ASSET / SLOPE-LEVEE
Slopes, levees and rivers: seeing long earth structures before the rain comes
Cut slopes, embankments, retaining walls, river levees, revetments and riverbeds. Earth structures run for kilometres, hide under vegetation and fail all at once in heavy rain. This guide sorts out how far drone imaging and laser scanning can go in inspection that has been done on foot, and where people’s feet and eyes are still needed.
On this page
Where slopes and levees start to deteriorate
Unlike concrete or steel structures, slopes and levees deteriorate through the struggle between soil and water. Rain soaks in and weakens the soil, surfaces erode, and water finds paths through the body. Protective shotcrete, frame works and revetments, and supporting retaining walls and ground anchors, crack and loosen over the years.
| Element | Typical defects | If missed |
|---|---|---|
| Cut slopes (shotcrete, frames) | Cracked or hollow shotcrete, voids behind, seepage, loose rocks | Shallow slope failure, rockfall |
| Ground anchors & retaining walls | Damaged anchor heads, tilting walls, opening joints | Deformation or failure of the whole slope |
| Embankments (roads & levees) | Cracks, settlement, sinkholes, bulging or slipping faces | Failure in heavy rain |
| Levee landside toe | Seepage, sand boils (during floods), wet ground | Piping through the levee leading to breach |
| Revetments & riverside face | Broken or displaced revetments, toe scour | Erosion cutting into the levee in floods |
| Around sluices & culverts | Voids around the box, displaced joints | A path for water through the levee |
| River channel | Sediment build-up, tree growth, bed lowering | Lost flow capacity, scour at piers and revetments |
Today’s inspection and its pain points
For river levees, the 2013 revision of Japan’s River Act clarified the duty to maintain, repair and inspect river management facilities; inspection at least once a year is the norm, with emphasis before the rainy and typhoon seasons. Road slopes and retaining walls are inspected under the national inspection guideline for road earthworks. In both cases the basis is walking, looking and feeling the ground underfoot.
The pain is length and grass. Levees run for tens or hundreds of kilometres and must be mown before inspection. Tall cut slopes are inspected hanging from ropes. After heavy rain, the most dangerous places are the hardest to approach. And the municipalities and prefectures managing local rivers and roads are short of civil engineers.
| Method | Strength | Pain point |
|---|---|---|
| Foot patrols & inspection | Catch cracks, wet ground, even animal burrows | Long distances, mowing first, slow |
| Vehicle patrols | Fast and wide | Cannot see faces hidden from the crest (toe, riverside) |
| Rope access (slopes) | Reach tall cuttings and tap surfaces | Fall and rockfall risk, few experts |
| Airborne laser scanning (crewed) | Large-area terrain at once | Expensive, cannot be frequent |
| Boats & sounding (channel) | Measure the riverbed | Hard in shallow or fast water |
What drones change
Long, wide earth structures are a natural fit for drones. Build 3D models from photographs and compare shape with last time. Drone LiDAR can recover the ground under vegetation to a useful degree, and aircraft carrying green (bathymetric) lasers can now measure the riverbed under shallow water.
| Approach | Typical aircraft & sensors | Best for |
|---|---|---|
| Photogrammetry & orthomosaics | RTK GNSS drone with high-resolution camera | Slope and revetment surfaces, crack patterns, change since last time |
| Drone LiDAR | LiDAR-equipped aircraft | Ground under vegetation, settlement, bulging and small slips |
| Green-laser bathymetry | Aircraft with green LiDAR | Riverbed shape in shallow water, scour at revetment toes |
| Thermal imaging | Drones with thermal cameras | Candidate wet areas suggesting seepage (condition-dependent, still exploratory) |
| Long-range patrol | BVLOS aircraft, drone ports | Rapid sweep of levees and channels after a flood |
The basic method is to repeat the same route, the same height, the same season. Vegetation changes how much ground you can see, so compare surveys taken after mowing or in winter dieback. Combine RTK with ground control and check points to keep positional accuracy consistent.
Overcoming the environment
Rivers and hillsides have open sky, so GNSS is usually good. The hard parts are height, water and wind. Tall cuttings mean big elevation differences, and valleys bring strong, shifting winds. Over rivers, water reflections fool sensors, and fog and rain linger after floods.
- Height: on tall cuttings, terrain-following flight keeps image detail consistent. Check the pilot’s line of sight too. Height
- Water & humidity: reflections disturb downward range sensors and image matching. Take care with altitude hold over water and choose weather-resistant aircraft. Water, spray & humidity
- Wind & turbulence: valleys and levee crests are windy, and post-flood inspections often meet bad weather. Set a wind limit in advance. Wind & turbulence
- Long distances: kilometres of flight mean beyond visual line of sight. Plan observers, or BVLOS approval and reliable links.
