Are Drone Point Clouds Effective for Dam Inspections? 7 Things to Know Before Introduction

Table of Contents

• What are drone point clouds?

• Improving inspection safety

• Streamlining inspection work

• Recording and analysis with detailed 3D data

• Equipment and skills required for introduction

• Flight planning and regulatory compliance

• Point cloud data processing and accuracy management

• Summary

Dams are essential social infrastructure for securing water resources, flood control, and power generation, and regular inspections are indispensable to maintain their safety. In particular, many dams built since the high economic growth period have been in service for decades, making measures against aging a major issue. The shortage of personnel responsible for inspection work is also becoming more serious. In conventional dam inspections, workers typically visually check for cracks and leaks or use measuring instruments to measure structural deformation. However, inspections at height or of large structures require considerable effort and involve danger. In recent years, new inspection methods using drones (unmanned aerial vehicles) have begun to emerge in this field. Among these, attention is focusing on whether the "point cloud data" (three-dimensional surveying data) acquired by drones is effective for dam inspections.

In practice, efforts have begun to fly drones equipped with cameras or laser scanners around dams to create three-dimensional models of the structures. However, when introducing new technology, you may have questions such as “Is it really effective?” and “What are the hurdles for introduction and operation?” This article explains seven points that practitioners considering the introduction of drone point clouds for dam inspections should know in advance. We will go through the benefits for safety and efficiency as well as operational cautions in order.

1. What are drone point clouds?

Drone point clouds are three-dimensional point cloud data obtained using drones. Point cloud data is a collection of countless points arranged to cover the surface of the target, with each point having XYZ coordinates (positional information). Point clouds obtained using laser scanners (LiDAR) or photogrammetry are digital data that precisely reproduce the shape and dimensions of a dam.

For example, if a drone is equipped with a high-resolution camera and takes many photos of the dam body that are then analyzed with dedicated software, a three-dimensional point cloud model can be generated from image overlap. Alternatively, if a drone carries a compact laser scanner and emits laser light during flight to measure reflection points, it is also possible to directly obtain a set of 3D coordinates. By viewing the resulting point cloud data, you can check detailed three-dimensional information that is difficult to grasp from photos or drawings, such as subtle irregularities on concrete surfaces and the overall dam shape. Also, because the acquired state at the time of acquisition can be saved as a digital record, it is useful for inspection history management.

Point cloud generation by photogrammetry is easy to interpret visually because it is derived from color images, but on surfaces with few patterns like concrete, feature points can be hard to detect and accuracy may decline. It is also affected by weather and lighting at the time of shooting. On the other hand, methods using LiDAR can measure in dark conditions within the range where light reaches and can capture shapes of textureless objects, which is an advantage. However, drones equipped with LiDAR are more expensive in terms of equipment cost than camera-only systems and require specialist knowledge for operation. Depending on purpose and budget, it is advisable to use photogrammetry and LiDAR appropriately.

2. Improving inspection safety

One of the primary expected benefits of using drone point clouds is improved safety. Regular dam inspections often involve workers visually checking the surface of embankments that can reach tens of meters (tens of feet) in height or performing sounding inspections. Traditionally, workers would descend the embankment using ropes or erect temporary scaffolding to perform high-altitude work. Such work at height is always accompanied by the risk of falls, placing a heavy burden on workers.

Drones can approach dangerous heights or areas where people cannot enter without a person being present. For example, the upstream slope of an embankment (the reservoir side) may be difficult to access because there is no access road, but a drone can approach from over the lake to capture images. Even locations that used to be checked by a person descending a 75 m (246.1 ft) ladder can be replaced by aerial photography with a drone, avoiding putting workers at risk. With drone point clouds, cameras or lasers can acquire conditions remotely, reducing the need for scaffolding or suspended work and greatly lowering the risk of fall accidents.

Also, immediately after a major earthquake or heavy rain when the safety of the area around a dam cannot be confirmed, drones can investigate damage without risking secondary disasters. Even when people cannot approach, obtaining video and point cloud data remotely can help determine the presence or absence of abnormalities early and consider appropriate countermeasures.

Furthermore, drone inspections are non-contact, so they do not place loads on structures. Instead of workers walking on concrete surfaces or striking with hammers, data can be collected from above without contact, eliminating concerns about damaging the structure during inspection. In infrastructure inspections where safety comes first, drone point clouds can be a groundbreaking risk reduction measure.

3. Streamlining inspection work

The introduction of drone point clouds is also expected to streamline inspection work. Inspecting a vast dam body in detail by manpower requires many personnel and days. For example, inspecting every part of a dam with a dam height near 100 m (328.1 ft) may take time to set up temporary scaffolding or ascend and descend with gondolas, and covering the entire structure can take several days to several weeks.

