Talk:Florida Landuse Import
Jump to navigation
Jump to search
" Data license: Public domain" - how we know this? Where it was specified? https://geodata.dep.state.fl.us/datasets/FDEP::current-landuse-landscape-support-index-lsi/about is missing indicator of that Mateusz Konieczny (talk) 21:45, 22 February 2022 (UTC)
God bless America. -------> https://en.wikipedia.org/wiki/Copyright_status_of_works_by_the_government_of_Florida SherbetS (talk) 02:13, 30 September 2022 (UTC)
- God aside I did a basic glance over some of the stuff you've been importing and it looks to be extremely inaccurate in a lot of places, if not completely bad. Especially the areas where you imported a bunch of scrub land in places that are actually trees and/or woods. You really should be doing quality checks on the data before you import it. Instead of just allowing clear garbage data to be added to the map in mass. It doesn't really benefit anyone to import bad information. Especially at the scale you seem to be doing it at. --Adamant1 (talk) 03:54, 19 August 2023 (UTC)
- I think that calling the entirety of this dataset "complete garbage" because of an isolated tagging mistake is a bit of an overreaction. The objects which I assume you are referring to are identified on the DB as 4430:Forest Regeneration Areas, which zluuzki figured was best represented by natural=scrub. I can change this tag if you think it'd be more appropriate as forest or wood. --SherbetS (talk) 18:28, 19 August 2023 (UTC)
- The areas of forests that are being imported as scrubland is garbage. Does that mean the dataset is "complete garbage"? No and I never said it was. There's clearly a difference between "clear garbage" and "complete garbage." I said the former, not the later, and I really don't appreciated being miss cited about it. In no way is it an "isolated tagging mistake" either. At least from what I saw it's an extremely large area taking up large parts of the state. Changing the tag is something. But it would have to be done on a area by area basis and I assume you can't do that because of the scale. Plus you'd still be left with thousands of tiny micro mapped forest areas inside of other ones. Which really isn't any better. Nor does it really mesh with the whole "One feature, one OSM element" thing. Although you already seem to have that problem in places like this where there's 30 or more different little swampland areas in OpenStreetMap to represent what's essentially only a couple of swamps IRL. If they even exist to begin with. There's other places where it's essentially the same issue but with forests and scrubland. So I think it needs to be dealt with on a broader level then just changing a tag. I'm not going to go as far as saying it's "complete garbage" like you claim I did, but it's definitely not very accurate in general and I don't think changing a tag without doing a broader review (that hopefully involves reverting at least some of it) is really going to solve things. --Adamant1 (talk) 06:14, 21 August 2023 (UTC)
- Hi Adamant, I did not misquote you intentionally. I composed a response based on a different line of reasoning, but changed my mind and quoted you again from memory of what I wrote before, instead of what was actually written, and I apologize. I still disagree that the data is clear garbage, considering that the polygon is still accurate to an individual area of landcover, and at one point in time was tagged accurately. I would only consider data such as a completely inaccurate polygon that has never been accurate to be clear garbage. Also, why would the tag have to be changed on an area by area basis? If the tag means the same thing everywhere, why can't we change the tag across the entire database? On the way you cited to violate the one feature one element, I disagree. On the database, the larger wetland is identified as '6210: Cypress' while the smaller wetland is identified as '6170: Mixed Wetland Hardwoods' Although they were both translated as swamp onto OSM, this database made a distinction that is MORE granular than what most OSM users would bother doing, and the only problem is that the tagging wasn't specific enough to separate these two objects. I don't feel that changing the scrub areas to be tagged to forest would create any problematic tagging, because most farmers harvest all the trees at once in a tract, and wherever the scrub ended most likely marked the end of the farmer's land. --SherbetS (talk) 12:58, 21 August 2023 (UTC)
- "I still disagree that the data is clear garbage." Again, some of the data is clear data. Stop acting like I said all of it is. I don't really care about the data is fine. That's not what I have an issue with. I'm not really sure what your point about it being accurate at one time is. The point in OpenStreetMap and one of it's guiding principles it that we map what's on the ground. So if the data was accurate 5 years ago but isn't now then it still shouldn't be in OpenStreetMap. If you want an example you can just look at this round marsh area. Maybe it was a marsh a few years ago, but it clearly isn't now and is partially over a house. The same goes for this heath area that's now a green house and some residents. Etc. Etc. Those are clearly inaccurate polygons and there's hundreds, if not thousands, of example out there. It doesn't what those areas used to be and the edits adding them should be reverted. Period. It's ridiculous to act like those areas should stay on the map just because they might have been accurate 15 years ago when the state of Florida created the information your importing and it really should have been reviewed for quality before you did the import.
