The general consensus in the previous changeset is that an area like the east part of Hilltop Hanover can not be leisure=park. That is why I changed it to leisure=nature_reserve. Following the "front of park"/"back of park" scheme, the outer border here would be boundary=protected_area perhaps with leisure=nature_reserve, while the farm section would be tagged separately.

Again, I would like to see a tagging scheme that works better for these borderline cases, rather than calling the whole thing either a leisure=park or a leisure=nature_reserve. But currently, it seems like this is what the consensus is.

As for Hilltop Hanover: the current boundary seems like it would definitely fall under leisure=nature_reserve, since it is natural. If the boundary were to be changed to extend across the street and include the farm section, it would admittedly feel strange to call the whole thing a nature reserve. I would be interested in having the community improve the tagging scheme here.

As for Onatru Park and Preserve: this seems to fit the "front of park"/"back of park" scheme that was discussed in the Cranberry Lake changeset. If we were to follow that, then the area with the sports fields would get leisure=park, but the outer boundary would still be boundary=protected_area perhaps with leisure=nature_reserve. Again, using leisure=nature_reserve on the outside here does feel a little weird to me, but that seems to be the usual tagging at the moment.

I've changed the leisure value on the outer boundaries, though we could use some natural=wood polygons. I haven't touched the places where a leisure=park subarea is necessary, yet.

Sure, I suppose I could have said spaces left in a natural state. In the region here though, that is usually wooded.

Well, it seems like the general consensus is that wooded preserves should get leisure=nature_reserve instead of leisure=park, so I will go ahead and make the change. I won't touch borderline cases for now though.

Hi,

Could you confirm where you got the shape for the park from? It looks like it came from the county tax parcels, which is an acceptable source, but I just want to confirm before adding it to the source tag.

I support the hybrid tagging approach, at least in cases where there really is a "front of park" section. Here is an example of a nearby park where I used this scheme: way/1302389330

In that case there is definitely a leisure=park there, though the location in this changeset is more borderline. Some of the other locations in questions have no amenities at all except for perhaps a small dirt parking lot, so in those cases I would say there shouldn't be a leisure=park subarea.

Yes, the tags of the features inside of a park/preserve are important. But think about it from the perspective of someone from out of town who is searching the map for a place to go. If they want to look for a good place to hike and see nature, they wouldn't want a more managed park to come up in their search. Similarly, if they were looking for a park for more general recreation, they probably wouldn't want a nature preserve that is meant for hiking to come up.

I very much appreciate the work that you have been doing for the trails in the region. But working on tagging is something that can and should happen concurrently with collection of new data, especially when there is a dispute.

Since the previous geometry which drew from aerial imagery better follows the median of the two traces, I would like to restore it. What do you think?

I would agree that Rockefeller State Park Preserve is more of a borderline case, as it is more managed and has more amenities when compared to some of the other locations in question. I would not change it from leisure=park without getting consensus first.

However, I think that consistency in this case is undesirable, since we want to be able to tell the difference between a leisure=park (is landscaped, has amenities, and is a good gathering place for miscellaneous recreation), and a place that is more suited for hiking or nature observation.

Some examples of what I would consider a leisure=park:
- Croton Gorge Park
- Glen Island Park
- Kensico Dam Plaza
- Central Park
- Flushing Meadows–Corona Park

Some things I would consider a leisure=nature_reserve or otherwise not a leisure=park (all of these have been changed from nature_reserve to park):
- Crawbuckie Nature Preserve
- Hardscrabble Wilderness Area
- Brinton Brook Sanctuary
- Cranberry Lake Preserve

As for tags that would be consistent between all Westchester County parks, I would support `operator=Westchester County`, and possibly `boundary=protected_area`. I would be open to other suggestions here, but I do think that the leisure value should be allowed to vary so as to allow data consumers to treat the locations accordingly.

Either way, as jmapb said, these sorts of places had already been tagged as leisure=nature_reserve by a variety of mappers prior to being changed over the last few months. If we want to leave things as-is until we reach consensus on which tags to use, then I think the tags should be set to what they had been prior to the retagging here.

Yes, your track certainly has more points and better represents the layout of the trail. I was mainly trying to show that our two devices got noticeably different readings, for example on the northernmost part of the Hudson Loop. The median between all passes, yours and mine, is pretty close to the path on imagery, so I do think the imagery is accurate here.

The southern part of my track looks weird because I took a few paces onto the center east-west path from each side of it, just so that it would be captured on the trace so I'd know to add it. Yours is more accurate particularly in this area, but this is a good example of how GPS can have drift, since the path that is shown on my trace isn't the exact path that I took. It averaged out some of my movement.