On the other hand, using drones enables rapid data collection over wide areas. Flying drones overhead or along slopes and taking photos can accumulate short flights to acquire point cloud data of the entire dam. It is not uncommon for areas that would have taken a full day by manual labor to be covered in a few hours with drones. This can significantly shorten inspection schedules and improve operational efficiency for dam managers.

Also, once point cloud data has been acquired, necessary measurements and checks can be performed in the office, contributing to efficiency. For example, if something was overlooked on site, that area can be zoomed in on later in the point cloud data to measure dimensions or recheck conditions. This reduces the need to return to the site for additional surveys, saving travel time and labor costs. By obtaining detailed data in a short time with drone point clouds, limited time and resources can be used more effectively.

Furthermore, reducing workload may allow for more frequent regular inspections. Whereas annual inspections were once the limit, using drones makes it easier to plan inspections every six months or quarterly, enabling continuous monitoring systems. With drones handling data collection, workers need spend less time on dangerous high-altitude tasks and can concentrate on higher-value tasks such as data analysis and planning repairs.

4. Recording and analysis with detailed 3D data

Data obtained from drone point clouds can record the current condition of a dam in detail and be used for various analyses. Unlike ordinary photos, point cloud data contains coordinate information in real scale for each point. Therefore, it is easy to measure distances and areas of arbitrary parts on the data or create cross-sections. Tiny changes that are not obvious to the naked eye can be captured as numerical values, enabling inspections and evaluations based on scientific evidence.

For example, for concrete surface cracks, combining high-resolution orthophotos allows accurate measurement of their length and width. Measurements of crack lengths that used to be done manually with measuring tapes can be read easily on the point cloud. Additionally, by accumulating point cloud data over multiple years, it is possible to grasp trends in displacement or settlement of the dam body by comparing with past data. In practice, overlaying regularly acquired dam point cloud models and analyzing them has detected slight bulging or tilting changes, enabling early repair decisions.

Moreover, point cloud data is useful for repair design and long-life planning of dams. If a current 3D model or drawings are created from point clouds, discrepancies with the actual site during renovation design of aged dams can be reduced. Even for old dams without remaining past drawings, point clouds can provide detailed as-built diagrams, enabling reliable planning. Thus, detailed data obtained from drone point clouds serves not only as inspection records but also greatly contributes to subsequent analysis and advanced maintenance management. Acquired 3D models and point cloud diagrams are also useful for explaining to stakeholders and informing residents. Using three-dimensional visuals makes it easier for non-experts to intuitively understand the dam’s current condition.

Initiatives such as i-Construction and CIM promoted by the Ministry of Land, Infrastructure, Transport and Tourism are also a tailwind, and infrastructure management using point cloud data will become increasingly important. Various guidelines are being developed, and industry-wide utilization of 3D inspection technology is expected to accelerate further.

5. Equipment and skills required for introduction

Introducing drone point clouds requires corresponding equipment and technical skills. On the equipment side, a drone that can fly stably and high-precision measurement sensors are indispensable. While point cloud creation is possible with general aerial photography drones, more capable aircraft are desirable for dam inspections. Examples include drones of a certain size that are less affected by wind, models equipped with high-resolution cameras, and, in more advanced cases, industrial drones fitted with laser rangefinders. Because GPS signals may be disturbed near the dam body, drones with sensors or programs that can self-control aircraft attitude are reassuring. It is realistic to start operations with aerial photography drones and photogrammetry-based point cloud creation, and consider introducing LiDAR-equipped machines as needed in a phased approach.

Next, shooting and surveying skills are also important. Acquiring drone point clouds is not as simple as just flying; careful planning is required in advance regarding flight routes and shooting angles. To cover the entire dam body thoroughly, methods such as setting up an autonomous flight plan to take photos at regular intervals or capturing the structure from oblique directions multiple times are effective. Operators need to be proficient not only in piloting drones but also in the basics of photogrammetry and the operation of point cloud processing software. If there are no specialists in-house, outsourcing to external surveying companies or drone inspection services is an option, but obtaining appropriate results can be difficult without a certain level of knowledge on the client side.

Cost considerations are also important. Introducing high-performance drones, LiDAR equipment, and point cloud processing software requires initial investment. Training operators also takes time and money. However, once a system is established, improvements in inspection efficiency and safety can bring returns. When introducing, compare the investment required for equipment and personnel with the expected benefits and judge from a long-term perspective whether it is economically viable.

