site in folders named asmb_05, init_05, ctrl_05.
  • (hg) prepended init
  • (hg) corrected sign of hfgeoubed


MPIM

  • (hg) extracted scalar init from first time step


UCIJPL

  • (hg) corrected masks sftgif=sftgif>0; sftgrf=sftgif*sftgrf; sftflf=sftgif*sftflf


ULB

  • FETISH1 and FETISH2 experiments were submitted.
  • This submission has more scalar variables than other submissions. It has variables such as: lim, limnsw,tendacabf,scalars for asmb and variables: iareag, lim,limnsw,tendacabf,scalars for ctrl.
  • These masks and/or the scalar files are not uniform across all three directories. (i.e. some files are only in init, and may not be in the other 2)
  • hfgeoubed and topg did not have time associated with it.
  • On August 29th Sainan updated the sftgif files for the asmb and ctrl directories for FETISH1 and FETISH2. I downloaded these and regridded just that variable.
  • I also ran my code to sweep for common errors and attribute name errors. I corrected a few problems such as "scalars" instead of "scalar", "sfrgrf" instead of "sftgrf", etc.
  • I changed the time units to be "seconds since 2005-01-01 00:00:00" because that what the original files had. However, the original files have no actual time data (the data fields were empty). I copied the time from the IMAU submission and used those numbers.
  • I uploaded the corrected file to the searise ftp server and named the directories asmb_05, ctrl_05, init_05.
  • (hg) extracted scalar init from first time step
  • (hg) correct masks sftgrf=sftgif; sftflf=0
  • (hg) dlithkdt in m/s not m/yr divide by spy

VUB

  • (hg) corrected magnitude hfgeoubed
  • (hg) Temp in K not C
  • (hg) corrected calving magnitude
  • (hg) dlithkdt in m/s not kg/s divide by 910.0

Collective change

labs=(ARC BGC BGC BGC DMI DMI DMI DMI DMI ILTS_PIK IMAU IMAU IMAU LGGE LGGE LSCE MIROC MIROC MPIM MPIM VUB VUB)

models=(PISM5KM BISICLES1 BISICLES2 BISICLES3 PISM0 PISM1 PISM2 PISM3 PISM4 SICOPOLIS IMAUICE05 IMAUICE10 IMAUICE20 ELMER ELMER2 GRISLI ICIES00 ICIES01 PISM0COMPUTED PISM0INITMIP GISMHOM GISMSIA )

  • correct sign of licalv


labs=(ARC BGC BGC BGC DMI DMI DMI DMI DMI IMAU IMAU IMAU LGGE LGGE VUB VUB)

models=(PISM5KM BISICLES1 BISICLES2 BISICLES3 PISM0 PISM1 PISM2 PISM3 PISM4 IMAUICE05 IMAUICE10 IMAUICE20 ELMER ELMER2 GISMHOM GISMSIA )

  • correct sign of libmassbf


labs=(ARC AWI AWI BGC BGC BGC DMI DMI DMI DMI DMI ILTS ILTS_PIK IMAU IMAU IMAU JPL1 LANL LGGE LGGE LSCE MIROC MIROC MPIM MPIM UAF UAF UAF UAF UAF UAF ULB ULB VUB VUB)

models=(PISM5KM ISSM1 ISSM2 BISICLES1 BISICLES2 BISICLES3 PISM0 PISM1 PISM2 PISM3 PISM4 SICOPOLIS SICOPOLIS IMAUICE05 IMAUICE10 IMAUICE20 ISSM CISM ELMER ELMER2 GRISLI ICIES00 ICIES01 PISM0COMPUTED PISM0INITMIP PISM151 PISM152 PISM301 PISM302 PISM451 PISM452 FETISH1 FETISH2 GISMHOM GISMSIA )

  • rename variables sfrgrf --> sftgrf