Soil Volume Calculation Point Cloud Introduction: Procedures and Precautions to Visualize Excavation and Embankment with Photogrammetry

Soil volume calculation accompanying excavation and embankment is indispensable at civil engineering and land development sites. Traditionally, survey data were used to create cross-sectional drawings and volumes were calculated using methods such as the average end area method. Recently, however, 3D measurement using point cloud data that captures the entire site in detail has attracted attention. If you photograph the site using photogrammetry and generate a point cloud model, you can intuitively “visualize” where and how much soil was removed by excavation and where soil was added by embankment. This article provides an introduction to using point cloud data for soil volume calculation, explaining concrete procedures and precautions for generating point clouds with photogrammetry and performing volume calculations. We also touch on ways to improve positional accuracy using RTK positioning and the new smartphone-based technology LRTK, and present a perspective on the latest methods that allow anyone to perform simple, high-precision surveying.

Table of Contents

• What are photogrammetry and point cloud data

• Advantages of soil volume calculation using point clouds

• Procedure for calculating soil volumes using photogrammetry point clouds

• Visualizing excavation and embankment: point cloud comparison and volume calculation

• Shaping surface models and defining reference planes

• Key points for improving accuracy: use of RTK positioning and LRTK

• Summary

• Frequently Asked Questions (FAQ)

What are photogrammetry and point cloud data

Photogrammetry is a technique that reconstructs the three-dimensional shape of an object from multiple photographic images to generate 3D models or point cloud data. By photographing the site from various angles using drones, digital cameras, or smartphones and analyzing the images with dedicated software, you can obtain point cloud data, which is a collection of a large number of points that make up the terrain or structures. Point cloud data contain each point's coordinates (X, Y, Z) and color information, and can be visualized as a precise 3D digital model representing the object's shape. For example, photogrammetric surveying of a construction site surface can generate a detailed point cloud model that includes fine irregularities of rocks and soil.

Advantages of soil volume calculation using point clouds

In traditional soil volume calculations, surveyors had to measure heights at fixed intervals on site and create cross-sections on drawings to compute volumes. This method required significant effort and time to measure wide areas, and areas inaccessible to people had to be estimated. In contrast, soil volume calculation using point cloud data obtained from photogrammetry allows you to create 3D models of terrain before and after excavation or embankment and compare their differences, capturing changes across the entire surface accurately. Because point clouds are dense data containing millions of points, high-precision calculations that account for small irregularities are possible. Also, since volumes can be calculated from a mesh generated from a once-acquired point cloud, you can easily recompute volumes for additional sections without re-surveying the site. In practice, some construction sites that previously required four people several days to measure soil volumes switched to creating point cloud models from drone aerial photos and computing volumes, completing the work in one day. Moreover, the point cloud method has shown comparable results to traditional methods, with field validation confirming high accuracy around 1%. Introducing point cloud technology offers the major benefits of improved safety and efficiency while maintaining high accuracy.

Procedure for calculating soil volumes using photogrammetry point clouds

Now let's look at the basic procedure for calculating soil volumes using point clouds obtained by photogrammetry. The following is an example of a typical workflow.

• Shooting plan and photography: First, create a photography plan for the target area. Take photos with high overlap to ensure coverage of excavation and embankment areas with margin. When using a drone, automated flight from altitude is efficient, and handheld cameras or smartphones can be used at ground level by surrounding the subject. The key is to ensure sufficient overlap between photos and to capture the subject from all directions. If the subject lacks patterns or features, placing artificial markers to increase feature points is effective. Also, avoid blur and out-of-focus shots during photography; obtaining high-resolution, sharp images leads to high-precision point cloud generation.

• Generating point cloud data (photogrammetry software): Import the many photos you took into dedicated photogrammetry software to generate point cloud data and 3D models. The software analyzes correspondences of feature points appearing in each photo and reconstructs a point cloud consisting of millions of points based on triangulation principles. With a high-performance PC or cloud service, even a large number of images can be processed in relatively short time. The generated point cloud is often represented in a relative coordinate system, so the next step is to perform scaling and georeferencing.

• Georeferencing to survey coordinates: Align the photogrammetry-derived point cloud model to real-world survey coordinates. To assign absolute coordinates (latitude/longitude or plane rectangular coordinates), it is common to use ground control points (GCPs) measured on site. For example, if multiple GCP targets are placed on site and their coordinates are measured with RTK-GNSS or a total station, you can assign those coordinates to corresponding points on the point cloud to correct the model's overall position and scale. With a sufficient number of control points, you can provide absolute coordinates to the point cloud with an accuracy of several centimeters (a few in). Note that technologies like LRTK, described later, that add high-precision position information to photos at the time of shooting can eliminate the need to install additional control points and allow rapid georeferencing.

