Wire vs Satellite: The Physics of a Dog Fence That Never Drifts

A GPS satellite's signal reaches your yard carrying less than one billionth of a watt. A federal cybersecurity presentation to the government's own GPS advisory board compares receiving it to trying to see a 20-watt light bulb in California from New York, at noon. Every GPS dog fence stakes your dog's boundary on that whisper. A buried wire doesn't, and the reasons live in physics you can check.

We covered the market data in GPS vs in-ground: the makers' own drift specs, the buffers, the batteries, the three-year cost. This post is the layer underneath, the science of why those numbers come out the way they do. Every physics figure here traces to a government spec, a peer-reviewed study, or an FCC filing, and where a number comes from a manufacturer's manual, we say so.

The wire makes a field. The collar just listens.#

Start with what a wired fence actually is, because the phrase "underground fence" undersells how physical the thing really is. A transmitter on your garage wall drives a signal through a loop of boundary wire in the ground. The manuals describe it plainly: the signal is carried by the wire, and the collar is a small radio receiver that responds when it gets close enough to hear it.

Here is the detail that explains everything else. FCC filings for one major maker's transmitters certify them at 10.65 kilohertz; these systems live in the low-kilohertz band, the very bottom of the radio spectrum (the FCC's legal definition of radio frequency energy starts at 9 kilohertz). The GPS signal your phone uses sits at 1,575.42 megahertz, roughly 150,000 times higher. At 10.65 kilohertz a radio wave is about 17 miles long, so a yard-sized loop is far too small to launch one. Instead of radiating away, the loop's energy stays wrapped around the wire as a local field, which is exactly what you want from a boundary: it soaks through soil and slush and stays home. GPS's microwave-range signal is the opposite animal. It travels in near-straight lines and is attenuated, reflected, or blocked outright by buildings, terrain, and foliage.

The practical consequence: the fence's field is generated in your yard, a few feet from your dog's collar. You set its width with the transmitter dial and a walk test (12-24 feet of total detection width, with the field splitting about 20 percent warning tone and 80 percent correction zone, per the manuals). In fairness to precision: that width can flex a little at corners, with how the collar sits on the neck, and near large metal objects, which is exactly why the walk test exists. What the field does not do is drift with the day. It is fixed in time, and after setup there is nothing to compute, nothing to update, and no sky to consult. The boundary is a physical object, a loop of copper (our store) with a transmitter (our store) feeding it.

dogfenceguide.com

What a GPS collar must do instead#

A GPS collar has no boundary to listen for. It must first figure out where it is, and the official GPS performance standard spells out what that takes: pseudorange measurements from at least four satellites, orbiting about 12,550 miles up. Why four? Three distances fix a position; the fourth is needed because the collar's inexpensive clock is itself an unknown the math has to solve. The collar measures how long each signal took to arrive, converts time to distance, and finds where the range spheres cross. It's an elegant system, and every error in it lands on your dog.

The government's own standard publishes the error budget, and the numbers deserve to be quoted exactly. For a single-frequency civilian receiver, the 2020 GPS Standard Positioning Service Performance Standard puts the illustrative 95 percent user equivalent range error at 11.9-20.7 meters, which is 39-68 feet of range uncertainty before satellite geometry multiplies it into a position error. The largest single line item is the ionosphere at 9.8-19.6 meters, and that is what remains after the correction model the satellites broadcast has already done its work; the European Space Agency's technical documentation pegs that model at removing only about half of the raw ionospheric error.

dogfenceguide.com

To be fair to modern hardware: premium collars listen to multiple satellite constellations, some now receive on two frequencies (which cancels most of the ionosphere term directly), and they fuse in motion sensors. Much of that improvement is real. What no second frequency fixes are the reflections off your own house, the canopy over the side yard, and the geometry multiplier, and the makers' own setup rules (the 15-25 foot road buffers we documented in the comparison post) tell you where their field experience says the practical floor sits.

The sky is part of the system#

The ionosphere deserves its own paragraph, because a GPS boundary owner lives with it forever. It is the layer of charged particles the signal must cross, and the peer-reviewed navigation literature identifies it as the largest positioning error source for single-frequency users. It is also never the same two hours running. Penn State's GPS course, a standard reference for surveyors, notes that midlatitude ionospheric delay during the day can run five times the nighttime value, that November's delay runs nearly four times July's, and that it all peaks with the solar cycle. NOAA's Space Weather Prediction Center puts a number on neglecting it: tens of meters of position error.

