Land Regridding

Regridding ELM/CLM land data files

Low Complexity

In this example, a directory of ne30 clm2.h0 files will be regridded. This will regrid all variables in the input files, using sub-grid-scale regridding, which uses the land fraction around coastal areas to better represent the values around complex coastal geometry.

#!/bin/bash
mapfile=map_ne30np4_to_cmip6_180x360_aave.20181001.nc   # path to mapfile
input_dir=clm2.h0                                       # path to input directory
output_dir=180x360                                      # path to output directory
land_file=<input file path>                             # this is the path to a single land file to pull the landfrac variable from

ncremap -m ${mapfile} -I ${input_dir} -O ${output_dir} --sgs_frc=${land_file}/landfrac

Medium Complexity

In this example, the same regridding operation as above is performed, but only on the DEADSTEMC and CDWC variables. Running with specific variables significantly speeds up the run and reduces output files size, as the rest of the variables are ignored.

#!/bin/bash
mapfile=map_ne30np4_to_cmip6_180x360_aave.20181001.nc   # path to mapfile
input_dir=clm2.h0                                       # path to input directory
output_dir=180x360                                      # path to output directory
vars=DEADSTEMC,CDWC                                     # the variables must be in a comma separated list with no spaces
land_file=<input file path>                             # this is the path to a single land file to pull the landfrac variable from

ncremap -v ${vars} -m ${mapfile} -I ${input_dir} -O ${output_dir} --sgs_frc=${land_file}/landfrac

High Complexity

In example, the regridding operation is performed and the selected variables are output in single-variable-per-file time-series files. The output will be compressed and deflated. The model data starts at 1850-01 and ends at 2014-12.

This example produces time-series files, to produce monthly and annual climatologies remove the “–ypf N” (years-per-file) flag, and optionally add the “-a sdd” (seasonally discontiguous december) flag if you don’t want to include the n-1th december from the climo.

When choosing variables to extract into time-series, its only possible to use variables with both a time and space dimension.

#!/bin/bash
mapfile=map_ne30np4_to_cmip6_180x360_aave.20181001.nc  # path to mapfile
input_dir=clm2.h0                                      # path to input directory
output_dir=180x360                                     # path to output directory
vars=DEADSTEMC,CDWC                                    # the variables must be in a comma separated list with no spaces
start=1850                                             # the first year of model data
end=2014                                               # the last year of model data
land_file=<input file path>                            # this is the path to a single land file to pull the landfrac variable from
flags="-7 --dfl_lvl=1"                                 # format and deflation

ncclimo \
  ${flags} \
  --var=${vars} \
  --yr_srt=${start} \
  --yr_end=${end} \
  --ypf=50 \
  --input=${input_dir} \
  --output=${output_dir} \
  --map=${mapfile} \
  --sgs_frc=${land_file}/landfrac