Skip to content

LAND_OWNERSHIP.md

Latest commit

 

History

History
114 lines (70 loc) · 9.38 KB

LAND_OWNERSHIP.md

File metadata and controls

114 lines (70 loc) · 9.38 KB
Raw

OurHike — Surrounding Land Ownership (Feature Design Draft v1)

Companion to MAP_OPTIONS.md, TRAIL_BLAZE_COLORS.md, ../OurHikeValues.md and ../WIREFRAMES.md.

In much of its length the AT is a narrow protected sliver with private property on both sides. A hiker stepping fifty feet off-trail to camp, find water, or take a shortcut may be leaving public land without any sign that they have. This feature shows what kind of land surrounds the corridor — national park, national forest, state forest, local park, AT-owned — so that leaving it is a visible act rather than an accident.

Research findings, before any design

ATC's own GIS does not have this

Checked the live NPS/ATC ArcGIS org (services1.arcgis.com/fBc8EJBxQRMcHlei) directly on 2026-07-30, the same way bridges/privies/at_treadway were originally found by listing the FeatureServer root rather than trusting the public map's layer list.

Seven AT-specific services exist:

APPA_ECBS_Monitor_Sections            APPA_NatResRriorityZones
APPA_HeleneStatusCenterline           APPA_Trail_Club_Sections   (registered)
APPA_HUC10_LargeLandscape_dissolve    SHEN_TRANS_AppalachianTrail
Appalachian_National_Scenic_Trail_500ft_Buffer

None carries land ownership. The closest candidate — Appalachian_National_Scenic_Trail_500ft_Buffer — turns out to be pure geometry: its only fields are FID, Shape_Area, Shape__Area, Shape__Length. It is a 500 ft geometric buffer drawn around the centerline, not a record of what is actually owned or protected. Useful later as a crude "nominal corridor" hint, useless for ownership.

So this is the first feature in a while that genuinely cannot be sourced from data we already hold.

PAD-US is the right source, and it has exactly the right fields

The USGS Protected Areas Database of the United States is the canonical national dataset for this. Verified live on services.arcgis.com/v01gqwM5QqNysAAi, which publishes several PAD-US views (Manager_Type_PADUS, Fee_Managers_PADUS, Manager_Name_PADUS, and others).

The attributes map almost one-to-one onto what was asked for:

field coded values (excerpt)
Mang_Type Federal · State · Local Government · Regional Agency · Private · Non-Governmental Organization · Joint · Unknown · American Indian Lands
Des_Tp National Park · National Forest · Wilderness Area · National Wildlife Refuge · National Scenic or Historic Trail · Conservation Area
Own_Name / Mang_Name National Park Service · Forest Service · TVA · Army Corps of Engineers · …
Unit_Nm the actual unit's name, for display
GAP_Sts protection status 1–4 (biodiversity mandate vs multiple-use)

Des_Tp = "National Scenic or Historic Trail" combined with Own_Name = "National Park Service" is how AT-owned corridor land identifies itself — so "AT owned land" is a query against PAD-US, not a separate source to find.

The thing this feature must not do

Absence from PAD-US does not mean private. PAD-US maps protected areas. Land that is unmapped is unmapped — it may be private, or it may be public land that a steward has not yet submitted, or a gap in a state's reporting.

This matters more than any rendering decision in this document, because the obvious implementation is the wrong one. Painting "not in PAD-US" as private and telling a hiker "do not go here, this is private land" would be:

  • Wrong sometimes, in a way the hiker cannot check and would have no reason to doubt.
  • Consequential when wrong in both directions — either implying a trespass that is not one, or (worse) implying by omission that unshaded ground is fine to walk on.
  • Against ../OurHikeValues.md #4, trustworthy above all: a confident claim we cannot actually support.

So the feature shows what land IS, and lets absence be absence. Protected land is drawn and named. Everything else is simply not drawn, and the legend says plainly that unshaded ground is unknown, not private. A hiker who sees the shading end knows they are leaving mapped protected land — which is the actionable fact, and is true.