• Editing and preparing point cloud data: After georeferencing, prepare the point cloud data for soil volume calculation. Remove unnecessary points and noise, and extract only the point cloud of the ground surface to be analyzed. For example, point cloud data of machinery or trees are unnecessary for volume calculation and should be filtered out. Using ground extraction functions (which classify ground and non-ground) included in photogrammetry software can automatically extract only the ground surface point cloud. For the obtained ground surface point cloud, interpolate locally missing data or delete obvious outliers (erroneous points). If necessary, generate polygon meshes or TIN (triangular irregular network) models from the point cloud to facilitate subsequent volume calculations.

• Setting the reference plane: Define the reference plane for calculating volumes. If you have point clouds from two time points to compare (for example, before and after excavation), prepare each ground surface model and directly compute the difference between the two models. If you want to calculate volume from a single terrain model—such as the volume of an embankment or a spoil heap—you can use a known horizontal plane or existing ground surface as the reference. For example, consider the surrounding existing ground level as the reference plane and compute the volume between that plane and the embankment surface point cloud. Alternatively, set a horizontal reference plane at an arbitrary height (a virtual reference plane) and compute the volume based on elevation differences from that plane. Choose an appropriate reference plane according to site conditions.

• Calculating volumes (embankment and excavation volumes): Once preparations are complete, perform volume calculations. Using software capable of handling point clouds or mesh models, calculate the volume difference between two ground surface models to be compared, or integrate the volume enclosed between a ground surface model and a reference plane. Specifically, this is a numerical integration process that computes differential volumes from height differences between two terrain models. Many software packages also allow you to specify the calculation area with a polygon to compute soil volumes only for that region. As a result of the volume calculation, you can obtain total embankment and excavation volumes (positive and negative volume values) and local volume changes per mesh.

• Reviewing and utilizing results: Verify the calculated soil volumes and use them for site management as needed. Check whether the calculated results are unreasonably large or small by comparing them with known values. Visualizing the point cloud together with volume differences allows intuitive understanding of where and how much excavation or embankment occurred. If necessary, record volume values in reports or drawings and use them to verify quantities with the client.

These are the basic steps. The next section explains visualizing soil volumes by point cloud differencing in more detail.

Visualizing excavation and embankment: point cloud comparison and volume calculation

By comparing point cloud data obtained via photogrammetry and calculating soil volumes, you can visually grasp excavation and embankment conditions. Specifically, overlay point cloud models from before and after excavation or compare the design surface and the as-built ground surface, then calculate volumes from their elevation differences. At this time, rather than obtaining only numerical volumes, displaying a color-coded map (heat map) showing where and how much excavation or embankment occurred provides an immediate visualization of site conformity. For example, if excavation at a certain point is 10 cm (3.9 in) deeper than the design, display it in blue; if embankment is 20 cm (7.9 in) higher than the design, display it in red on the point cloud so that excesses and shortages are intuitively understood.

Volume calculation for embankment and excavation typically aggregates the positive volumes (embankment portions) and negative volumes (excavation portions) between differential models. This allows calculation of how many cubic meters of soil were added or removed in total. Comparing with the design plan shows quantitatively whether soil is sufficient or surplus, aiding schedule management and revision of soil hauling and placement plans. A point cloud–based calculation also offers the flexibility to re-aggregate volumes by arbitrary sections as needed. For example, if the construction area is divided into a grid or arbitrary regions, you can calculate excavation and embankment volumes for each part, supporting detailed construction management. Using point clouds for soil volume calculations dramatically improves the accuracy and efficiency of as-built control.

Shaping surface models and defining reference planes

To perform high-accuracy soil volume calculations, shaping the surface model derived from the point cloud and defining an appropriate reference plane are important. First, regarding shaping the surface model, photogrammetrically obtained point clouds may contain some errors and noise. For example, point cloud data that include parts of vegetation or machinery can create irregularities different from the true terrain and distort volume calculation results. Therefore, removing non-ground points and, if necessary, smoothing the surface is recommended. After meshing, if there are locally abnormally sharp parts, smooth them or fill holes to shape the model to match the actual terrain. Such preprocessing increases the reliability of volume calculations.

Next, defining the reference plane determines what you compare volumes against. The basic approach is to compare “terrain at one time” with “terrain at another time,” but you may sometimes use a planned surface on design drawings or a hypothetical horizontal plane as the reference. For example, when calculating the volume of an embankment, you may treat the surrounding ground as the reference plane and calculate the volume of the part that has risen above it. For measuring the volume of voids created by underground excavation, you might consider the pre-excavation ground as the reference and take the difference with the post-excavation point cloud model. In any case, choosing the wrong reference plane can render the numbers meaningless, so set an appropriate reference according to the purpose of the site. Generally, in as-built management, use the design surface as the reference to take differences with actual results, while for progress management use the original terrain as the reference to compute hauling and placement volumes.

