How Accurate Is Phone GPS Under Tree Canopy?

Every phone-mapping method for orchards depends on one number almost nobody checks: how far off is the location the phone reports, standing under your trees? The honest answer, from published forest measurements, runs from about 2 metres where the canopy is open to more than 10 metres under full leaf, depending on the phone, the season, and the sky view. That range straddles the spacing of a planted orchard, so it is worth understanding properly. This article summarises what the research measured and translates it into decisions for a grower mapping trees at 5 to 10 metre spacing. One caution up front: the studies were run in forests, not orchards, so treat their numbers as a guide and measure your own block (the method is below).

What the studies measure

Researchers have tested consumer phones against surveyed reference points under forest canopy for years, and the numbers are consistent in shape if not in magnitude.

• A 2022 study in the journal Forests, led from the University of British Columbia, tested one single-frequency and two dual-frequency smartphones at 57 points in a mostly broad-leaved forest during the deciduous season, against survey-grade reference positions. With correction processing (PPP), the phones reached about 2.5 metres of horizontal error where canopy openness was above 70%. Below 70% openness, the authors say survey-grade equipment is still needed for sub-metre work.
• The same paper cites earlier forest trials: three smartphones that measured 6.7 to 11.5 metres of error with leaves on the trees and 4.5 to 6.7 metres without, and dual-frequency phones that measured about 6.1 metres with leaves and 4.1 metres without.
• A 2026 University of Georgia study in Forest Science tested an iPhone 13 Pro and a Pixel 6 Pro on a forested test course, the way a field professional uses a phone: a single quick fix, held at about waist height, with minimal waiting. The authors conclude that foresters may expect 2 to 6 metres of accuracy from a smartphone in some southern US forests. The Pixel outperformed the iPhone by 1 to 3 metres in both stands tested, a pine stand and a hardwood stand, leaf-on and leaf-off.
• A 2023 PLOS One study from the same research group asked whether the size and position of nearby trees produce a predictable error. They do not: forest variables correlated with position error for every receiver tested, but the trends were not consistent, so there is no correction you can apply on paper. The paper notes that multipath, signals bouncing off trunks and canopy before reaching the antenna, may be the largest source of error for any receiver in a forest, and that it is almost impossible to mitigate while moving among trees.

Summary of the figures above: 2 to 6 metres is what a current phone managed under forest canopy, older trials under full leaf measured 6 to 11 metres, and in every study more open sky meant a better fix.

What that means at orchard spacing

Here is the translation a forestry study will not do for you. Mango and avocado trees are commonly planted 6 to 10 metres apart; high-density citrus runs 4 to 5; intensively managed blocks can be closer still. If your phone's fix is 3 metres off in a block at 8 metre spacing, the marker still lands nearer the right tree than any other, and a human looking at the map and the ground resolves it instantly. If your fix is 8 metres off under a dense durian canopy, the marker could plausibly belong to either of two or three neighbours, and blind coordinate-matching starts assigning photos to the wrong tree.

A worked example makes the boundary concrete. The rule is simple: when the error is less than half the spacing, the nearest tree to the fix is always the right one. Suppose cocoa planted at 3 by 3 metres (a common commercial density), and your phone reports a photo location with 4 metres of error. A circle of 4 metres around the fix covers about 50 square metres, and at 9 square metres per tree that is five or six trunks. The coordinates alone cannot decide which tree the photo belongs to. Now take the same 4 metre error in a mango block at 9 by 9 metres: 4 is less than half of 9, so the fix is still closer to the right tree than to any neighbour. This is why low, tight crops need deliberate trunk-framing photos, and wide-spacing orchards can lean on coordinates much more.

Does your phone matter

Partly. The British Columbia study found its dual-frequency phones (L1 plus L5 signals) more accurate under forest than its single-frequency phone, and the Georgia authors cite the same result. Dual-frequency chipsets have shipped in Android phones since 2018, and Apple added one with the iPhone 14 Pro. Android also exposes raw GNSS measurements to apps, which is what makes post-processing and advanced correction possible; iPhones do not. If you are buying a phone that will double as your survey instrument, dual-frequency is the spec worth paying for. If you are not buying anything, the field habits below cost nothing.

You can also measure your own error for free, which beats trusting any published number. Pick a feature with a known position on a good satellite basemap (a gate, a pump, a lone post), stand there under representative canopy, and drop a pin with the phone. Repeat five times across a day. The spread of those pins is your real-world error at that spot, and it directly tells you whether your spacing tolerates coordinate-only mapping or needs the photo-first habits described below.

Field habits that keep fixes honest

1. 1Turn location tagging on before the survey and verify one test photo actually carries coordinates.
2. 2Stand beside the trunk when you photograph. Walking to the tree costs seconds and pins the photo to the right neighbourhood.
3. 3Let the phone settle for a few seconds before the first photo at each tree. The Georgia figures come from quick fixes with minimal waiting, so patience can only help.
4. 4Prefer fixes under gaps in the canopy. The British Columbia study found phone accuracy followed canopy openness, so when a step to the side gives you more sky, take it.
5. 5Photograph each trunk deliberately, especially for cauliflorous crops (durian, jackfruit, cacao) and any tree standing close to its neighbours. The photo content then disambiguates what coordinates cannot.
6. 6Take two or three frames per tree from different angles. Redundancy is cheap; a wrong assignment is expensive to find later.
7. 7Season matters. The forest trials above measured smaller errors without leaves than with them, so in a deciduous orchard a winter survey should give better coordinates than one under full summer leaf.

When phone GPS is not enough

Be clear about the jobs phone GPS cannot do. It will not settle a legal boundary (that is a surveyor). It will not give repeatable sub-metre positions under dense canopy on its own. The 2026 Georgia study reports that a high-precision external antenna paired with the phone may reach sub-metre accuracy in some forests, but at a cost comparable to a mapping-grade receiver, which most small growers do not need on day one. What phone GPS does support, at orchard scale, is a tree inventory whose positions are roughly right and whose identity comes from the combination of location, photos, and tags. That is enough for the decisions an inventory feeds: health scouting, fruit counts, replanting choices, and yield history.

This is how CanopyTwin treats location. Survey photos are geotagged, GPS clusters them into individually tracked trees on a live satellite map, and each tree keeps its own page with its photo history, so the photos back up what the coordinates say. The method is described in our orchard survey guide, it works on the web and on iOS, and every plan starts with a 30-day free trial. The demo orchard at canopytwin.com/map shows the end result before you survey your own rows.