Note: I’ve done large data imports of glacier data into OSM from GLIMS (http://glims.colorado.edu/glacierdata/, also osm.wiki/GLIMS_Glacier_Database), specifically for the area called the Wakhan Corridor in NE Afghanistan that I know relatively well. All these changesets contain individually checked changes, polygon by polygon. I’ve also done improvements on the ‘poorly vectorization’ polygons before import, as this is a known issue with much of GLIMS data.