The same restraint appears elsewhere in this project already: MAP_OPTIONS.md refuses to compute detours, and SPUR_TRAILS.md says nothing rather than "Unknown destination". This is that principle applied where the stakes are legal rather than logistical.

A second, related honesty limit: PAD-US is a compiled national dataset, not a legal boundary survey. Its polygons are accurate enough to say "you are in Nantahala National Forest" and not accurate enough to say "you crossed a property line eight metres ago." The UI must never imply parcel-level precision — no "you are now on private land" alerts, no boundary-crossing notification. Which is also consistent with OurHike's one-notification policy.

Design

1. Pipeline: a new corridor-clipped source

Follows the existing pattern exactly — a sources.json entry, a fetch, corridor clipping, an export:

pipeline/fetch_land_ownership.py     PAD-US, queried per corridor cell
pipeline/export_land_ownership.py    clip + simplify + emit

Clipped to the same 30-mile corridor every other layer uses. PAD-US nationwide is very large; the AT's 14 states are a small fraction of it and the corridor is a small fraction of those.

Simplified aggressively. These are background context polygons, not safety-critical geometry — nothing about them needs the 1 m tolerance export_trails.py uses for the trail line. A much coarser tolerance is appropriate and should be measured the same way, since polygon vertex counts dominate the size of a dataset like this.

Fields kept: Mang_Type, Des_Tp, Own_Name, Unit_Nm. Everything else dropped — GAP_Sts is interesting to a conservationist and noise to a hiker.

Freshness: PAD-US publishes on a versioned cycle (PAD-US 3.0, 4.0…), so check_freshness.py gains a source whose marker is the published version rather than a per-file timestamp — closest in shape to the elevation edition-set marker already there.

2. Rendering: quiet, and beneath everything that matters

A translucent fill per Mang_Type, under every existing layer. Ordering is not cosmetic here — this sits below the topo background's features, the trail lines, closures and warning pins, all of which are either safety-relevant or the reason the app exists. A land-ownership wash that competes with a closure band would be a real regression.

Colour should not reuse the blaze palette or the closure red. TRAIL_BLAZE_COLORS.md already spends the saturated end of the palette on line colours that carry meaning, and closureStyle.ts deliberately made closures structurally distinct from a red blaze — a new fill layer must not undo that by introducing a large red region.

Off by default is worth considering and is left open below: this is context, not navigation, and a permanently-shaded map costs legibility for something most hikers want occasionally.

3. What the hiker sees

Tapping an area names it plainly, using PAD-US's own words rather than our interpretation:

Nantahala National Forest National Forest · managed by the Forest Service

Where shading stops, the legend explains the boundary honestly:

Shaded areas are mapped public or protected land. Unshaded means we don't have ownership data — not that it's private.

That second line is the whole feature's integrity in one sentence, and should survive any copy edit.

What this deliberately isn't

  • Not a trespass warning. No alerts, no boundary-crossing notification, no "private land ahead". The data does not support it and the notification policy forbids it.
  • Not parcel data. No owner names of private individuals, ever — PAD-US does not carry them and it would be a privacy problem if it did.
  • Not a routing constraint. It does not affect snapping, wrong-way detection, or anything that computes.
  • Not a substitute for signage and blazes. On the ground, the blazes and the landowner's own signs are authoritative. This is orientation, not permission.

Open questions (for you, not decided here)

  • On or off by default. Argued weakly for off above; genuinely depends on how heavy the shading reads on a real topo background, which wants seeing rather than deciding.
  • How many Mang_Type categories to actually distinguish visually. Nine coded values is far too many fills to tell apart. Federal / State / Local / Private-protected / NGO is probably four too many already — the honest minimum may be just "protected" vs "not mapped".
  • Whether to include the 500 ft APPA buffer as a separate hint. It is the nominal corridor and cheap to add, but showing a geometric buffer next to real ownership polygons risks reading as though it were also ownership.
  • Size after clipping and simplification. Unmeasured. Polygon layers can be much heavier than they look, and this ships inside a download people already weigh against phone storage — worth measuring before committing to a tier.
  • Whether this is MVP. It needs a new source, a new fetch, a new export and a new client layer, none of which exist. Reads Post-MVP, but it is a scope call.