The boundary your dog learned in July is being computed through a different atmosphere in November, at breakfast versus midnight, and at solar maximum versus solar minimum. None of this touches a buried wire. We watched it play out for real on May 10, 2024, when a G5 solar storm sent centimeter-grade farm tractors wandering in circles; the comparison post tells that story with NOAA's numbers.

How weak is weak?#

The GPS signal's real enemy is arithmetic. Each satellite's civilian signal leaves the antenna at an effective radiated power of 479 watts. It then spreads over the 12,550-mile trip, losing about 184 decibels to distance alone, and the official standard guarantees only that it arrives no weaker than -158.5 dBW. Written out, that floor is 0.00000000000000014 watts. Penn State's course draws the engineering conclusion in one line: because the received signal is so weak, it is easily degraded by vegetation canopy, urban canyons, and other interference.

Peer-reviewed field work shows what "easily degraded" means for the signal itself:

• Measurements across 15 US forests found tree canopy costs GPS signals 2.8-10.8 decibels, and a Forest Service crew under heavy hardwood canopy found satellites below 35 degrees elevation simply unusable.
• Snow research (on receivers buried for avalanche and snowpack studies, not on a dog's neck) found GPS loses 11-13 decibels in the first 1.5 meters of snow, and that a meltwater layer of about an inch and a half of slush above the antenna can block reception entirely. Your dog's collar rides above the snow; the point is how thin the margin is on a signal this faint.

Now the contrast. When mine safety engineers needed a radio that works through solid rock, the NIOSH-sponsored solution was through-the-earth radio operating at 300-5,000 hertz, because only low frequencies penetrate ground; tested units pushed voice through roughly 1,000 feet of overburden. That is a far harder job than any lawn will ever pose, and your fence speaks a similar low-frequency language. Four to six inches of buried depth, half a foot of snow, a rainstorm: the field does not notice. The physics that forces miners to the bottom of the spectrum is the physics your fence was built on all along.

dogfenceguide.com

Your dog's brain wants a fixed landmark#

The physics argument has a biology twin, and it's the one that decides whether the fence actually works. Dogs navigate by landmarks. In a controlled study published in Learning and Memory (Milgram and colleagues, 1999), dogs learned to find a reward by its spatial relationship to a landmark, and every one of the nine dogs succeeded while that relationship stayed fixed. As the researchers loosened the landmark-to-target relationship, accuracy fell apart. Fixed spatial relationships are what dog brains are built to learn.

A boundary fence is exactly this task. The flags, the tone line, the correction zone: your dog is learning a landmark map of the yard. A buried wire gives that map a landmark that never moves, which is why the 14-day training protocol converges: same flags, same tone, same line, every session. A GPS boundary asks the dog to learn a landmark that, by the makers' own mid-2026 setup specs, wanders 10-16 feet with the sky.

The welfare research says the same thing from the other direction. In the Schalke study of 14 Beagles, the only group that showed no stress-hormone rise was the group whose corrections were predictably tied to their own behavior; dogs corrected unpredictably showed elevated cortisol. Veterinary behaviorists (DACVB) write that predictability contributes to a dog's emotional comfort and is foundational to learning. A line that holds still is what makes the training humane and the learning permanent. The correction itself, on any system, should stay what it is designed to be: a brief, low-energy static interrupt, delivered through a collar worn no more than 12 hours a day and repositioned every 1-2 hours (collar fit and skin care covers the wear rules).

What the physics cannot fix#

Building on the physics#

If the science pushes you toward wire, it also tells you what to buy. The field is only as reliable as the loop, so the money goes into the physical layer: solid boundary wire at a gauge that survives your soil, waterproof splices, and surge protection.

Not sure what your perimeter is? Draw it in the free fence planner on satellite imagery and you'll get footage, layout, and cost in a few minutes, or use the wire calculator if you already know the number. If trenching isn't your weekend, the planner can also route your yard to a professional installer for a quote.

The bottom line#

One system's boundary is a computation performed on a billionth-of-a-watt whisper that crossed 12,550 miles of space weather, clouds, and tree canopy before your dog's collar computed a fix from it. The other's boundary is a piece of copper in your dirt, carrying a field too low-frequency to care what the weather is doing, at a spot that has not moved since the day you buried it. Both are real technology, and there are yards where GPS is the right call. But if the question is which boundary your dog can trust on every day of every season, the physics only points one way.

---

Want the market-data half of this argument, prices, batteries, and the makers' own drift specs? Read GPS vs in-ground. Ready to see wire on your own lot? Start with the planner.