From images to decisions
For earth structures, the difference over time matters more than any single image. Overlay point clouds to reveal settlement and bulging; compare orthomosaics to follow crack growth. AI is good at proposing cracks, slips and vegetation change. Judging whether a change is *dangerous* still needs people who know the geology, past disasters and rainfall records.
| Step | What machines do | What people own |
|---|---|---|
| Terrain | Build point clouds, DEMs and orthomosaics from images and LiDAR | Check control points and accuracy |
| Change | Compute settlement, bulging and slips against last survey | Tell real movement from error or vegetation |
| Detection | Propose cracks, damaged revetments, seepage spots | Decide which spots to verify on foot |
| Localisation | Tie to chainage markers and stations on the river or road register | Confirm the register match |
| Assessment & action | — | Risk evaluation, repair priority, input to evacuation and closure decisions |
Japan’s public works have built standards around drone surveying (i-Construction). If inspection keeps terrain in the same coordinate system and formats, construction and inspection data join into one record. When disaster strikes, being able to pull up the *previous shape* immediately is the greatest strength.
Rules and responsibility
River administrators (the national government, prefectures and others) are responsible for levees under the River Act; road administrators are responsible for road slopes. In disasters, teams such as MLIT’s TEC-FORCE now routinely survey damage by drone. If drone-derived terrain is used as survey output, it must also follow surveying rules such as the Geospatial Information Authority of Japan’s UAV survey manuals.
Rivers and hillsides are mostly outside densely inhabited districts, but long sections tend to become beyond-visual-line-of-sight flights. Japan allows BVLOS over unpopulated areas (Level 3) and, since December 2023, “Level 3.5”, where an onboard camera checking for people below can replace ground spotters and signage. Abroad, the US Army Corps of Engineers runs a levee safety program and a national levee database (about 6,000 systems, some 23,000 miles), and the Netherlands periodically assesses its primary flood defences against statutory safety standards. Compare countries in world regulations.
What drones cannot do
- What your feet feel: soft ground, small depressions and animal burrows are often found only by walking.
- Inside the levee: soil conditions and seepage paths inside do not show in photos or laser scans; boreholes and geophysics are needed.
- Deep or turbid riverbeds: green lasers struggle with turbidity and depth; boat-based sonar is still needed.
- Judgement during a flood: identifying seepage and sand boils and flood-fighting remain jobs for people on site.
- Risk evaluation and evacuation decisions: interpreting terrain change is a human job.
2035: the world where inspection works
The Airffic editors imagine it like this. Drone ports sit on the roofs of floodgates and pumping stations along the river. Before the flood season, and right after heavy rain ends, aircraft fly fixed routes over levees and slopes. The difference from the previous terrain is coloured station by station. “Five centimetres of settlement here”, “more seepage there” reach the river office and city hall by morning.
People walk only the coloured sections to confirm, and decide on repairs and evacuation preparations. Hillsides and levees shift from “repair after failure” to “act when movement starts”. Nights of heavy rain become a little less frightening.
PREPARATION
Pre-flight preparation checklist
Ticks are saved on this device (and to your shared account if signed in). This is a learning aid, not proof of permission or safety.
Cases for this asset
- MLIT Kurayoshi office: drone trial for levee inspection on the Tenjin RiverYurihama, Tottori, Japan · 2020
- Hong Kong Geotechnical Engineering Office: regulatory-sandbox pilots of LiDAR and thermal drones for slope and natural-terrain monitoringHong Kong, China (first pilot at Tai Mong Tsai, Sai Kung) · 2025
Related papers
- Survey on Coverage Path Planning with Unmanned Aerial VehiclesCabreira, T. M. 2019
Country sources(40)
- Guidebook for Promoting Drone Use and Development in the Construction SectorJP · 国土交通省(大臣官房技術調査課系)
- MLIT expands the Inspection Support Technology Performance CatalogJP · 国土交通省 道路局
- Road management: drones to raise safety and efficiencyKR · 데일리안
- MOLIT to use drones for railway facility inspectionKR · 산업일보
- Busan Port Authority uses drones in winter port safety checksKR · 아시아경제
- Facility management with IoT and drones: railways go smartKR · 세계일보
- Geumho River and Sincheon: stronger facility safety with dronesKR · 경북매일
- Gyeongnam uses drones to check steep slopesKR · 뉴스서울
- Drones patrol mountain railways in Heilongjiang's flood seasonCN · 科技日报
- Technology for flood-season expressway inspectionCN · 扬子晚报/紫牛新闻
- Technical standard for UAV dynamic monitoring of open-pit minesCN · 科技日报
- Miyi: 7 million data points weave a 3D safety net for a mineCN · 川观新闻(四川日报)
Check the evidence
Rules, figures and cases cite the sources below. Future scenarios and design proposals are the editors’ own view.
- MLIT: Evaluation of pre-flood-season inspections of river management facilities (FY2025)MLIT Water and Disaster Management Bureau · Full text read
- MLIT: Performance catalogue of inspection-support technologiesMLIT Road Bureau · Full text read
- GSI: UAV initiatives and latest trends (including the UAV public survey manual)GSI · Full text read
- MLIT: First Level 3.5 drone flight approved in Japan (8 December 2023)MLIT Civil Aviation Bureau · Full text read
- USACE: National Levee DatabaseU.S. Army Corps of Engineers · Full text read
- Dutch House of Representatives paper 34436 no. 3: Explanatory memorandum on amending the Water Act (new standards for primary flood defences), 2016Ministry of Infrastructure and the Environment (Netherlands) · Full text read
- MLIT: Permission and approval procedures for unmanned aircraft (specified flights)MLIT Civil Aviation Bureau · Full text read
Sources checked: 2026-10-22