Proposal talk:Leaftype

From OpenStreetMap Wiki
Jump to navigation Jump to search

Caution

Whilst fully in agreement with what these tags seek to achieve. I'm not sure that having more tags helps. I'm against proliferating tags when there is already a tag which broadly speaking does the job, but needs a certain degree of refinement.

Broadly wood=* should be used for Coniferous or Broad-leaved woodland or mixed (both types). wood=deciduous is clearly an anomaly which should gradually be encouraged to disappear. Additional tags such as leaftype or leafcycle may help in this process, but unless there is extensive support from both editors & renderers I doubt it.

Note that there are coniferous trees with non-needle like leaves (Gingko balboa, comes to mind), and probably vice-versa. All leaves are ultimately deciduous, in that they fall. We are concerned with seasonally determined leaf fall. In practice a simple tag deciduous=* supports this. SK53 (talk) 13:05, 29 April 2014 (UTC)

You are fully right. There are two ways: installing new tags, while deprecating old ones, or changing the value of existing tags, while installing additional tags. In both cases you need extensive support from mappers, but we don't tag for renderers. I don't know which way is the best, but I prefer new tags to avoid confusion. Shurely this is a gradually process.
I don't see the advantage in a new tag deciduous=*. Do you want to tag evergreen as deciduous=no?--Rudolf (talk) 06:58, 30 April 2014 (UTC)
Why not, it's just not necessary: broadly speaking we only need to mark deciduous woodland differently (most other woodland classifications need to be supported by other tags). Firstly deciduous=yes is already in use, secondly, I have raised & discussed the issue at length but most importantly, it is trivial to migrate from wood=deciduous to deciduous=yes without risk of misinterpreting the original tagging. This is non-trivial because there are over 400k objects so tagged. In addition there is already a tag trees=broad-leaved with over 18,000 instances (almost entirely from Corine imports). Tag proliferation is always a problem, even if the tag is reasonably sensible for a number of reasons: user familiarity, tool support, renderer support. Another worry with leaftype is that it looks more technical than the current use of deciduous which I presume most people assume is synonymous with broad-leaved woodland and therefore may not be used. SK53 (talk) 19:59, 4 May 2014 (UTC)
If the majority of mappers prefer a migration from wood=evergreen to deciduous=no, I will accept it. Please take in consideration, that then every missing tag deciduous=* implies that the forest is evergreen.
The tag trees=broad-leaved is nowhere documented. trees=* is used for the kind of trees in orchards. IMHO the value broad-leaved is mismatching.
A clear definition should not base on your presumption what other people assume. --Rudolf (talk) 05:51, 5 May 2014 (UTC)
It might not be documented, but a simple inspection of the relevant CORINE codes and the use of the tag shows that's what it means Furthermore it should be possible to check with the people who wrote the code for the French CORINE import. I have nothing to do with it, but it is not possible to ignore actual usage of existing tags. I assume the choice was made to use an existing tag rather than create a new one. Semantically absence of a deciduous tag is no better or worse than absence of wood or leaftype. The idea of deciduous and other tag usage is to keep close to words used in normal speech: so Deciduous trees/woods, Broad-leaved (Laubwald), Coniferous (Nadelwald) etc. Not every mapper will be a specialist and it's important to try and use terms which will prompt recognition and familiarity. For instance my natural response to leaf type would be to expect pinnate or simple as valid values! Lastly how do you identify a majority of mappers: wiki users let alone wiki editors are a very small minority. SK53 (talk) 14:14, 8 May 2014 (UTC)
A mass-import must be thoroughly discussed before importing. I don't know where the discussion took place. Perhaps you can show me. IMO a mass-import should respect the convention of the wiki, not otherwise.
I also prefer a close-to-speech tagging. IMO "broadleaved" and "needleleaved" are easy to determine by every mapper. The determination of "deciduous" is not that easy. Sometimes you need special knowledge.
I don't have a definition of "majority of mappers". Perhaps you can help me. --Rudolf (talk) 05:55, 9 May 2014 (UTC)
The French Corine Import did not use the tag trees=broad-leaved. We used the already existing tag wood=deciduous/coniferous/mixed. I dont know witch country imported trees=broad-leaved.
trees=* was introduce with landuse=orchard to have a key less precise than species=* to describe an orchard, and more descriptive of what is on the ground than produce=*, crop=*
I'm happy that there is a try to clarify the jungle of trees, wood, type, deciduous, broad-leaved , palm, coniferous, evergreen. The set of tags : leaf_type=* + leaf_cycle=* is a good way. I like the reference to phenology. Maybe a little hard for beginners. But being precise is not easy.
And, maybe, a little too close to the leave question... Can we imagine a set of tag expressing the kind of vegetation, with a prefix like vegetation' (or tree, but it would restrict the use) vegetation:cycle, vegetation:shape=palm/..., vegetation:density=dense/sparse, vegetation:height ?
And, it would be helpful to have some examples based on the main forest types (alpine forest, equatorial forest...) People learns more with examples than with theory.
--FrViPofm (talk) 13:38, 18 May 2014 (UTC)
IMO the use of "vegetation" is possible but too complicated. Then we need "vegetation:leaf_type" and "vegetation:leaf_cycle" because the single use of type or cycle can refer to the leaves, the trunc or the roots. I want to keep the tags as simple as possible. leaf_type=* can be determined by every mapper. The key leaf_cycle=* is additional and the determination not common knowledge in every case. --Rudolf (talk) 06:45, 20 May 2014 (UTC)