Key points for improving accuracy: use of RTK positioning and LRTK

Improving positioning accuracy is key to further enhancing the accuracy of point cloud–based soil volume calculation. Photogrammetry alone can produce relatively precise models, but high-precision position information is essential to align the entire model to accurate survey coordinates. One useful technology is RTK-GNSS positioning. RTK (real-time kinematic) combines data from a base station and a rover to correct errors in satellite positioning and perform real-time centimeter-level positioning (cm level accuracy (half-inch accuracy)). Using RTK in point cloud measurement makes acquiring coordinates for ground control points easier, and by assigning high-precision geotags to aerial photos you can directly reflect absolute coordinates in photogrammetry results. Recently, solutions called LRTK, which incorporate RTK receivers into smartphones, have emerged. By attaching a small GNSS antenna to a smartphone and taking photos or scans, you can immediately assign global positioning coordinates to the acquired point clouds. In other words, without specialized equipment or complex post-processing, anyone can easily perform high-precision point cloud surveying.

With LRTK, while scanning the site with a smartphone camera or LiDAR sensor, RTK positioning coordinates are overlaid on the point cloud data in real time. This means the resulting point cloud model is from the start an accurate terrain model referenced to survey coordinates, eliminating the need for later alignment with control points. Also, acquired point cloud data and measurement results can be stored and shared on the cloud, allowing immediate office review of on-site scans and smooth collaboration with stakeholders. LRTK, which enables high-precision and efficient surveying by anyone on site, is expected to greatly simplify future soil quantity and as-built management.

Summary

Using photogrammetry-derived point cloud data is a powerful way to understand the as-built conditions of earthworks such as excavation and embankment in detail and intuitively. Calculating and color-coding soil volumes on a 3D model makes it easy to see at a glance what and how much was done on site, contributing to faster schedule management and prevention of rework. This article explained the workflow and key points for using point clouds in soil volume calculations, but the key to achieving high accuracy remains appropriate data acquisition and processing. High-precision photogrammetry requires sufficient overlapping photos and georeferencing with control points, and recently LRTK that combines smartphones and RTK has dramatically simplified this process. By adopting new technologies, soil volume management that once relied on surveying specialists is becoming something site personnel themselves can easily perform. Consider actively using point cloud–based soil volume calculation to support smarter and more efficient construction management.

Frequently Asked Questions (FAQ)

Q1. What is soil volume calculation using point cloud data? A1. Soil volume calculation using point cloud data is a method of determining terrain volume changes using 3D data (point clouds) composed of numerous points obtained by laser scanners or photogrammetry. Rather than creating cross-sections from limited survey points as in traditional methods, this approach compares high-density point cloud models that measure the entire ground surface, enabling soil volume calculations that accurately capture fine topographic changes due to excavation and embankment.

Q2. How do you calculate volume from a point cloud obtained by photogrammetry? A2. To calculate volume from a point cloud obtained by photogrammetry, first convert the point cloud into a ground surface 3D model (mesh or TIN), then perform differencing with another model or a reference plane. For example, overlay the pre-excavation terrain model with the post-excavation model, compute elevation differences, and integrate those differences across the area to determine the excavated volume. Similarly, embankment volumes are obtained from model differences before and after embankment. Dedicated software can automatically calculate volumes from point clouds and display results numerically or as heat maps.

Q3. Can point cloud data for soil volume calculation be acquired without a drone? A3. Yes. While drones are effective for efficiently photographing wide areas, you can still generate point clouds with photogrammetry by taking many photos from the ground with a camera. Also, many modern smartphones and tablets include LiDAR sensors that can quickly scan point clouds at short range. By using a small GNSS receiver attached to a smartphone (such as LRTK), you can obtain high-precision point clouds with handheld smartphone shooting, so even without drones or expensive laser scanners, the environment for on-site soil volume measurement is becoming feasible.

Q4. How accurate are soil volume calculations using point clouds? A4. It depends on conditions, but with proper measurement and processing, errors in point cloud–based soil volume calculations are often within a few percent. Field validations have reported differences from traditional survey results of about 1% in some cases. However, ensuring accuracy requires high-quality photos (or scan data), a sufficient number of ground control points (features), and accurate georeferencing. Georeferencing with RTK-GNSS or known control points improves the model’s absolute accuracy and enables reliable soil volume calculations.

Q5. What is LRTK and how does it help with soil volume calculation? A5. LRTK is a new solution that combines a smartphone with RTK-GNSS technology to achieve high-precision surveying. By attaching a small RTK-capable GNSS receiver to a smartphone and performing photography or LiDAR scanning, the acquired point cloud data can be assigned high-precision position information in real time. The advantage of LRTK for soil volume calculation is that the point cloud model is obtained already aligned to the survey coordinate system, removing the need for post-processing alignment. That means anyone can easily acquire accurate point clouds and immediately compute excavation and embankment volumes on site. LRTK is a technology that can greatly improve efficiency and reduce labor in soil quantity management.

Q6. Can point cloud soil volume calculation results be displayed as a heat map? A6. Yes. Results of point cloud–based soil volume calculations can be displayed as heat maps (colored maps). Dedicated point cloud processing software and cloud services provide functions to color-code elevation differences between design models and surveyed point clouds. This makes it easy to identify where large differences exist. For example, green indicates almost as designed, blue indicates over-excavation (lower than design), and red indicates overfill (higher than design), allowing intuitive evaluation of construction as-built results. Heat maps visually identify problem areas, aiding corrective action identification and quality control.