Earthwork Volume Calculation — The New Standard in Point Clouds: Rapid As-Built Computation with Photogrammetry × High-Precision GNSS

Table of Contents

• Introduction

• The new trend in earthwork volume calculation using photogrammetry and point clouds

• Photogrammetry vs. laser scanning: comparison of methods and accuracy

• Point cloud surveying advanced by high-precision GNSS

• Procedure for calculating earthwork volumes using point cloud data

• MLIT as-built management guidelines and CIM utilization

• LRTK enabling simple surveying and AR construction support

• FAQ

Introduction

In civil engineering, earthwork volume calculation is a crucial process for accurately determining “how much soil and earth was moved” during filling and excavation. Traditionally, ground elevations before and after construction were measured to create cross-sections, and volumes were calculated using the average cross-section method. However, that approach relied heavily on manual surveying and hand calculations, demanding significant labor, and as the area increased, variations in accuracy and workload became problematic.

Recently, attention has turned to calculating earthwork volumes using point cloud data. By scanning an entire site and processing the countless 3D measurement points (point cloud) obtained, terrain volumes can be measured quickly and with high accuracy. Especially with advances in photogrammetry and the proliferation of drones, detailed point clouds can now be acquired for large-scale earthworks in a short time. Moreover, combining this with high-precision GNSS (such as RTK) has dramatically improved the positional accuracy and coordinate assignment of point clouds, enabling unprecedentedly fast as-built (final shape) assessment. This article explains the innovation of using photogrammetry combined with high-precision GNSS for as-built measurement as a cutting-edge “earthwork volume × point cloud” approach, and outlines concrete procedures and application points.

The new trend in earthwork volume calculation using photogrammetry and point clouds

Using point clouds for earthwork volume calculation offers overwhelming improvements in both efficiency and coverage compared to traditional methods. For example, at one large site that previously required four people seven days to measure and calculate volumes, switching to a drone photogrammetry workflow to generate point clouds and compute volumes allowed the same work to be completed by just two people in one day. Work time was reduced to roughly 1/14, while the calculated as-built quantities showed errors of about 1% compared to conventional methods, confirming high accuracy. Point cloud data capture fine terrain undulations, reducing oversights and enabling reliable volume estimation.

Point cloud–based volume calculations are also useful for construction progress management. For instance, if you compute the as-built volume immediately after excavation is completed, you can identify differences from the design quantities on the spot. Decisions such as arranging additional backfill soil or planning disposal of surplus soil can be made quickly without stopping work. Traditionally, compiling survey data took time and as-built management tended to lag behind, but with point clouds you can quantify site status in near real-time. In the current era where DX (digital transformation) is demanded, this is a powerful tool for construction management.

Furthermore, the acquired point cloud itself becomes a precise 3D record, making it easy to perform additional volume calculations for arbitrary areas after measurement. Once scanned, you can respond immediately in the office to requests like “recalculate the volume for a particular section.” Detailed as-built comparisons that were difficult with paper drawings or photos can be done freely on a digital point cloud model. For these reasons, point cloud–based earthwork volume calculation is becoming the new standard in civil construction.

Photogrammetry vs. laser scanning: comparison of methods and accuracy

Representative methods for obtaining point clouds are photogrammetry and 3D laser scanning. Each has its characteristics, but in recent earthwork volume projects, photogrammetry adoption has increased due to cost and ease of use. Here is a brief comparison.