6. Flight planning and regulatory compliance

When conducting dam inspections with drones, prior flight planning and compliance with regulations are essential. For flight planning, preparations that consider the dam’s unique environment are necessary. Mountainous dams are prone to strong winds and turbulence, so it is important to assess weather conditions and make reasonable plans. GPS instability near the dam body may require precise manual control or placing visual observers to ensure safety. Battery remaining management is also crucial; for large dams where it is not possible to cover everything in one flight, plan to divide the area and fly multiple times. In case of a drone crash, measures such as equipping flotation devices to prevent submersion and enable recovery are sometimes taken.

Next, regulatory compliance. When flying drones within Japan, procedures under the Aviation Act and related laws are required. Although the airspace above dams is often not densely populated, depending on the situation there are multiple permits and approvals required—for example, approval for flights below 30 m (98.4 ft) when third-party access cannot be restricted. It is also important to obtain individual flight permission from managers of critical infrastructure facilities such as dams. In many dams, prior application to the management office is required, and you cannot fly drones without permission. Also, if you plan to fly the drone where it cannot be seen by the pilot, such as behind the embankment, this falls into beyond-visual-line-of-sight flight and requires separate authorization. Nighttime inspections likewise require flight permits and approvals. Allow sufficient time for application procedures when planning flights under such conditions.

In addition, a new system that started in 2022 requires registration of aircraft above a specified weight and has introduced a pilot licensing system for Level 4 flights. Check the latest regulations and complete the necessary qualifications and notifications before operation. Thorough legal compliance and consideration for surrounding safety will allow safe use of drone point clouds.

7. Point cloud data processing and accuracy management

It is important to properly process and analyze the large amount of point cloud data acquired by drones and manage its accuracy. Generating point clouds from hundreds or more photos requires processing steps in dedicated software to align images and create a 3D model. Processing uses high-performance PCs or cloud services and can involve calculations that take from several hours to several days. The resulting point cloud models are denoised and, as needed, transformed to the site’s survey coordinate system.

Accuracy management is particularly important. If point clouds are generated relying only on the drone’s GPS, coordinate shifts of several tens of centimeters may occur. Since structural integrity evaluations for dams require accuracy on the order of a few centimeters (cm-level accuracy (half-inch accuracy)), measures to ensure higher accuracy are indispensable. Generally, known coordinate points called control points (ground control points, GCPs) are installed on site, and by including those points in the point cloud data, the overall model accuracy is corrected. For example, targets can be placed in locations with good visibility around the dam and their coordinates surveyed with high-precision GNSS or a total station. In addition, drones equipped with RTK-GNSS that perform high-precision positioning in real time have appeared. Using such aircraft can reduce the number of required control points while increasing overall accuracy, but since errors are not completely eliminated, it is desirable to set several verification points to check accuracy. Recently, GNSS receivers that can be attached to iPhones and easily achieve centimeter-level positioning have also emerged. Utilizing such devices can help obtain high-precision reference point information on site without specialized surveying instruments and contribute to improving the accuracy of drone point clouds.

When point cloud data is thus refined for high accuracy, it becomes reliable enough for long-term monitoring and detailed deformation analysis. Back up and store processed data appropriately. Because point cloud files are large, consider using cloud storage or data compression when sharing within the organization. Sharing 3D models and drawings generated from point clouds with stakeholders also facilitates smooth information sharing of inspection results. Establishing an end-to-end flow from acquisition to processing and utilization and enforcing data quality management are keys to successful drone point cloud introduction.

Summary

Point cloud surveys using drones have the potential to bring significant transformation to dam inspection sites. While there are many advantages—ensuring safety, improving operational efficiency, and leveraging high-precision data—there are also challenges to clear, such as introducing specialized equipment, acquiring technical skills, and complying with regulations. If you carefully prepare by addressing the points raised here before introduction, drone point clouds can become a powerful tool for dam maintenance and management.

Of course, drone inspections alone cannot detect all abnormalities. Final judgments still require visual confirmation by experienced engineers and other inspection methods, but by efficiently collecting current condition information with drones, tasks that must be performed by people can be carried out more safely and accurately.

In terms of accuracy management, complementing drone positioning accuracy with high-precision GNSS is key. For example, using the high-precision GNSS receiver “LRTK” that can be attached to an iPhone allows you to easily obtain reference point coordinates on site. Combining such new measurement technologies with drone point clouds will further improve the accuracy and reliability of dam inspections. Moreover, cases of drone inspections being introduced domestically and internationally are increasing. In the future, this may become a new standard technology to compensate for the decline in skilled technicians and the shortage of personnel. While incorporating cutting-edge tools, consider using drone point clouds to help maintain and manage infrastructure in the future. Actively adopt new technologies to achieve safe and resilient infrastructure maintenance and management.