SK53 ignores tag voting procedures. I see you changed your principles. Congratulation. --Rudolf (talk) 11:39, 6 June 2014 (UTC)

leaf_cycle values

After reading the proposal a few more times, I feel that the leaf_cycle values don't feel perfect yet. I see two issues with them: 1. Why is there no "mixed" tag as with leaf_type? 2. The meaning of the "semi_" values is unclear without reading the definition. I don't have a suggestion ready, but could they be exchanged for values that describe the meaning more clearly? --Tordanik 19:41, 5 May 2014 (UTC)

This values refer to the LCCS. The LCCS include also a value "mixed" for leaf phenology. The definition is "This category is limited exclusively to a layer with a mixture of broadleaved deciduous and needleleaved evergreen vegetation". IMO this value refers also to the leaf type. So I shifted this value to "leaf_type". This implies that the leaf phenology is also mixed. The average mapper can use this tag without problem. For most maps it is sufficient to tag the leaf type.
The attribution of the phenology is not trivial. Sometimes you need special knowledge. The LCCS-definition of "semi-deciduous" is "This term applies to a combination of broadleaved deciduous that is dominant and broadleaved evergreen being more than 25 percent." I doubt whether this is easier to understand.
I suggest to use the values of the most common classification systems, instead of creating new ones. --Rudolf (talk) 21:44, 5 May 2014 (UTC)
Meanwhile I searched several maps with forests and found no entry with "mixed leaf phenology". The value "mixed" is theoretical possible, but in life praxis subordinated. If we use the LCCS-definition of "semi-deciduous" and "semi-evergreen" then we have two kind of mixed phenology. IMHO this may be sufficient. --Rudolf (talk) 21:07, 6 May 2014 (UTC)
To be honest, I still don't understand the "Wikipedia" definitions of the semi_* values. Are these statements about individual plants or about a group of plants? I believe I do understand the LCCS definition of "semi-deciduous" that you quoted, although it appears to say something completely different than the Wikipedia ones? --Tordanik
At first sight, I considered wikipedia to be generally understandable. At second sight, this is not the case. You are right and I will remove the wikipedia concept. So we can keep a close correlation to LCCS. --Rudolf (talk) 06:01, 9 May 2014 (UTC)
Ok, that's some progress and I would find this acceptable now. However, if you are open to some more feedback, I still wonder why
  1. the definition of the semi values mixes broadleaved and deciduous (isn't that what we trying to get away from?)
  2. the values use "semi" (which means "half" and would thus suggest 50%, not 75%) rather than e.g. mostly_evergreen and mostly_deciduous
  3. the scale from 0% to 100% is separated into 3 values with leaf_type and into 4 values with leaf_cycle
--Tordanik 14:16, 9 May 2014 (UTC)
The distinction of the semi-values is still complex. Perhaps we must distinguish between single plants and habitats.
I found following descriptions: "Semi-deciduous is a botanical term which refers to plants that lose their foliage for a very short period, when old leaves fall off and new foliage growth is starting". - "Plants may be referred to as semi-evergreen when they lose most, but not all, of their foliage for a fraction of the year." I don't know whether these descriptions are right.
LCCS adopted these terms but gave it a new definition, with mixed broadleaved phenology. The LCCS definition is used for landcover usage and makes only sense for habitats, not for single plants.
Perhaps we can keep the botanic definition for single plants or groups of identical plants. Here the "semi" makes sense. Then we need a new value for landcover-usage. As you suggested "mixed" may be a solution, or "dominant_decidiuos" or "mostly_deciduous". I agree that the confinement to "broadleafed" is not needed.
I don't understand your third point. Do you think, we need more leaftypes?
--Rudolf (talk) 16:25, 10 May 2014 (UTC)
Thank you for your explanations, the different definitions mare more sense to me now. With that knowlege, it would probably be best to keep the semi_* values in their meaning for describing individual plants (or groups of plants that share those properties), and invent one or more values for mixed groups of plants. I like the idea of just allowing "mixed" for both keys best, due to its simplicity.
My third point is obsolete now that I understand the definitions better. --Tordanik 18:47, 10 May 2014 (UTC)
Thanks for your critical questioning. Without that, there will be no improvement.--Rudolf (talk) 09:10, 11 May 2014 (UTC)

