How to Calculate As-Built Condition and Earthwork Volume from Point Clouds | 10 Checks to Prevent Measurement Errors

Table of Contents

• What is point cloud data?

• Main methods of point cloud surveying (drone LiDAR and terrestrial laser scanner)

• How to calculate as-built shape from point clouds

• How to calculate earthwork volume from point clouds

• 10 checks to prevent measurement errors

• Summary

In recent years, the use of point cloud data for as-built management and earthwork volume calculation at construction and civil engineering sites has attracted attention. A point cloud is a collection of numerous three-dimensional points obtained by scanning a site with lasers or photogrammetry, a form of digital data that records the shapes of terrain and structures in detail. Traditionally, measurements were taken manually with tape measures and levels, cross-sections were drawn, and volumes were calculated using the average cross-section method. However, by utilizing point cloud data, the entire site can be captured in 3D and accurate as-built conditions and volumes can be calculated, helping to prevent missed measurements and improve efficiency. In this article, we explain the specific methods and workflow for calculating as-built conditions and earthwork volumes from point clouds, and finally introduce ten checkpoints to prevent measurement errors.

What is point cloud data?

Point cloud data (point cloud) is three-dimensional (3D) data that represents the surfaces composing real-world space with a large number of points. Each point is assigned X, Y, and Z coordinates (and sometimes color information), and as an assembly of points it reproduces the shape of an object with high precision. For example, if you convert an earthwork site for civil engineering into a point cloud, you can digitally copy the site in detail—from the irregularities of the ground surface to the positions of structures. Because point cloud data contains high-density information amounting to millions to hundreds of millions of points, it can capture subtle undulations and dimensional errors that manual surveying cannot fully detect. In as-built management it is necessary to confirm whether the completed structure matches the design shape, and by using point clouds you can record the site as it is and later measure arbitrary cross-sections and dimensions, dramatically improving the reliability of quality control.

Main Methods of Point Cloud Surveying (Drone LiDAR and Terrestrial Laser Scanners)

Methods for acquiring point cloud data primarily use drone-mounted LiDAR (laser scanners) and terrestrial laser scanners (TLS). Drone LiDAR can scan large areas from the air in a short time and is often used for topographic surveys of forests and development sites. Aerial laser measurement has the advantage of capturing the ground surface through obstacles, making it effective for understanding the terrain surface geometry beneath tree cover. On the other hand, terrestrial laser scanners are devices mounted on tripods that measure 360 degrees around from the ground, characterized by millimeter-level high-precision point clouds. They are suitable for as-built surveys of structures and close-range precision measurements, but on large sites they require scanning multiple positions several times.

In addition to these laser scanners, a method called photogrammetry (photogrammetry), which converts drone aerial photographs into point clouds using software, has also become widespread. Photogrammetry extracts feature points from overlapping image photos in dedicated software and generates point cloud models through triangulation calculations. Drone photogrammetry has relatively low equipment costs and can efficiently capture wide areas, but to obtain high-precision results it requires sufficient image overlap and position correction using ground-installed control points. In recent years, simple LiDAR scanning functions built into iPhone and iPad have also appeared, increasing the situations where point clouds can be obtained easily. However, point clouds obtained with smartphones and the like often include errors on the order of a few centimeters (a few inches), and to achieve reliable accuracy for practical use, survey-grade GNSS and coordinate alignment and verification with known points are indispensable.

How to calculate as-built geometry from point clouds

The basic principle of calculating as-built quantities using point cloud data is to compare the as-built point cloud and the design data and evaluate the differences. First, the site after construction is surveyed with a laser scanner or drone to acquire point cloud data of the finished terrain and structures. Next, prepare the design drawings or a 3D design model and align it to the same coordinate system as the point cloud (georeference it). Representative comparison methods are cross-section comparison and 3D difference checking.

In cross-section comparisons, cross-sectional shapes are extracted from the point cloud at predetermined positions (for example, at regularly spaced survey points in road construction) and overlaid with the design cross-section lines to check for deviations. For each cross-section, the as-built dimensions (height, width, thickness, etc.) are checked to see whether they meet the design values, and if there are deficiencies or excesses the construction is corrected. This approach is similar to conventional methods, but because many cross-sections can be verified at fine intervals that would be difficult to do manually, the risk of overlooking issues is greatly reduced.

