Oddities in single-point mapalgo defaults

[samsrabin]

Various mapalgo settings are bilinear by default, with some single-point sites set to nearest-neighbor. It's not clear:

• Why those sites are set to nearest-neighbor
• Why other sites aren't

popdensmapalgo has some sites missing (e.g., 1x1_smallvilleIA):

CTSM/bld/namelist_files/namelist_defaults_ctsm.xml

Lines 1818 to 1825 in a943377

	<popdensmapalgo use_cn=".true." >bilinear</popdensmapalgo>
	<popdensmapalgo use_cn=".true." hgrid="1x1_brazil" >nn</popdensmapalgo>
	<popdensmapalgo use_cn=".true." hgrid="1x1_mexicocityMEX" >nn</popdensmapalgo>
	<popdensmapalgo use_cn=".true." hgrid="1x1_vancouverCAN" >nn</popdensmapalgo>
	<popdensmapalgo use_cn=".true." hgrid="1x1_urbanc_alpha" >nn</popdensmapalgo>
	<popdensmapalgo use_cn=".true." hgrid="1x1_asphaltjungleNJ" >nn</popdensmapalgo>
	<popdensmapalgo use_cn=".true." hgrid="5x5_amazon" >nn</popdensmapalgo>

Same for lightngmapalgo:

CTSM/bld/namelist_files/namelist_defaults_ctsm.xml

Lines 1724 to 1731 in a943377

	<lightngmapalgo >bilinear</lightngmapalgo>
	<lightngmapalgo hgrid="1x1_brazil" >nn</lightngmapalgo>
	<lightngmapalgo hgrid="1x1_mexicocityMEX" >nn</lightngmapalgo>
	<lightngmapalgo hgrid="1x1_vancouverCAN" >nn</lightngmapalgo>
	<lightngmapalgo hgrid="1x1_urbanc_alpha" >nn</lightngmapalgo>
	<lightngmapalgo hgrid="1x1_asphaltjungleNJ" >nn</lightngmapalgo>
	<lightngmapalgo hgrid="5x5_amazon" >nn</lightngmapalgo>

Same for urbantvmapalgo:

CTSM/bld/namelist_files/namelist_defaults_ctsm.xml

Lines 1877 to 1884 in a943377

	<urbantvmapalgo >nn</urbantvmapalgo>
	<urbantvmapalgo hgrid="1x1_brazil" >nn</urbantvmapalgo>
	<urbantvmapalgo hgrid="1x1_mexicocityMEX" >nn</urbantvmapalgo>
	<urbantvmapalgo hgrid="1x1_vancouverCAN" >nn</urbantvmapalgo>
	<urbantvmapalgo hgrid="1x1_urbanc_alpha" >nn</urbantvmapalgo>
	<urbantvmapalgo hgrid="1x1_asphaltjungleNJ" >nn</urbantvmapalgo>
	<urbantvmapalgo hgrid="5x5_amazon" >nn</urbantvmapalgo>

lai_mapalgo has some sites repeated (e.g., 1x1_brazil) and some sites missing (e.g., 1x1_smallvilleIA):

CTSM/bld/namelist_files/namelist_defaults_ctsm.xml

Lines 1674 to 1687 in a943377

	<lai_mapalgo>bilinear</lai_mapalgo>
	<lai_mapalgo hgrid="1x1_brazil" >nn</lai_mapalgo>
	<lai_mapalgo hgrid="1x1_mexicocityMEX" >nn</lai_mapalgo>
	<lai_mapalgo hgrid="1x1_vancouverCAN" >nn</lai_mapalgo>
	<lai_mapalgo hgrid="1x1_urbanc_alpha" >nn</lai_mapalgo>
	<lai_mapalgo hgrid="1x1_asphaltjungleNJ" >nn</lai_mapalgo>
	<lai_mapalgo hgrid="5x5_amazon" >nn</lai_mapalgo>
	<lai_mapalgo hgrid="1x1_brazil" >nn</lai_mapalgo>
	<lai_mapalgo hgrid="1x1_mexicocityMEX" >nn</lai_mapalgo>
	<lai_mapalgo hgrid="1x1_vancouverCAN" >nn</lai_mapalgo>
	<lai_mapalgo hgrid="1x1_urbanc_alpha" >nn</lai_mapalgo>
	<lai_mapalgo hgrid="1x1_asphaltjungleNJ" >nn</lai_mapalgo>
	<lai_mapalgo hgrid="5x5_amazon" >nn</lai_mapalgo>

[samsrabin]

From CTSM SE meeting today: Might be because single points should be thought of as POINTS, not gridcells. Effective gridcell size shouldn't give different answers. That's what's being done in subset_data.

Is there a better way to do this than having to enumerate every single point we want to do this for? Could do it in the build-namelist script instead. Have it set the mapalgo to nn if it's a single-point, as identified by nx and ny. (Can probably let 5x5amazon be bilinear.)

@ekluzek will add to this issue with an idea of how a solution might look.