On Wednesday 17 April 2019, Grigory Rechistov via Imports wrote: > > In the PDF on pages 52-55 there are English descriptions of classes > used in the original raster: > https://wiki.openstreetmap.org/wiki/File:NMD_Produktbeskrivning_NMD20 >18Basskikt_v1_0.pdf . Note that we used not all of these classes, and > in fact merged several classes of forests into a fewer sets of groups > of OSM-tags.
That does not really help much - except for recognizing that the classes probably (based on the English description - which might be incomplete of course) do not match the tags in OSM in most cases, for example * landuse=farmland is not the same as "arable land" * landuse=grass is not the same as "vegetated other open land" - which is not limited to herbaceous vegetation and includes everything with >10% vegetation coverage (which frankly can also be lichen alone) * landuse=forest is not for every area with more than 10% crown cover in trees >5m height (which can be almost anything - gardens, parks, orchards, scrub - as well as grassland and heath with scattered trees obviously). Note these are just the "wishful thinking" specifications of course - the data does not actually have to represent them. How they actually want to distinguish 10% tree cover >5m high from lower woody vegetation based on multispectral imagery along is doubtful. Typically such classifications have error rates of at least 5-10 percent, often much higher. > This import is meant to convert > areas with *no data* into areas with *some data*. That is the standard "bad data is better than no data" argument. With that we could have completely tiled the world with data from some global land cover classification project a long time ago. Luckily for OSM we have refrained from doing that. -- Christoph Hormann http://www.imagico.de/ _______________________________________________ Imports mailing list [email protected] https://lists.openstreetmap.org/listinfo/imports