A 3D difference check of point clouds is effective for more comprehensively grasping deviations across an entire site. This method calculates, for each point in the acquired point cloud, the difference from the corresponding design surface (the design's finished 3D model) and creates a color-coded heat map of the height differences. If a point in the point cloud is above the design it is treated as a positive deviation, and if below as a negative deviation; for example, areas with deviations of +○ cm (○ in) or more are colored red, and those with -○ cm (○ in) or less are colored blue. By doing so, one can immediately and intuitively see which parts of the finished shape are raised (overconstruction) or excessively cut (underconstruction) compared with the design. Because the results are color-coded they are easy to understand even for non-specialists, and they are useful as explanatory materials for clients and for self-checks prior to inspection. In fact, the Ministry of Land, Infrastructure, Transport and Tourism (MLIT) has introduced a new standard called "surface management" to evaluate the entire as-built using areal data such as point clouds, making quality inspections far more comprehensive than conventional point measurements. For example, in pavement work, measurements used to be limited to thickness at a few locations, but with point cloud heat maps one can analyze the entire finished surface for undulations and thickness distribution, directly enabling early detection and correction of out-of-spec areas.

With as-built management using point clouds, once the data are captured you can record the entire site, including parts that will be backfilled after construction. With manual surveying there have been cases where missed measurement points became problems later, but with point clouds, even if additional dimensional checks are needed afterward you can always extract cross-sections or remeasure dimensions from the saved data. Also, because the point cloud data itself remains as a digital record, there is the advantage that during future renovation planning or additional construction the saved data can be used to efficiently assess current site conditions.

How to calculate earthwork volume from point clouds

Calculation of earthwork quantities (volume calculation) has also been greatly transformed by the use of point cloud data. Traditionally, surveyors measured heights in the field at regular intervals, created cross-sections from those point clouds (point sets), and calculated volumes using the average-section method. This approach required a great deal of effort and time to determine earth volumes for large sites, and steep slopes that were inaccessible sometimes had to be estimated.

With point-cloud-based methods, the terrain before and after excavation or before and after embankment construction is each modeled in 3D and the differences are calculated, allowing the overall shape changes of the site to be captured accurately and completely. Specifically, the pre-excavation surface point cloud and the as-built post-construction point cloud are each converted into mesh surfaces such as triangulated irregular networks (TINs), and the volume difference arising from their height differences is calculated by numerical integration. It is also possible to calculate the enclosed volume from the height differences between one point-cloud model and an arbitrary reference plane (horizontal plane or design plane). For example, in embankment work, comparing the pre-development terrain model with the design finished-surface model will yield the required fill volume, and the total volume of the embankment can also be obtained from the height differences between the finished-surface point cloud and a horizontal reference plane.

The advantages of volume calculation using point cloud data are improved accuracy from high-density data and flexible recalculation. Because point clouds include fine terrain undulations, they yield volumes closer to reality than estimates based on a limited number of points. In an actual case where earth volumes were calculated from a point cloud model generated from drone photographs, accuracy comparable to conventional methods was confirmed, with errors of about 1–2%. Also, if acquired point cloud data are meshed once, it is easy to later recalculate earth volume for additional sections as needed. Without re-surveying the site, there is flexibility to immediately respond to requests such as "I want to know the fill volume of just a particular part." In terms of efficiency, there are reports that earth volume measurements that previously required four people and several days were completed in one day by switching to drone aerial photography and point cloud analysis. Point cloud technology, which can measure wide areas in a short time, contributes greatly to shortening schedules and reducing labor.

Furthermore, using point clouds enables visual volume management. Not only can you obtain numerical volume results, but you can also "visualize" the entire site with a color-coded heatmap showing where and how much excavation or filling has occurred. For example, if areas excavated deeper than the design are shown in blue and areas that have been overfilled and raised too high are shown in red on the point cloud, you can intuitively grasp the distribution of excess and deficient soil volumes. This allows site supervisors to assess construction status at a glance and, if necessary, immediately order additional excavation or fill adjustments. In this way, earthwork volume calculation using point clouds is valuable not only for producing accurate figures but also for directly linking that information to on-site decision-making and quality control.

10 Checks to Prevent Measurement Errors

Using point clouds to calculate as-built conditions and earthwork volumes is a powerful technique, but if not properly planned and executed there is a risk of errors and oversights. Below are 10 checkpoints that site personnel should keep in mind to prevent measurement mistakes.

