Ground Penetrating Radar Depth: How Deep Does GPR Go in Real Surveys?

In July 2018, a drone flew low and slow over the Greenland ice cap carrying a low-frequency ground penetrating radar. Its cargo was a mission run with the Fallen American MIA Repatriation Foundation to locate Echo, a P-38 Lightning that went down during World War II and had spent over 80 years getting swallowed by the ice sheet. The radargrams came back with anomalies at roughly 100 meters depth.

If you've spent any time researching GPR depth, you've seen claims ranging from 0.5 m to 30+ m. GPR really does vary across two orders of magnitude, and the reason is almost entirely physics. A 100-meter detection through polar ice and a 50-centimeter scan over wet clay can both be accurate, as they might be describing wildly different environments.

The following section outlines the key factors that determine depth, presents the measurements obtained from field observations, and provides a framework for estimating expected conditions at a given site prior to allocating resources such as personnel and UAV systems or extended billable time.

What determines how deep GPR can go

Material conductivity, water content, antenna frequency, and the target form determine the depth of ground penetrating radars. Let's examine the roles these factors play in GPR depth measurement.

• Material conductivity: This is the key factor because radar pulses lose energy when passing through conductive materials. Substances like clay, saline soils, and seawater all absorb the signal rapidly, whereas, dry sand, granite, and ice have minimal impact on it.
• Water content: Water increases conductivity, making it the most common reason a survey falls short of expectations. However, water isn't entirely bad. It also slows the wave propagation, which can improve resolution, and in some scenarios, a moist medium produces better imaging than a dry one. As a general  rule, saline water severely degrades performance while fresh moisture represents a trade-off.
• Antenna frequency: Lower frequencies travel way further but resolve less. Higher frequencies give you sharper images of small targets near the ground surface. This is a hard trade-off. A 1 GHz antenna cannot reach the depth of 10 meters in any soil; a 50 MHz antenna cannot resolve a 5 cm pipe no matter its depth below the ground. Increasing transmit power won't solve this issue, as regulatory caps limit how much energy can be transmitted into the ground, and beyond a certain point, additional power yields diminishing returns.
• Target size, shape, and orientation: Larger reflectors (bigger surface area) return more energy. Flat, horizontal surfaces reflect the most amount of signal. A metal plate tilted more than 45° could be harder to detect than a smaller horizontal one. This is worth remembering before a survey, in case you don't know the orientation of what you're looking for.

These variables do not add up independently. You have to put all four into consideration when measuring GPR depth.

GPR penetration depth by antenna frequency

Here's how a typical GPR unit performs under field conditions, based on reference data across hundreds of surveys. All values below assume what we call "average soil", a medium with relative dielectric permittivity around 9, low conductivity, and low water content. In soils with unfavorable conditions, halve these numbers. In ice or dry sand, double or even triple them.

Terrestrial (ground-based) GPR penetration

Airborne (drone-mounted) GPR penetration

If you fly the same antenna on a drone instead of pushing it in a cart, you lose roughly half your penetration depth. The physics of the air-ground interface is doing the work here. When the antenna is held above the ground, part of every pulse reflects off the air/ground boundary before entering the subsurface.

The "max antenna altitude" column is critical and often ignored. As a rule, your drone should fly low enough that the antenna height stays below the wavelength of the EM pulse in air. For a 500 MHz antenna, that's about 60 cm above ground. For 100 MHz, about 3 m. 

Fly higher, and the data quality collapses sharply. This is why True Terrain Following matters more than most operators realize: a drone that can't hold centimeter-level altitude over rough ground is a drone that can't collect usable GPR data.

An exception occurs during surveys conducted on glaciers, where it is necessary to capture reflections from both the glacier's surface and it's base. To determine when a higher flight altitude is required (flying above the dead zone), the SPH Engineering GPR calculator can be used.

