Radar: below the ground, through the smoke, and watching from orbit

Radar emits radio waves and listens for echoes. Strong echoes come from sudden changes in electrical properties (relative permittivity): air to soil, soil to water, concrete to steel. Voids contrast strongly, so they show well. Waves slow down inside materials, v = speed of light ÷ √relative permittivity, and depth is v × round-trip time ÷ 2. Water’s relative permittivity is about 80, so moisture changes the velocity dramatically — the core difficulty.

A major cause of road collapses is damaged sewer pipes. Through joints and cracks, surrounding soil is washed into the pipe with each storm and flow, and a void grows under the pavement. The surface often looks flat until the very end. The large road collapse in Yashio, Saitama, in 2025 involved a damaged sewer main, and at the national government’s request a special priority survey of old sewers 2 m or more in diameter followed nationwide.

What is mature for road-void surveys today is vehicle-mounted GPR: it scans widely while driving with traffic, and suspicious spots are confirmed by drilling a small hole and inserting a camera. Drones come in where vehicles cannot go — riverbanks, levees, slopes, roads blocked after disasters — and inside sewer pipes with in-pipe drones.

• Height: the further the antenna is from the ground, the more energy reflects at the surface. Drones must skim the ground at a constant low height.
• Weight and antennas: lower frequencies need larger antennas. The deeper you want to see, the harder it is to carry.
• Soil: wet clay and saline soils attenuate strongly. The same kit sees very different depths from site to site.
• Positioning: mapping a reflection needs centimetre-level position and height logs.
• Interpretation: whether an echo is a void, a pipe, a stone or a soil boundary is read from waveform shape and experience. False alarms are common.

So for now, drone GPR is realistically a tool for rough screening where neither vehicles nor people can go. Research is active in snow, glaciers and landmine detection, and inspection trials are growing.

High-frequency GPR is widely used in handheld form to measure rebar position and cover in bridge decks, tunnel linings and building walls. Vehicle-mounted systems also screen for deck deterioration beneath asphalt. These are operational techniques.

Doing this by drone is hard: accurate measurement needs the antenna in contact with, or very close and steady to, the surface. Research exists on drones with contact arms that press sensors against walls or deck soffits, but it remains at the research and trial stage. For now the realistic split is drones screening with visible or thermal imaging, and radar applied up close by people or ground robots.

Millimetre-wave radar is familiar from car collision avoidance. Its longer wavelength lets it measure distance and speed through smoke, dust, fog and darkness. On drones it is used, for example, by agricultural spraying drones for terrain following and obstacle avoidance.

Its weakness is coarse angular resolution: it knows something is there but not its shape. So it works best combined with LiDAR and cameras, each covering when the others are blind. Expectations are high for dusty sites like boilers and mines. Dust & smoke

InSAR (interferometric SAR) images the same area repeatedly with satellite synthetic-aperture radar and measures ground motion from phase differences. Time-series analysis of many acquisitions reveals slow, millimetre-scale displacement of subsiding ground, embankments, levees, dams and bridge surroundings over wide areas. Europe’s Sentinel-1 (C-band, open data), Japan’s ALOS-2 and ALOS-4 (L-band) and commercial small SAR satellites are used.

For rivers and slopes, see slopes, levees & rivers and dams.

Where pictures made to be seen split from records made to measure is told in how aerial imaging divided into pictures and measurements (Airffic Aerial). InSAR is a recent arrival on the measuring side.

• Deep penetration everywhere: wet clay and saline soils attenuate quickly.
• Identifying an echo for certain: void, pipe or stone is confirmed only by digging.
• Close-range sensing by drone: rebar cover and similar need the antenna close and steady — hard in flight.
• Fast change from orbit: InSAR suits slow motion; a collapse that develops in hours is over between acquisitions.

By 2035, satellite InSAR will likely watch city-wide settlement monthly, with vehicle GPR and in-pipe drones dispatched first where movement appears. Drone GPR may have grown into a fast survey tool for roads and levees after disasters. Millimetre-wave radar will probably be standard on aircraft that must keep flying in smoke and dust. This is the editors’ view; drone GPR in particular still needs much more validation.