process second batch of SCIP data

actions/scip-remove-scalar/DESCRIPTION.md

@@ -0,0 +1,40 @@
+# SCIP data update
+
+Scripts in this folder do two things:
+* update metadata by changing the domain from `bccoast+fraser` (the domain used in the old dataset) to `bccoast+ungauged+fraser`
+* remove a CRS scalar variable that our indexing scipt did not parse correctly
+
+## update domain
+
+First run the file-renaming bash script in the directory containing the files to rename all of them. While filenames are not load-bearing, leaving them with the wrong domain could be confusing.
+
+```bash
+#! /bin/sh
+
+for n in *.nc; do
+	x="`echo $n | sed s/bccoast\+fraser/bccoast-ungauged-fraser/`"
+	mv "$n" "$x"
+done
+```
+
+Secondly update domain attributes with `ncatted`:
+```bash
+for file in ./*.nc ; do echo $file ; ncatted -O -a domain,global,o,c,"bccoast+ungauged+fraser" $file ; ncatted -O -a hydromodel__domain,global,o,c,"bccoast+ungauged+fraser" $file ; ncatted -O -a hydromodel__forcing_domain,global,o,c,"bccoast+ungauged+fraser" $file ; done
+```
+
+
+# remove scalar variable
+
+NetCDF supports "scalar variables" - variables with no data. The CF Standards indicate that a scalar variable is in fact the preferred way to indicate the grid mapping of a dataset - the various mapping parameter are stored in attributes of the variable.
+
+However, in this case, our indexing script did not understand the grid mapping, and I elected to just remove it, since it was WGS 84, the default assumption of the SCIP portal. A python script was needed to remove it, because NCO does not recognize scalar variables as variables that can be edited or deleted.
+
+After removing the CRS variable with the python script, references to it were removed from attributes with `ncatted`.
+
+```bash
+for file in ./*.nc ; do python3 remove_variable.py $file crs ; done
+
+for file in ./*.nc ; do echo $file; ncatted -a grid_mapping,,d,, $file $file.tmp ; mv $file.tmp $file; done
+```
+
+These tweaks were all these files needed. Their metadata was otherwise perfect.