GPS Stakeout  ·  Docs
Using the App

Accuracy & Anchoring

Everything the app shows you on the ground is measured from your anchor. Get the anchor right and navigation lands on the stake; get it wrong and every corner inherits the error. This page covers how to anchor well, how to read the confidence meter, and why you shouldn't re-anchor every time you arrive somewhere.

The one rule. The quality of your anchor sets a floor on every arrival. A boundary placed from a ±15 ft anchor can never land closer than about 15 ft, no matter how good your GPS is when you later walk to a corner.

Occupying a corner (averaged anchoring)

When you tap Set as Reference (I'm Here), the app doesn't take a single GPS reading. It occupies the point: it collects fixes for a set period, throws out the outliers, and anchors on the average. This is what land surveyors do, and for good reason — a single instantaneous fix carries the receiver's full scatter, which under tree canopy can be 10–15 ft.

While it counts down you'll see the number of fixes collected, the scatter (how much they disagree with each other), and which receiver is supplying them. Stand still and hold the receiver over the corner.

Occupation screen with countdown ring, fixes collected, scatter and source

Set the duration in Settings → Anchor averaging: Instant, 15 s, 30 s, or 60 s. The default is 30 s. Longer occupation averages away more of the random error — use 60 s for the anchors you plan to solve a rotation from.

Honest limitation. Averaging removes the random part of GPS error well. Multipath — signals bouncing off ridges, buildings or wet canopy — is partly correlated over time, so it doesn't average away completely. Expect real improvement, not perfection. Standing longer helps, because satellite geometry shifts as you wait.

The confidence meter

Confidence tells you how much to trust the current position. On the Navigate screen it's a colored ring around the compass (green / amber / red) with an engraved “±ft” reading; it also appears as a four-bar meter in the GPS Source sheet. It's driven mainly by how far your fix wanders while you stand still — which is the honest measure, because a receiver's own accuracy claim can't detect multipath. A unit can report ±5 ft while sitting 14 ft from the truth.

LevelWhat to do
HighSteady, and both receivers agree. Good moment to anchor.
GoodFine for navigating; acceptable for anchoring.
FairNavigate, but wait before anchoring a corner.
LowDon't anchor. Let the position settle first.

When confidence drops to Fair or Low, a short line appears explaining why — "Position steady to ±3 ft", "Receivers disagree by 14 ft", or "Moving — stand still to measure". While you're walking the meter can't measure wander at all (your own travel would look like scatter), so it falls back to reported accuracy and won't claim High.

Confidence measures steadiness, not correctness. A reflected signal can hold a wrong position very steadily. That's why High requires a second receiver to corroborate — if only one receiver is live, the meter caps at Good. A good reading is necessary but not sufficient.
GPS Source sheet showing the active receiver, a four-bar confidence meter reading Good, the 20-second wander figure, and fix latency

Open the full breakdown from the GPS Source sheet: it shows the active receiver, the confidence level and what's driving it, the 20-second wander figure, and fix latency — the age of the last fix. If latency sits at 2–4 seconds while you walk, the position is lagging, which reads as being short of the pin on arrival and then counting down once you stop.

Anchor quality

Once a corner is anchored, the Corners tab shows how good that anchor was — "Anchored at ±3 ft". If it exceeds about 10 ft it turns orange and suggests re-anchoring. That number is the floor on everything downstream, so it's worth glancing at before you spend an afternoon walking a parcel.

Don't re-anchor every time you arrive

This is the single most common way to get inconsistent results, and it's counter-intuitive.

Tapping I'm at [corner] on a corner that's already placed slides the entire boundary so that corner lands on your current fix. That's exactly what you want when you're deliberately re-placing the plat from a known monument. It is not what you want after walking to a corner and finding the distance reads 20 ft short — because in that case you're re-anchoring on a fix you already have reason to distrust, and dragging the whole parcel with it. The next leg then starts from a displaced boundary, and results swing back and forth.

The app now warns you: if anchoring would move the boundary further than your current fix quality justifies, you'll get a "Move the boundary?" confirmation showing how far it will shift and how good the fix is. The safe choice — Keep current placement — is the default.

Rule of thumb. Anchor once, deliberately, at the corner you're most confident of. After that, treat arrival distances as information, not as corrections to apply.

Anchoring vs. rotating — how the workflow actually works

This trips people up, so it's worth stating plainly: tapping “I'm Here” places the plat — it never rotates it.

Rotation is a separate, manual step. It only happens when you open ⋯ → Bearing Basis and tap Solve rotation. Nothing else turns the parcel — we made it opt-in because auto-solving from a noisy reading was throwing navigation off by 15–25 ft.

Most parcels never need it. If the corners land on the stakes from a single anchor, leave rotation alone — a correct deed shouldn't be rotated.

If the parcel plots crooked (older magnetic or “assumed” bearings), the sequence is:

  1. Stand on a known corner → I'm Here (places the plat).
  2. Walk to your farthest known corner → I'm Here (records a 2nd anchor; the plat slides but does not rotate yet).
  3. Open ⋯ → Bearing Basis → Solve rotation from 2 corners — this is the step that rotates the deed to fit both corners.
  4. Confirm it shows a green check, then tap Apply.

When to “Clear GPS Positions”

Clear GPS Positions (the ⋯ menu, or the top of the Plat sheet) is a full reset: it un-places the plat and erases every anchor and any solved rotation. You do not need it for ordinary re-anchoring — “I'm Here” already re-places things. Reach for it when:

PLSS & KML plats. On an imported plat this button reads Reset to BLM / KML Position instead — it restores the authoritative surveyed coordinates rather than wiping the placement, because those plats always have an official position to fall back to.

Solving rotation: baseline matters more than anything

If your deed plots rotated, the two-point solve fits the parcel to the corners you've physically anchored (see Understanding Survey Systems). Its accuracy depends almost entirely on how far apart your anchors are.

The angular error is roughly (position error ÷ baseline). The same 14 ft of GPS scatter produces:

Baseline between anchorsRotation uncertainty
142 ft±5.6° — unusable
300 ft±2.7°
500 ft±1.6°

So: anchor the two most distant corners you can reach, occupy each for 60 s, and solve once. Anchoring more corners helps further — a fit over five corners has far more redundancy than one over two. The app warns you when your anchors are too close together for the result to mean anything.

Anchor everything with the same receiver. Corner positions live in the coordinate frame of whatever receiver measured them. Mixing an iPhone anchor with an external-GPS anchor folds the difference between the two receivers straight into the rotation. The app flags this if it detects it.

Reading the solved rotation

The solved rotation is shown in the same convention as magnetic declination: + is East, − is West. So −2.34° W means the deed's bearings sit 2.34° west of true north.

That's useful because you can compare it against historical declination for your area. If a deed used magnetic bearings, the rotation should match the declination at the time of the original survey — which may be many decades ago. A rotation that matches no plausible declination usually means an assumed meridian: an arbitrary basis tied to a fence, a road, or a prior plat. Both are common; only the document tells you which.

A small rotation proves little. A degree or so is well within what an arbitrary assumed meridian could be by chance, so don't read a survey date out of it. Check the plat for a basis-of-bearing statement — that's the only thing that settles the question.