• Photogrammetry: A technique that reconstructs 3D shapes from many photos taken with a camera. The site is photographed from various angles using drone aerial photography, DSLR cameras, or smartphones, and specialized software analyzes feature points to generate point clouds. Because the model is derived from color photos, it is visually intuitive, and the equipment is relatively inexpensive, centered on cameras. Large areas can be photographed quickly, and airborne captures allow data collection over steep slopes or high areas that are hard to approach on foot. However, to achieve high-accuracy results you need to shoot with sufficient resolution and overlap, and the subject should have adequate texture or features for matching. Glass surfaces, water, or uniformly white walls may lack detectable feature points and be difficult to reproduce. Also, photogrammetry’s algorithms mean that reconstruction accuracy is influenced by camera placement and shooting angles, with vertical accuracy often inferior to horizontal accuracy. Therefore, as described later, combining photogrammetry with control points (ground control points) or high-precision GNSS can achieve on the order of several centimeters (cm level accuracy (half-inch accuracy)) in survey accuracy.

• 3D laser scanning: A method that directly measures distances to targets by emitting laser light and acquiring point clouds. There are various types, including tripod-mounted terrestrial 3D laser scanners, vehicle- or drone-mounted mobile scanners, and handheld scanners. Laser scanning can acquire very dense points—hundreds of thousands to millions of points per second—and can measure in dark conditions, being less affected by lighting. Point accuracy per measurement can be on the order of millimeters (for high-end equipment), providing a more consistently stable accuracy compared to photogrammetry. Laser scanning is particularly advantageous for vertical accuracy and for capturing ground beneath forest canopy, and it is reliable in complex terrain or nighttime surveys. However, laser scanners are expensive, and large models require operator expertise. Materials like glass or mirrors (reflective/transparent) and very dark, light-absorbing objects can be difficult to measure. Overall, photogrammetry is advantageous for initial cost and ease of operation, while laser scanning often wins in accuracy and environmental adaptability. In practice, it is ideal to use or combine both to complement each other’s weaknesses depending on site conditions and required accuracy.

Point cloud surveying advanced by high-precision GNSS

What dramatically improves the positional accuracy challenge of photogrammetry is the use of high-precision GNSS. Typically, a point cloud generated by photogrammetry is floating in an arbitrary coordinate system and may not align with real-world geodetic coordinates. Traditionally, several ground control points were placed on site and their marker positions were used to adjust the model to provide absolute coordinates, a time-consuming task. Using RTK-GNSS (real-time kinematic GNSS) makes high-precision workflows much smarter.

RTK-GNSS is a surveying method that reduces positioning errors to below a few centimeters by adding correction information from a ground base station to GNSS satellite positioning. With improvements in communications infrastructure, network RTK services allow centimeter-class positioning nationwide. Mounting an RTK receiver on a drone lets you high-precision tag the capture positions (geotags) of each aerial photo, and the point cloud generated by photogrammetry can be obtained in an accurate coordinate system from the start. Even for ground photography, recording camera positions with a GNSS device synchronized to the camera allows automatic alignment of the entire model to real coordinates during post-processing. In a sense, it’s like “surveying at the same time as taking photos.”

The biggest advantage of combining high-precision GNSS is that it reduces the effort required for coordinate alignment in the point cloud workflow. Time spent installing control points and surveying known points can be drastically reduced, simplifying the work flow. The resulting point clouds can be immediately overlaid with electronic maps and design data, facilitating downstream use. For example, as-built point clouds obtained by an RTK-equipped drone can be compared directly with 3D design models (BIM/CIM models) for as-built inspection without coordinate transformation. Combining the ease of photogrammetry with GNSS precision makes efficient, high-accuracy as-built measurement possible even over large areas.

Procedure for calculating earthwork volumes using point cloud data

Now let’s review the concrete procedure for calculating earthwork volumes using point cloud data. In earthworks, the basic approach is to compute the “volume difference between a reference surface and the current surface.” A surface model of the ground is created from the point cloud, and volumes are derived by comparing surfaces.

• Acquire as-built conditions with point cloud surveying: First, acquire point cloud data for the target area. Survey at each timing you want to compare—before and after construction, or before and after fill, for example. Methods such as drone photogrammetry or ground laser scanning are acceptable, but it’s important to cover the terrain with sufficient extent and point density. Clean the acquired point cloud of unnecessary noise and distant objects (such as machinery or people) so it is suitable for analysis.