- Hi Adamant, I did not misquote you intentionally. I composed a response based on a different line of reasoning, but changed my mind and quoted you again from memory of what I wrote before, instead of what was actually written, and I apologize. I still disagree that the data is clear garbage, considering that the polygon is still accurate to an individual area of landcover, and at one point in time was tagged accurately. I would only consider data such as a completely inaccurate polygon that has never been accurate to be clear garbage. Also, why would the tag have to be changed on an area by area basis? If the tag means the same thing everywhere, why can't we change the tag across the entire database? On the way you cited to violate the one feature one element, I disagree. On the database, the larger wetland is identified as '6210: Cypress' while the smaller wetland is identified as '6170: Mixed Wetland Hardwoods' Although they were both translated as swamp onto OSM, this database made a distinction that is MORE granular than what most OSM users would bother doing, and the only problem is that the tagging wasn't specific enough to separate these two objects. I don't feel that changing the scrub areas to be tagged to forest would create any problematic tagging, because most farmers harvest all the trees at once in a tract, and wherever the scrub ended most likely marked the end of the farmer's land. --SherbetS (talk) 12:58, 21 August 2023 (UTC)
- The areas of forests that are being imported as scrubland is garbage. Does that mean the dataset is "complete garbage"? No and I never said it was. There's clearly a difference between "clear garbage" and "complete garbage." I said the former, not the later, and I really don't appreciated being miss cited about it. In no way is it an "isolated tagging mistake" either. At least from what I saw it's an extremely large area taking up large parts of the state. Changing the tag is something. But it would have to be done on a area by area basis and I assume you can't do that because of the scale. Plus you'd still be left with thousands of tiny micro mapped forest areas inside of other ones. Which really isn't any better. Nor does it really mesh with the whole "One feature, one OSM element" thing. Although you already seem to have that problem in places like this where there's 30 or more different little swampland areas in OpenStreetMap to represent what's essentially only a couple of swamps IRL. If they even exist to begin with. There's other places where it's essentially the same issue but with forests and scrubland. So I think it needs to be dealt with on a broader level then just changing a tag. I'm not going to go as far as saying it's "complete garbage" like you claim I did, but it's definitely not very accurate in general and I don't think changing a tag without doing a broader review (that hopefully involves reverting at least some of it) is really going to solve things. --Adamant1 (talk) 06:14, 21 August 2023 (UTC)
- I think that calling the entirety of this dataset "complete garbage" because of an isolated tagging mistake is a bit of an overreaction. The objects which I assume you are referring to are identified on the DB as 4430:Forest Regeneration Areas, which zluuzki figured was best represented by natural=scrub. I can change this tag if you think it'd be more appropriate as forest or wood. --SherbetS (talk) 18:28, 19 August 2023 (UTC)
- "Also, why would the tag have to be changed on an area by area basis?" Because not every scrub is mapped wrong. You can't just change everything tagged as a scrub into a forest in Florida because 60% or whatever of it is wrong, obviously. Even in the context of the import I'm sure not every area mapped as scrub land is bad. So if you re-tag all of them you'll just create the same issue, but in reverse.
- "This database made a distinction that is MORE granular than what most OSM users would bother doing." That's fine, but it clearly didn't transfer over to OpenStreetMap and it's not suppose to be a 1/1 recreation of state derived anyway. If anything that's exactly not OpenStreetMap isn't suppose to be. Especially at the state level. Landuse also isn't supposed to mapped at the parcel level anyway. So it doesn't really matter how the state has landuse classified on the per farmer level. There's no way to maintain or verify that information anyway even if it did. But I don't think that's what the polygons represent anyway. More then likely they are just random areas based on surveys that were done at the time, not actual parcel data. You really should have known all that before you did the import though. It's pretty clear that a good portion of the imports need to be reverted. I don't think just re-tagging every scrubland in Florida to forests is really going to solve the problem. More so since it's even the only example. You really need to take a serious look at the quality of the imports. Instead of acting like there's nothing wrong with them or that the issues can be solved simply by shuffling a few tags around. --Adamant1 (talk) 05:58, 23 August 2023 (UTC)
- If you think we shouldn't have data that was accurate 5 years ago, by that logic, should we revert all changes by any user or import added more than 5 years ago because they are out of date? The reason we don't is because their presence provides more value than their absence, which is the case here.
- By the way, the data is 5 years old, not 15. The data WAS reviewed for quality before import, and it was determined that the data was of acceptable quality to add, and have to do the minor maintenance over time, rather than having to spend hundreds of data making an incomplete and still subject to time duplicate of the landuse data that was already collected.
- I have the skill to only replace the scrubs that were originally tagged with this database tag, it won't be hard. I can change the tagging to use landuse=forest instead.
- The data isn't mapped at the parcel level.
- I've never thought there were no issues with the import. I just know that what was already on OSM and whatever might've been added by other mappers was of worse quality than what the import added, in some cases by large margins.
- I'd be happy for you to notify me of any other tagging errors present, and I will fix them. --SherbetS (talk) 14:46, 31 August 2023 (UTC)
- "This database made a distinction that is MORE granular than what most OSM users would bother doing." That's fine, but it clearly didn't transfer over to OpenStreetMap and it's not suppose to be a 1/1 recreation of state derived anyway. If anything that's exactly not OpenStreetMap isn't suppose to be. Especially at the state level. Landuse also isn't supposed to mapped at the parcel level anyway. So it doesn't really matter how the state has landuse classified on the per farmer level. There's no way to maintain or verify that information anyway even if it did. But I don't think that's what the polygons represent anyway. More then likely they are just random areas based on surveys that were done at the time, not actual parcel data. You really should have known all that before you did the import though. It's pretty clear that a good portion of the imports need to be reverted. I don't think just re-tagging every scrubland in Florida to forests is really going to solve the problem. More so since it's even the only example. You really need to take a serious look at the quality of the imports. Instead of acting like there's nothing wrong with them or that the issues can be solved simply by shuffling a few tags around. --Adamant1 (talk) 05:58, 23 August 2023 (UTC)