alpha-ctsm5.4.CMIP7.18.ctsm5.3.082: fsurdat validation tool

tools/mksurfdata_esmf/validate_fsurdat_files.sh

@@ -0,0 +1,98 @@
+#!/bin/bash
+
+# Validation/verification of new fsurdat files
+# --------------------------------------------
+# WRITTEN by slevis after discussions with ekluzek.
+# DATE of original script 2025/11/11.
+#
+# LOCATION: I first used this script in the directory
+# .../inputdata/lnd/clm2/surfdata_esmf/ctsm5.4.0/validation
+#
+# CAVEAT: For new CTSM versions, use this script as a template with the
+# understanding that aspects of the code will need to change.
+#
+# PURPOSE and DETAILS
+# -------------------
+# This script
+# 1) uses cprnc to compare ctsm5.4 versus ctsm5.3 files by generating
+# cprnc.out files.
+# 2) greps for fields with differences (RMS or NORMALIZED) that are >=1.
+# The strictest grep threshold that I found isolates potentially
+# unexpected changes is E-03. I determined this empirically using two
+# types of problematic fsurdat files from the recent past:
+# - No LAI, SAI, and heights for pfts 15 and 16.
+# - No soil textures in parts of the world in unstructured grids.
+# These "unexpected" fields appear alongside expected diffs (discussed
+# below) when grepping for E+. Grepping for less than E-03 starts to
+# capture fields with smaller differences and is likely to miss
+# unexpected problematic fields.
+#
+# Step after running the script
+# -----------------------------
+# Interactively and iteratively build this grep command to confirm that
+# all fields in the script's grep output are expected. This list of
+# fields is ctsm5.4-specific:
+# >>> grep NORM grep_E+_surfdata_cprnc.out | grep -v ROOF | grep -v WALL | grep -v URBAN | grep -v BUILDING | grep -v abm | grep -v CANYON | grep -v CONST_HARVEST | grep -v ROAD | grep -v UNREPRESENTED_PFT | grep -v PCT_NATVEG
+#
+# If the grep command reveals unexpected fields, investigate.
+#
+# -----------------------------
+#
+# Separate subjective comparison
+# ------------------------------
+# >>> ncdiff surfdata_new.nc surfdata_old.nc surfdata_new_vs_old.nc
+# >>> ncview surfdata_new_vs_old.nc
+# Focus on fields with larger RMS diffs in the cprnc output.
+# For ctsm5.4 I got approval for expected diffs from
+# - Keith Oleson for diffs in urban fields
+# - Peter Lawrence for diffs in const_harvest, unrep_pft, pct_natveg, pct_nat_pft
+# - Fang Li (a few months ago) for diffs in abm.
+# ncvis works like ncview for unstructured grids (e.g. ne30), though it
+# seems to crash when I give it a "diff" file generated by ncdiff.
+#
+# Another validation step
+# -----------------------
+# Run mksurfdata_esmf with a different number of processors and confirm
+# bit-for-bit same results.
+
+echo "starting grids loop"
+
+# The first loop of grids has different ctsm5.3 date stamp.
+# Skip ne3np4.pg2 as present only in ctsm5.4.
+grids=("ne3np4")
+
+for grid in "${grids[@]}"
+
+   do /glade/campaign/cesm/cesmdata/cseg/tools/cime/tools/cprnc/cprnc ../surfdata_$grid\_hist_1850_78pfts_c251022.nc ../../ctsm5.3.0/surfdata_$grid\_hist_1850_78pfts_c240925.nc >& surfdata_$grid\_hist_1850_78pfts_ctsm5.4v5.3.cprnc.out
+   /glade/campaign/cesm/cesmdata/cseg/tools/cime/tools/cprnc/cprnc ../surfdata_$grid\_hist_2000_78pfts_c251022.nc ../../ctsm5.3.0/surfdata_$grid\_hist_2000_78pfts_c240925.nc >& surfdata_$grid\_hist_2000_78pfts_ctsm5.4v5.3.cprnc.out
+   echo "done $grid"
+
+done
+
+# Second loop of grids.
+grids=("4x5" "10x15" "0.9x1.25" "1.9x2.5" "mpasa120" "mpasa480" "ne16np4.pg3" "ne120np4.pg3" "ne3np4.pg3" "ne30np4" "ne30np4.pg2" "ne30np4.pg3")
+for grid in "${grids[@]}"
+
+   do /glade/campaign/cesm/cesmdata/cseg/tools/cime/tools/cprnc/cprnc ../surfdata_$grid\_hist_1850_78pfts_c251022.nc ../../ctsm5.3.0/surfdata_$grid\_hist_1850_78pfts_c240908.nc >& surfdata_$grid\_hist_1850_78pfts_ctsm5.4v5.3.cprnc.out
+   /glade/campaign/cesm/cesmdata/cseg/tools/cime/tools/cprnc/cprnc ../surfdata_$grid\_hist_2000_78pfts_c251022.nc ../../ctsm5.3.0/surfdata_$grid\_hist_2000_78pfts_c240908.nc >& surfdata_$grid\_hist_2000_78pfts_ctsm5.4v5.3.cprnc.out
+   echo "done $grid"
+
+done
+
+# Third loop of grids.
+# Skip mpasa30 as present only in ctsm5.4.
+# Skip 1850 as only 2000 is present.
+grids=("mpasa3p75" "mpasa15" "mpasa60")
+for grid in "${grids[@]}"
+
+   do /glade/campaign/cesm/cesmdata/cseg/tools/cime/tools/cprnc/cprnc ../surfdata_$grid\_hist_2000_16pfts_c251022.nc ../../ctsm5.3.0/surfdata_$grid\_hist_2000_16pfts_c240908.nc >& surfdata_$grid\_hist_2000_78pfts_ctsm5.4v5.3.cprnc.out
+   echo "done $grid"
+
+done
+
+# grep for E+ to catch larger diffs.
+for file in surfdata_*cprnc.out
+   do grep -H NORM $file | grep 'E+' >> grep_E+_surfdata_cprnc.out
+done
+
+exit