• Extract ground surface and create a DTM: Extract only points that correspond to the ground surface from the point cloud. If points include non-ground objects such as construction machinery, trees, or structures, use filtering or automatic classification to remove them. Once you have ground points, generate a digital terrain model (DTM) such as a TIN (triangulated irregular network) or mesh model. Creating a surface that represents ground undulations with a network of triangles enables area and volume calculations.

• Set the comparison surface (reference surface): Prepare the reference surface to compare with the current terrain. Depending on the case, representative examples include comparing the “pre-construction original ground” with the “post-construction as-built ground” to determine volume changes, or comparing the “designed planned ground” with the “as-built ground” to check for excesses or shortages. The comparison target should likewise be converted to a DTM from point clouds or created from design data such as CAD drawings or a BIM model.

• Compute differences and calculate volumes: With both surfaces ready, overlay them and compute the volume difference. Specifically, calculate the volume by integrating the vertical distance (height difference) between the two surfaces over the area. Dedicated earthwork calculation software or 3D analysis tools can automatically compute the differential volumes between surfaces. The results will show total volume change and can also break down fill volumes (positive volumes) and cut volumes (negative volumes) by area.

• Validate and utilize results: Validate the calculated volumes as needed. For example, place independent check points at key locations and compare their height differences with the point cloud differential results to verify errors. If accuracy is confirmed, reflect the volume results in as-built quantity determinations and construction management documents. Another advantage of point cloud–based volume calculations is that the 3D model used as the calculation basis can be stored as evidence. If future disagreements arise with inspectors, you can present the point cloud data and say, “This is the shape and this is the volume,” making consensus building smoother.

That is the overall flow. The point is that point cloud–based workflows measure wide areas at high density, so there are fewer “hidden errors.” Small terrain depressions that sectional methods might miss are reflected, improving volume estimation accuracy. Once modeled, changing the calculation area or recalculating can be done with the press of a button, enabling efficient quantity calculations without rework.

MLIT as-built management guidelines and CIM utilization

The government has taken note of the effectiveness of as-built measurement and earthwork volume calculation using point clouds, and official operational standards are gradually being established. Since the late 2010s, the Ministry of Land, Infrastructure, Transport and Tourism (MLIT) has promoted ICT-utilized construction as part of the i-Construction initiative, supporting the introduction of 3D technologies on sites. For example, the guideline “As-Built Management Using Aerial Photogrammetry (UAV) (Earthworks edition)” (draft 2020) outlines procedures and accuracy management methods for using drone photogrammetry-generated point clouds for as-built management. It also specifically describes high-precision shooting methods using RTK-GNSS, and error verification using control points and check points, paving the way for treating photogrammetric point clouds as an official as-built measurement method. With these standards in place, contractors can more confidently use point cloud technology. When measurements are performed in compliance with the as-built management guidelines, there are increasing cases where clients accept point cloud–derived deliverables.

Beyond as-built measurement, there is also movement to utilize CIM (Construction Information Modeling). CIM, akin to BIM for civil engineering, involves using 3D models consistently from design through construction and maintenance. Importing as-built terrain captured as point clouds into a CIM model makes it easy to compare with design models and reuse the data for maintenance. For example, overlaying as-built point clouds on design data to verify quality or using the completed 3D terrain for planning future renovation works are possible—digital data enable lifecycle reuse. In the context of CIM, point cloud data is valuable not just as measurement results but as a “digital record of construction.” MLIT has positioned point clouds and 3D model utilization within CIM guidelines following the as-built management guidelines, and it is envisaged that 3D data submission will become standardized in the future. For surveying and management engineers at sites, point cloud skills will become increasingly important.

