ArbNation Join free

Ground-penetrating radar for tree roots: accuracy statements in three journal studies

Three studies tested 900 MHz ground-penetrating radar on tree roots. One reports r2 of 0.76 and 0.81 under pavement, and two report errors that grow with depth or shrink with root size.

Checked against the sources at the bottom of this page on October 16, 2026. Rules, fees and pay change: the source has the last word.

Three journal papers tested ground-penetrating radar, or GPR, on tree roots. The studies compare radar output with roots of known number or placement. All three used a 900 MHz antenna. None tested trunks.

The three studies side by side

GPR root detection, as the authors report it
Study and setupReported result
Bassuk, Grabosky, Mucciardi and Raffel, July 2011. Norway maples planted in 2003 in two compacted soils, native silt loam and a structural soil, under 10 cm of concrete. Scanned in 2008, then root systems dug out with an air excavation tool and countedRoot count correlations of r2 = 0.76 for the native soil and r2 = 0.81 for the structural soil. In the native soil the radar overestimated roots at the minimum detection diameter. The authors conclude that GPR can identify root presence, if not size
Hirano and co-authors, 2009. Roots of Cryptomeria japonica in sandRoots wider than 19 mm showed up in the radar output. Roots with under 20 percent volumetric water content were not detected. Two close roots were not told apart. The authors say these factors lead to underestimating root biomass
Moore and Ryder, September 2015. Roots of 10, 20 and 40 mm buried at 200, 400 and 800 mm; paired roots; a small Chinese pistachio dug out afterwardAccurate detection at 200 mm, with error rising at 400 and 800 mm and phantom roots in some scans. In the tree trial, GPR missed many small roots near the trunk and read several roots as one

Depth and soil in the field results

Moore and Ryder give depth figures for their buried roots. At 800 mm the largest predicted depth was 440 mm and the average was 345 mm, about half the burial depth. They suggest the antenna available for the tests was probably a cause, with interference or weak signal processing as the most likely sources of error.

In the pistachio trial the outer trenches returned 52 predicted roots in total, and excavation found one. The authors describe the soils as disturbed urban soils. Bassuk and colleagues call for caution in reading root colonisation where the soil has a massive structure with small cracks full of very small roots.

Where the studies differ

The papers use different measures, so the figures do not stack. Bassuk and colleagues report a regression of radar counts against excavated counts. Hirano and colleagues report a size and moisture threshold. Moore and Ryder report position and depth errors. The Bassuk study scanned through concrete in built soil, Hirano in sand and Moore and Ryder in trenches of alluvial sandy loam and around one urban tree. Trunk scanning is covered in the guide to sonic and acoustic tomography.

Sources