Proposal talk:Landcover Barren

Existing tags such as natural=sand / bare_rock / shingle / scree should be used

At OSM we want to map specific features that are real and current and that individual mappers can verify on the ground or with good quality aerial imagery.

Landcover=barren does not meet these standards. It's a mix of several different features that already have established tags: natural=sand, natural=beach, Tag:natural=scree, natural=shingle, natural=mud, natural=glacier and natural=scrub.

It's best to leave an area unmapped if you are not certain what type of soil is in that area. But often you can see if it is sand, small rocks, or large rocks: this can be tagged natural = sand or scree or shingle or bare_rock as appropriate.

In most tropical regions, if a patch of forested land has been recently cut or burned, it will quickly regrow into the previous vegetation type.

Forested areas take several years to grow, but even after 6 months the ground will no longer be barren, but covered in small herbs, young trees and shrubs; this can be tagged natural=scrub. In OSM we generally try to map the long-term features that should be on a map, not transient features that only last a few months (for example, temporary road closures should not be mapped, unless they will last most of a year or longer).

Similarly, we do not map areas of snow that last for 6 to 9 months out of the year, but we do map natural=glacier locations, which change slowly over many years.

Some mappers also use natural=fell or natural=tundra for alpine areas with low or sparse vegetation and natural=desert for arid regions in the tropics with little vegetation. These tags are less specific and are usually not the best choice, but would be better than landcover=barren and they are already established.

This was also discussed on the Tagging mailing list. See the discussion: https://lists.openstreetmap.org/pipermail/tagging/2019-March/043702.html --Jeisenbe (talk) 04:35, 10 April 2019 (UTC)

General features have their advantages, especially for the remote mapping. For explaining better, I will take example of mapping deforestation in Amazonian rainforest. In the first place, it is very hard to obtain on-site information through citizen science in this case. Luckily for humanity, Amazon is still considered as a wild-life environment, but this means that presence of human is not common there, and also it indicates danger for human beings. For these reasons, we cannot expect many volunteers on-site. So, if we leave out the possibility of local mapping, the remote mapping remains. However, remote mapping is constrained by the cloud coverage. Namely, rainforests are known for the frequent cloud coverage (their name is based on this fact in one way), therefore having high-quality and cloud-free imagery might happen extremely rarely. In a case of scarce information, it would be good to omit some constraints related to information acquirement, extending in this way the possibility to map more.

Speaking about citizen science, it is not expected that volunteers are skilled to distinguish such a specific classes like shingle and scree from satellite imagery, even if it is high-quality. Of course, some of the volunteers are experienced and trained for photo-interpretation, but much larger number of volunteers are not so skilled. Thus, it is expected to have the quite a few errors in the tagging.

If general class like barren land is not included in a tag, it might mean that the Amazon will remain unmapped (or mapped incorrectly) for many years to come. More specific tags such as natural=sand and natural=beach are very welcome, but in their absence any other (less specific, but correct) information linked to Amazon deforestation is appreciated in understanding underlying processes in order to prevent and stop them.

Furthermore, specific type of barren soil that is a result of clear cutting is very prone to erosion. Soil erosion in rainforests is expected as heavy rains are usual here. This can significantly delay, if not completely interrupt, process of revegetation. This means that the period in which this type of barren land will be revegetated is uncertain, therefore mapping it (by including it into the landcover=barren ) can be significant.

Due to the unpredictable duration of barren land cover, the tag landcover=barren be accompanied with key {tag|fixme}} to point out uncertainty about duration of this land cover type. i.e. landcover=barren, fixme=”Prone to changes. Revise feature if older than 6 months” or key end_date=* to let user’s know when the tag is might not be valid anymore and needs to be checked in order to confirm its validity.

In this way, if feature is revised regularly, period of revegetation can be observed as well, and estimate how serious the consequences of deforestation are.

Barren is intended to describe soil without vegetation therefore sand, rock, cliff, debris, mud etc. Glaciers were not meant to be included in this class as they are not soil type. Example is given for Amazonian rainforest, but can be applied to almost any forest subject to deforestation. In addition to deforestation, contamination of the soil can be another example, where access to the contaminated sites is dangerous for health. --LorenzoStucchi (talk) and Gorica7 (talk) 9:45, 17 April 2019 (UTC)