Drone GPR penetration over freshwater

For freshwater with conductivity below 200 µS/cm:

Saltwater is a different story. Saline water is conductive enough to absorb radar energy within centimeters of the surface, which rules out GPR for seawater and brackish environments. For those surveys, a drone-mounted echo sounder is the right tool. Sonar works in the exact conditions that radar doesn't, and SPH's bathymetry systems integrate with the same SkyHub and UgCS workflow you'd use for a GPR mission.

How site conditions can either improve or reduce signal penetration

The GPR depth tables above assume a baseline soil. Real sites deviate from that baseline, often dramatically. Rather than memorize a new table for every soil type, we approach this in multipliers applied to the baseline:

Two practical consequences follow from this.

First, schedule wet-soil surveys for late in the dry season. A thin wet clay layer near the surface can block most of the radar signal all by itself, even if the deeper ground would otherwise be cooperative.

Second, don't confuse ice with water. A frozen lake is an excellent GPR target. The liquid water underneath is usually fine for surveying too, if freshwater. A brackish river mouth or the seawater underneath sea ice will cut your signal off at the first saline interface, and that's where an echo sounder takes over the job.

How deep has GPR actually reached?

Stock frequency-depth tables are starting points. What matters is what happens in the field. Here are three projects that bracket the range from under a meter to more than 100 meters.

100+ meters through ice, the P-38 in Greenland

During World War II, a P-38 Lightning with the call sign Echo went down on the Greenland ice cap. Eighty years of snowfall buried it. When the Fallen American MIA Repatriation Foundation went looking, the survey crew flew a low-frequency drone GPR over the ice under UgCS control. The radargrams came back with anomalies at roughly 100 meters depth.

A number like that would be unthinkable in most mediums. What makes it plausible is the ice itself. Greenland's ice cap is cold, clean, and almost non-conductive, which is about as friendly as it gets for a radar signal. Try the same mission in temperate soil and you'd be lucky to see a tenth of that range.

0.4 to 0.7 m deep, plastic irrigation pipes in the UAE

On 29 October 2024, a DJI M350 RTK carrying a RadSys Zond Aero 500 flew sixteen 50-meter lines at 1-meter spacing over an artificial lawn at Al Lisaili RC Flying Field, UAE. The drone was running with SkyHub, a radar altimeter, and UgCS flight planning. The target was a grid of plastic irrigation pipes.

Plastic is one of the harder targets for GPR. Its dielectric properties are so close to the surrounding soil that the radar signal barely differentiates it. After processing the data in GeoHammer and Geolitix, assuming a dielectric permittivity of 6 for the dry sand, the pipes showed up clearly at 0.4 to 0.7 meters depth. Three conditions lined up to make that work: favorable soil conditions , optimal survey parameters, and the grid was tight enough to cross-check the pattern from line to line.

40+ meters through a debris-covered glacier, Galena Creek Rock Glacier, Wyoming

Galena Creek and Sourdough are two rock glaciers in the United States: Galena Creek in the Absaroka Mountains of Wyoming, Sourdough in the Wrangell Mountains of southeast Alaska. Each one is a solid ice core covered by a thick layer of boulders and sediment. For years, the only way to survey them was on foot with a ground GPR, and coverage stopped wherever the slope got too dangerous to traverse.

A drone-based GPR survey changed what was possible there. The system detected the glacier base at depths greater than 40 meters in Galena Creek, and picked up internal debris layers in cirque areas that nobody had ever walked. A low-frequency ground system can push deeper in absolute terms. The drone's advantage at Galena Creek was 40+ meter depth coverage across the whole glacier, including sections a ground team couldn't safely reach.

How to estimate GPR depth before you fly

If you're trying to predict what you'll get at your site, work through these five steps in order.

Step 1. Characterize the medium

Identify the soil type, its likely water content, and roughly what its dielectric permittivity will be. Approximate values worth knowing are: dry sand ε ≈ 4-6, average soil ε ≈ 9 peat ε ≈ 30-80 , ice ε ≈ 3.2, concrete ε ≈ 6-8, fresh water ε ≈ 80. The higher the permittivity, the slower the wave, and typically the more attenuation.