I noticed that you added that tree, so when I made my modifications to restore some of the previous geometry, I kept that section since you clearly had knowledge that I missed.

@ZLima12/traces/12083672

It was collected 6 months ago. It would be best to look at the trace in your editor so you could see it alongside your trace and the existing data.

I have uploaded my GPS trace. The app I was using didn't record as many points as yours, but with what is there, you can see that there is some disagreement between the two traces. The geometry I had in previously lines up better in the median between both traces than what you updated it to.

If you would like to see what I'm looking at, I would be happy to provide screenshots showing the paths in 2021 vs 2024, and how the traces don't exactly line up with them.

Again, I did not make aerial imagery the sole factor. If there is a significant deviation between the GPS track and the imagery, more than what is typical for normal GPS inaccuracy, then I will investigate and not blindly follow the imagery.

Even within the same GPS track that you made here, there is as much as a 20ft difference between the two passes you made over some trails. All GPS receivers are affected by this kind of inaccuracy, it is just something that is inherent to GPS.

Yes, the southern part of this park is less maintained and trail visibility is not as good. But for the rest of it, there is inarguably a skew between the actual location of the path and your GPS track. That is expected, as no GPS data is perfect. It is one thing to use GPS tracks without much alteration when that is the only data source available, but I am looking directly at the trails which haven't changed between 2021 and 2024, and the GPS track weaves in and out of alignment with them.

I appreciate that, but still, I ask that you not overfit trails to GPS for the reasoning I provided above. GPS helps a lot, but aerial imagery is going to give better resolution. Looking around nearby GPS traces on roads, you can find examples of drift and distortion, and that is only going to be more prevalent with the conditions in hiking areas. If we are confident that the trail alignment has changed since imagery was taken, then I would disregard that imagery, but I don't think that's the case here.

I am not relying on the imagery too heavily. I did not map these trails until I did a GPS trace several months ago, so that I could be more confident in what I was seeing on imagery.

I compared the imagery from 2021 and 2024, and where the trails are clearly visible on both, there appears to be essentially no change to their alignment. If you look at a road that has many traces, it appears that the NYS imagery is well aligned, with the median track following the road well. Therefore, it is likely that the GPS trace here is not entirely accurate, especially since the elevation changes and tree cover in this preserve will cause distortions in the readings.

In this case, I believe the previous detail in almost all places was sufficient. There were a couple places where I benefited from your GPS track in addition to mine, and I incorporated those changes. But please, overfitting to GPS when there was already a good level of detail does not improve the data quality.

Also, I did not change the geometry of the path by Avalon. I only split the way and modified the part that was in the woods. The part that went on the grass, that you said went through the parking lot, was added by you and not touched by anyone else.

I do appreciate the work that you are doing, as you have done quite a few good GPS traces and added trails that weren't there at all before. But I also put quite a bit of effort into this park to make sure things were accurate, and I don't want that to go to waste.

Actually I see that you didn't revert, you just removed the attribution I added. I added it back since the current geometry does indeed come from both the aerial imagery and the GPS tracks.

I see now that you have reverted these changes in changeset/172939770 . Please take a look at what I've said here to understand why I made those changes.

Hi,

I previously added this preserve following a survey. I did a GPS recording (though I didn't upload it), but then supplemented this by using the NYS Orthos color IR imagery which gave a fairly good view of the trails here. I should have better indicated that I used imagery to get the path geometry here, but I would just ask that you try to make sure that the previous geometry is inaccurate before fitting paths to your GPS recordings. Such recordings are subject to drift and other inaccuracies, so it is best to combine it with other sources when possible.

In this case, I restored much of the old geometry, since your changeset moved paths off of alignment. However, I incorporated changes using your GPS track, which did show that I had the path wrong in a couple places. I will resurvey sometime soon to double check things.

The imagery I used here is actually the 2021 version of the color IR imagery, since for whatever reason the newer versions don't give as good of a view of the paths. If you want to see this in iD, you can paste this link into the box that comes up when you click "Custom" under the background options: https://orthos.its.ny.gov/arcgis/services/wms/Latest_cir/MapServer/WmsServer?FORMAT=image/jpeg&VERSION=1.1.1&SERVICE=WMS&REQUEST=GetMap&LAYERS=0&STYLES=&SRS={https://wiki.openstreetmap.org/wiki/Tag:proj}&WIDTH={width}&HEIGHT={height}&BBOX={bbox}

Please also take a look at this, which I think explains the situation well: osm.wiki/United_States/Public_lands

"In particular, leisure=park is widely misused on areas more properly tagged with boundary=protected_area or leisure=nature_reserve. OSM's leisure=park tag (a smaller, manicured, urban park) conflicts with "park" as used in US English to mean "a large area of land kept in its natural state for public recreational use.""