Documenting changes

I appreciate the work you've put into this, something I have tried and abandoned in the past. I would consider including photos (closes ups and wide view) or all the features, leaf types etc so non-experts can quickly familiarise themselves without also delving into wikipedia and other documents. Also a diagram mapping the old tags with the new tags where possible could also be useful. Think about documenting any edge cases such as palm trees, josuha trees etc. And getting this translated into other languages (sometimes meanings are lost in translation). Perhaps try a bit of the tagging in your local area (keep the old tags in place) and make a 'trees' map in Tilemill/mapbox to show how the old and new tags can be rendered and handled etc and what the benefits will be. But ultimately, it could be a long path to adoption...but I don't think anyone is really happy with the existing system. --Hawkeyes 17:24, 10 May 2014 (UTC)

Thanks for your support and suggestions. I will add some items. IMO a proposal page cannot include every edge case, furthermore I'm no botanist.--Rudolf (talk) 09:14, 11 May 2014 (UTC)

Terminology

It has not been explained why common terms "broad-leaved" and "needle-leaved" should be written without a hyphen.

This proposal uses the terminology of the "Land Cover Classification System (LCCS) by FAO".
The notation "broadleaved" is common usage in the oxford dictionary. The notation "needleleaved" is not mentioned. Maybe "needle-leaved" is a grammatical better term. When you look at the google-search you will find "needleleaved" 1.320.000 times and "needle-leaved" 40.100 times. Therefore "needleleaved" is common usage in the world-wide-web.
I prefer a tagging scheme relating to a worldwide used classification system. I prefer also a minimum usage of hyphens and underlines, due to a simple tagging scheme and to avoid typing errors. --Rudolf (talk) 09:31, 26 May 2014 (UTC)
My "Langenscheidt New College German Dictionary"(~760pages / 1995) does neither know "broad-leaved" nor "needle-leaved" but "deciduous tree" and "conifer / coniferous tree". leo.org knows "broad-leaved" but not "needle-leaved". So the question: are these tags a really english nouns?--Hfst (talk) 19:55, 26 May 2014 (UTC)
See my answer above. --Rudolf (talk) 05:56, 27 May 2014 (UTC)

Grundtags

Falls es zu einer Einigung und Zulassung kommen sollte, sollten aber die Grundtags (wood, tree, forest, tree_line, ...) überarbeitet werden und leaf_type/leaf_cycle eingearbeitet werden. Wichtig finde ich schon die Unterscheidung - auch zur Orientierung und zum Aufteilen riesiger Waldgebiete. --Geri-oc (talk) 06:37, 27 May 2014 (UTC)

+1 --Rudolf (talk) 10:11, 27 May 2014 (UTC)

Implementation

Good work. See also: http://josm.openstreetmap.de/wiki/Presets/Leaftype --Rudolf (talk) 10:22, 7 June 2014 (UTC) it


Fails for New Zealand

This scheme falls down badly for New Zealand, where the domiant forests are exotic Pinus Radiata (Monterey pine) and some Douglas fir plantations, and much of the native forest is dominated by Hebes and Beeches. Both the hebes and beeches are evergreen, but the main problem is that many hebes are/look like 'needleleaved' too, but not coniferous. So for NZ broad/needle is a useless distinction and only makes things much much worse, and the old way was working quite well. Moreover, trying to conflate the world's flora into two or three broad categories is always bound to fail, so being overly precscriptive about the categories is silly. Suggest to talk to a botanist before proceeding. --Hamish (talk) 20:41, 23 September 2014 (UTC)

Retrieved from "https://wiki.openstreetmap.org/w/index.php?title=Proposal_talk:Leaftype&oldid=2515304"