Step 2. Pick an antenna frequency

Use the depth tables above. The practical rule is to choose the highest frequency that can still reach your target depth. Higher frequency means sharper resolution, and you want as much of it as you can afford.

Step 3. Apply a site multiplier

If you're over dry sand, double the baseline. Ice or snow, triple or quadruple. Wet clay, cut it to a third. For saline ground or saltwater, GPR won't do the job at all, and a drone-mounted echo sounder is your best option.

Step 4. Account for the dead zone

Single-antenna GPR systems (like the Zond Aero LF) have a blind spot directly under the antenna, where the system is still transmitting when reflections from shallow targets arrive. For low-frequency antennas, the dead zone can be a couple of meters. If your target is shallower than that, you need a higher-frequency system or a different geometry.

Step 5. Validate with a calculator

The GPR Calculator we built for our customers takes antenna model, flight altitude, estimated target depth, and soil type, and returns expected resolution, detectability, dead zone depth, and two-way travel time. It's free, takes thirty seconds, and saves you from finding out you picked the wrong antenna after mobilizing to site.

A couple of detection rules of thumb are worth keeping in the back of your head. First, as a general heuristic: a reflector should be at least 10% of the distance between antenna and target, or half the wavelength in the host medium, whichever is larger. 

For pipes specifically, a non-conductive (plastic) pipe is detectable when its diameter is roughly half the radar wavelength in air at the antenna's central frequency. A conductive pipe (metal or water-filled) is detectable when its diameter is about 40% of the wavelength in the surrounding medium.

GPR depth accuracy: how precise is the measurement?

GPR is a time-measuring instrument. Converting that time into depth is a second step, and the accuracy of the conversion depends on how well you know the wave velocity in the medium.

What the instrument records is two-way travel time (TWT) in nanoseconds, the interval between when a pulse is emitted and when a reflection comes back. Depth is calculated afterwards by multiplying half the TWT by the wave velocity in the medium.

That velocity depends on the dielectric permittivity of the medium. Every velocity assumption is an educated guess, and the accuracy of your depth estimate is bounded by how good that guess is.

Some velocities worth knowing:

If a reflection arrives at 80 ns in dry sand with an assumed velocity of 0.15 m/ns, the target is at approximately (80 × 0.15) / 2 = 6 meters depth. If the real velocity turns out to be 0.12 m/ns (slightly wetter sand than assumed), the actual depth is 4.8 m, a 20% error from one bad assumption.

Ways to improve accuracy:

• Calibrate velocity on site using a common-midpoint (CMP) survey or by fitting hyperbolas to a known reflector (a manhole cover, a pipe at known depth). This is the single biggest accuracy improvement available for terrestrial surveys. For drone surveys, applying that same ground-derived permittivity to the full recorded TWT produces large depth errors, because part of the travel time is in air, not ground. Subtract the air-path travel time first (using antenna altitude and c ≈ 0.3 m/ns), then apply the ground velocity to the remainder. Alternatively, process the data with a two-layer velocity model that treats the air-ground interface explicitly.
• Use RTK-GNSS positioning with drone-mounted or terrestrial systems. Centimeter-level horizontal and vertical positioning removes one of the largest error sources in GPR surveys.
• Hold antenna height constant over rough terrain. True Terrain Following makes this automatic on drone surveys.
• Process with proper velocity models rather than a single assumed ε value across the whole survey. Multi-layer velocity models reduce depth error in stratified ground.

With a good on-site velocity calibration, depth accuracy within a few percent is achievable. Without calibration, assume the depth numbers have a 15-30% uncertainty attached, and say so when you hand the report to the client.

Airborne vs terrestrial: which GPR goes deeper?

The ground-based system will always penetrate deeper. If they both have the same antenna, same medium, the cart will out-penetrate the drone by roughly 2×. The reason is mechanical: when the antenna sits on the ground, nearly all of the pulse couples into the subsurface. When it hangs below a drone, fraction of every pulse bounces off the air-ground interface and never enters the ground in the first place.