（1）Confirm the purpose and required accuracy: Before starting point cloud surveying, clarify what you want to obtain and verify the accuracy requirements that correspond. Identify whether it is for as-built inspection or for earthwork volume calculation, and what error range is acceptable. Because the appropriate measurement range and point cloud density differ depending on the purpose, defining the goal at the outset leads to efficient, non-wasteful planning.

(2) Selection of methods and equipment: Choose the optimal measurement methods and equipment according to the purpose and required accuracy. If you only need a rough estimate of the volume of soil over a wide area, drone photogrammetry may be sufficient; if structural inspections requiring accuracy of a few centimeters or less are needed, a high-performance terrestrial laser scanner is necessary. If you will be measuring at night, also consider the equipment’s lighting capability. Taking into account the characteristics of each type of equipment (for example, laser scanners are highly accurate but expensive, while photogrammetry is economical but requires care in alignment), consider the optimal combination.

(3) Pre-site environmental preparation: It is also important to arrange the environment of the area to be measured. Remove or relocate any unnecessary obstacles within the survey area as much as possible in advance (e.g., heavy machinery and vehicles, materials, overgrown vegetation, etc.). Reducing items that block the line of sight lowers the risk of missed measurements due to blind spots and reduces noise points. You should also take weather and time of day into account. Rain and fog disturb laser distance measurements, and strong sunlight contrasts or the dimness at dusk affect photogrammetry results. Optimize environmental conditions by choosing a clear daytime whenever possible, or by installing lighting as needed.

（4）Establishment of control and check points: High-accuracy point cloud surveying requires known control points (reference points). If multiple targets with known coordinates are placed on site in advance, alignment between point clouds and between point clouds and design data can be performed accurately after surveying. Especially when integrating multiple scans or drone photogrammetry, without common reference control points the datasets can become misaligned. It is also effective to install check points separate from the control points and verify the coordinate errors at those locations after point cloud processing. By checking whether the errors at the check points fall within the required tolerance, you can objectively evaluate the reliability of the resulting point cloud data.

（5）Equipment calibration and confirmation of settings: The equipment to be used should undergo accuracy checks and calibration in advance. For laser scanners, perform horizontal and vertical adjustments; for photographic equipment, complete lens distortion correction and focus checks before departure. Parameter settings during measurement are also important. For laser scanning, set the resolution (angle step), measurement range, and scan speed appropriately. Increasing resolution raises point density but increases the amount of data to process, and increasing speed too much increases the number of missed points, so strike a balance according to the purpose. For photography, adjust shutter speed and ISO sensitivity to capture sharp images without motion blur or underexposure. When equipment and settings are fully prepared, they greatly contribute to ensuring surveying accuracy on site.

(6) Improvements to measurement methods (overlap and blind-spot countermeasures): In point cloud surveying, it is essential to ensure sufficient data redundancy (overlap). For drone imagery, low overlap between adjacent photos tends to introduce errors in point cloud reconstruction, and with laser scanning, if the coverage of adjacent scans is too far apart, alignment becomes unstable. Carefully plan flight paths and scan placements so that adequate overlap is achieved. Also pay attention to missing data caused by equipment blind spots. A terrestrial laser scanner cannot measure the hidden sides of objects from a single direction, so scan multiple times from different positions as needed to compensate for blind spots. For complex-shaped objects or sites with significant elevation differences, combining different viewpoints—such as from above and from the side—produces better results. During measurements, movement of people or vehicles can cause noise and data gaps, so it is important to take measures to ensure stable measurement conditions, for example by temporarily restricting access.

(7) Immediate on-site data verification: Once surveying is complete, make it a habit to check the acquired data on-site. If you only notice missing data or equipment problems after returning to the office, you may need to revisit the site and remeasure, causing losses of time and cost. The minimum items to check on-site are: "whether any areas were missed," "whether there is any data loss due to blind spots or obstructions," and "whether any obvious anomalies or noise have occurred." If a problem is found, immediately perform supplemental surveying, such as changing the equipment position and conducting additional scans. Fortunately, recent software allows point clouds to be previewed on tablet devices and even provides basic analysis on-site. By using such tools, you can detect defects immediately on-site and correct them the same day, minimizing rework.