LRTK enabling simple surveying and AR construction support

A new technology that further lowers the barrier to point cloud utilization is the emergence of high-precision surveying devices using smartphones × compact GNSS. A representative approach is called LRTK. LRTK involves attaching a palm-sized RTK-GNSS receiver to a smartphone and using a dedicated app to perform photo capture and point cloud scanning—an up-to-date solution. This makes centimeter-class 3D measurement, which previously required expensive survey equipment, easily achievable with an everyday smartphone.

LRTK makes small-site and routine as-built checks significantly more efficient. For example, a site manager can walk around a grading area with a smartphone and quickly scan it, generating a high-precision 3D model with positional coordinates on the spot. Data can be shared to the cloud immediately, allowing measurements of volume or distance and decisions to be made without returning to the office. Tasks that used to be outsourced to specialists for as-built surveying can now be done in-house in a short time, giving small and medium-sized contractors a powerful tool to advance DX.

Moreover, mobile measuring devices like LRTK have high affinity with AR (augmented reality) technology, and they are expected to evolve into construction support tools. By overlaying acquired point clouds or design models onto the real scene through a smartphone or tablet screen, as-built checks and layout work can be performed intuitively. For example, holding up a tablet over a partially completed embankment and color-coding differences between the design model and the as-built point cloud instantly shows where to cut or fill. AR visual assistance helps prevent operational mistakes while fully leveraging the accurate data obtained from point cloud measurement. The fusion of smartphone RTK devices with point clouds and AR has the potential to elevate site management to the next stage. As easier-to-use measurement technologies spread, a time will come when everyone routinely handles 3D data.

FAQ

Q1. How accurate is earthwork volume calculation using point clouds? A1. It depends on measurement conditions, but combining drone photogrammetry with RTK-GNSS typically yields on the order of a few centimeters (cm level accuracy (half-inch accuracy)). There are cases where volumes calculated from photogrammetric point clouds fell within ±1–2% of traditional methods. Point clouds from laser scanners can achieve millimeter-level accuracy, but for general earthwork as-built quantity management, photogrammetry + RTK is sufficiently practical.

Q2. Is a drone essential for photogrammetry? A2. Not necessarily. Aerial photogrammetry is effective for quickly capturing large areas in large-scale sites, but photogrammetric point clouds can also be generated from close-range photos taken with a smartphone or handheld camera. Recently, solutions like LRTK that attach small GNSS units to smartphones and enable easy high-precision point cloud generation have appeared. Choose the optimal capture method for the site scale and target.

Q3. Does MLIT accept photogrammetry in its as-built management guidelines? A3. Yes. MLIT has provided trial guidelines for as-built management using UAV photogrammetry, and if certain conditions are met, as-built evaluations can be performed using photogrammetric point clouds. Specifically, processes such as verifying accuracy with check points before and after shooting and demonstrating that errors are within prescribed ranges (for example, around 5 cm (2.0 in) in the vertical direction) are required. Local bureaus and municipalities have begun operations in line with these guidelines, so check the latest procedures before use.

Q4. What are the pros and cons of photogrammetry compared to laser scanning? A4. The advantages of photogrammetry are lower equipment cost and the ability to quickly measure large areas. It also produces intuitive 3D models from color images. The downsides are that results are more susceptible to shooting conditions (lighting and subject texture), leading to reduced quality at night or for textureless targets, and processing can take time. Laser scanners require higher initial investment but provide stable, high-precision measurement and can more easily acquire data in dark conditions or under vegetation. In short, photogrammetry offers “speed and ease,” while laser scanning offers “accuracy and versatility.”

Q5. Can point cloud volume calculations be done without special software or skills? A5. User-friendly point cloud processing software and cloud services have become more common, and environments that allow semi-automatic volume calculations without specialized knowledge are emerging. Some drone manufacturers’ analysis software and construction-oriented 3D volume tools can generate point clouds and compute volumes automatically just by uploading photos. Nevertheless, it is important to follow proper measurement procedures and perform accuracy validation. Although tools are becoming easier to use, having basic surveying knowledge and a habit of checking data is advisable.