The airborne penalty is the smallest in ice, snow, and dry sand, where the dielectric contrast between air and surface is mild. In those media, drone GPR can approach ground-system performance closely enough that the trade-off disappears.

What drone GPR gives you in exchange for lost depth:

• Safety. Nobody walks onto unstable glaciers, contaminated sites, or collapse-prone voids.
• Speed. Flight lines are pre-programmed; you cover km² per hour instead of m² per hour.
• Access. You can survey lakes, rivers, and cliff faces where carts won't go.
• Repeatability. Automated missions with RTK and TTF produce datasets you can compare year-over-year without the operator variability that hand surveys introduce.

In a dense city centre, use the cart. On a glacier, fly the drone. Over a solar farm scanning for bedrock? The drone wins on pure economics. Our universal GPR systems are designed so the same antenna deploys either way.

Choosing the right GPR depth for your job

To close the loop:

• Shallow utilities, rebar, thin features, ice thickness (<1 m): 1 GHz.
• Mixed utility mapping, void screening, archeology (1-4 m): 500-600 MHz.
• Geology, bedrock, stratigraphy (4-15 m): 150-300 MHz.
• Glaciology and deep geology (15 m+): 50-100 MHz.

A universal system like the Zond Aero LF family lets you change antennas for the same GPR in a few minutes, which matters when a single job spans multiple depth ranges. Think of a utility survey that also needs bedrock profiling, or a glacier survey that transitions from shallow snow to deep ice in the same flight.

If you want to test your specific scenario before mobilizing, the GPR Calculator will give you a depth and detectability estimate in under a minute. If you'd rather talk through a site with someone who's probably seen a similar problem, get in touch. We've run these surveys on multiple continents and can tell you quickly whether GPR is the right tool for the job.

The honest answer to "how deep does ground penetrating radar go" is: as deep as your medium lets it, and no further. Know your soil, match your antenna, leave a margin, and the number you get will be the one you can defend to your client.

FAQs on GPR depths

How deep can ground penetrating radar scan?

From roughly 0.3 m to over 100 m, depending on antenna frequency and medium. Most utility and archaeology surveys work in the 1-5 m range; glaciological surveys routinely reach 50-100+ m in polar ice.

How far can GPR go in ice?

With a 50 MHz low-frequency antenna, hundreds of meters. Ice has very low conductivity and is effectively transparent to radar. This is why glaciology is the single most depth-favorable application of GPR.

What is "deep GPR"?

A loose industry term for low-frequency systems (typically 50-150 MHz) used for geology, glaciology, and large-scale subsurface mapping at depths beyond ~10 m. Deeper penetration comes at the cost of resolution, so deep GPR sees far but not sharply.

How accurate is GPR depth measurement?

Depth accuracy is dominated by how well you know the wave velocity in the medium. With an on-site calibration (CMP or hyperbola fitting), a few percent error is achievable. Without calibration, assume 15-30% uncertainty and document the assumption.

Can GPR measure depth directly?

No. GPRs measure two-way travel time in nanoseconds. Software converts TWT to depth using an assumed or calibrated wave velocity for the medium.

Does moisture reduce GPR depth?

Yes, moisture reduces ground penetration radar depth significantly. Water raises conductivity, which absorbs the signal. A thin saturated clay layer near the surface can block most of the radar energy even if the deeper ground is cooperative.

Can drone-mounted GPR match ground-based depth?

In ice, snow, and dry sand, yes, nearly. In typical soil, drone GPR penetrates about half as deep as the same antenna on a cart, because some energy reflects off the air-ground interface.

What's the deepest a drone GPR has detected in the field?

In operations we've been part of, anomalies have been detected at approximately 100 m depth through Greenland ice during a P-38 aircraft search. Polar ice conditions make those numbers possible; in typical soils, expect 7-10 m from a low-frequency drone system.