(8) Point cloud data processing and noise removal: The collected point cloud data should be preprocessed appropriately to improve quality. Specifically, filter out obvious erroneous points (outliers) that are located in clearly wrong positions and noise points caused by reflections from airborne particles, birds, and the like. Because remaining unnecessary points can adversely affect subsequent analysis, perform noise filtering carefully. However, removing too much is also prohibited. Excessive thinning (downsampling) that eliminates subtle ground undulations will lead to decreased accuracy in as‑built evaluation and earthwork volume calculations. It is important to strike the right balance: remove only unnecessary points while maintaining the required point cloud density. Also remove unrelated point clouds that are outside the measurement target. For example, including adjacent buildings, power lines, temporary structures, and so on—items irrelevant to as‑built evaluation—not only increases data size but also interferes with comparisons to design data. Employ clipping to narrow the analysis area and automatic classification functions that classify and extract only the ground surface and structures, and prepare a clean dataset that leaves only the point cloud necessary for analysis.

(9) Point cloud registration and coordinate system verification: When integrating multiple measurement datasets or design models with point clouds, perform precise alignment (registration). During georeferencing using control points, match the control points in the software and calculate the errors, adjusting until the mean error is within the specified tolerance. Even when relying on automatic alignment algorithms (such as ICP), perform a coarse alignment first so there are no large initial offsets; this improves accuracy. After alignment, verify for several known points that the point cloud coordinate system (e.g., plane rectangular coordinate system ◯ system or a proprietary local coordinate system) and the vertical datum have been correctly applied. Discrepancies in coordinate systems can lead to serious errors. For example, if the vertical datums between point clouds are offset, the calculated earthwork volumes will be completely incorrect, and if the horizontal positions differ from the design model, acceptance decisions regarding as-built conformity will be mistaken. Always check the consistency of coordinates and scale, and if there is any doubt, investigate the cause by comparing with field survey measurements.

(10) Verification and Sharing of Deliverables: We also make sure to double-check the final as-built dimensions and earthwork quantities. Even if the point-cloud analysis software has automatically calculated the results, visually verify that there are no obviously anomalous values. For example, visually inspect whether an unreasonably large volume of earth has been reported or whether the error color map of the as-built contains any unnatural patterns (e.g., unmeasured areas left as blank/white holes). If possible, it is also prudent to compare a few representative locations with measurements obtained by conventional methods. Measure the dimensions of important cross sections with a tape measure or total station and verify how closely they match values computed from the point cloud to increase confidence in the results. Finally, share the acquired point-cloud data and analysis results within the team and with the client, and have them checked from a third-party perspective. Point-cloud digital data is easy to share by email or via the cloud, allowing multiple people to review the same 3D model. Multiple-person verification helps further prevent human error and avoid transcription mistakes in inspection documents.

By thoroughly checking the above 10 items both before and after, you can greatly reduce errors in as-built management and earthwork volume calculations using point cloud measurements. Common issues such as forgotten measurements, recording mistakes, and coordinate errors can also be prevented in advance through point clouds and appropriate procedures.

Summary

We explained methods for calculating as-built conditions and earthwork volumes using point cloud data, and the key points for ensuring accuracy. By incorporating point cloud surveying, you can digitally record the entire site geometry and calculate any dimensions or volumes with high accuracy and efficiency. Small surface irregularities and differences in earth volume that were overlooked in manual surveying can be reliably detected with point clouds, directly improving quality control and preventing rework. Of course, it is essential to properly carry out everything from preplanning through data processing and validation, but if you proceed based on the checklist items introduced in this article, you should be able to avoid major problems.

In recent years, technologies that allow anyone to easily perform 3D point cloud measurements have also emerged. For example, using LRTK (an iPhone-mounted GNSS high-precision positioning device) enables high-precision site surveying with just a smartphone. LRTK gives RTK-class positioning accuracy to surveying performed with a phone camera or LiDAR scanner, delivering precise positioning that previously required specialized equipment in a palm-sized form. Simply walking a site with a smartphone and LRTK can acquire high-precision point cloud data and immediately perform as-built checks and earthwork volume calculations—it is no longer a dream. The use of the latest tools is steadily advancing labor savings in surveying work and providing real-time feedback. With the spread of point cloud technology and digital surveying, construction management is becoming more reliable and efficient. Please consider adopting these new measurement methods in ways that suit your company’s operations to help improve on-site management accuracy and operational efficiency.