How to Use Drone Point Clouds in Cultural Heritage Surveys: 5 Items Explaining Accuracy and Procedures

In cultural heritage surveys, there is a strong demand to preserve the configuration of buildings, historic sites, stone structures, and entire sites as accurately as possible. Traditionally, it was common to record data by combining photographs, field measurements, plans, and elevations, but in recent years the use of point cloud data that can preserve shapes in three dimensions has advanced, and the very approach to surveying is changing. Among these methods, the use of drone point clouds is attracting attention as a way to quickly grasp large areas and to easily incorporate high and bird’s-eye viewpoints.

However, drone point clouds in cultural heritage surveys are not something that can be obtained simply by shooting from the air. Cultural properties differ from ordinary buildings and civil engineering structures: they are subject to preservation constraints, have complex shapes, are affected by trees and topography, and require consideration for visitors and managers, among other unique conditions. While they are suitable for grasping roofs and overall exteriors, there are parts that are difficult to confirm except at close range, such as under the eaves, detailed ornamentation, and interior spaces. Therefore, before introducing the method, it is essential to understand concepts of accuracy and the implementation procedures, and to clarify which parts should be handled by drone point clouds.

Also, what matters in cultural heritage surveys is not just producing visually appealing three-dimensional data. It is important whether the acquired point clouds can be connected to practical work such as drawing production, current condition assessment, conservation and repair planning, temporal comparisons, and public use. If acquisition conditions are vague or accuracy standards are unclear, the data may become difficult to use later. Especially for cultural properties, where each target has different conditions, the key to success is to organize survey objectives first, determine the required accuracy, and then assemble feasible procedures, rather than starting with the method.

This article organizes and explains in five items how to utilize drone point clouds in cultural heritage surveys from both accuracy and procedural perspectives. It is written for cultural heritage officers, survey companies, surveying firms, conservation and repair personnel, and others who will need to make practical decisions, summarizing practical ways of thinking to avoid failures in the field as clearly as possible.

Table of Contents

• Background: Why drone point clouds are sought in cultural heritage surveys

• Item 1: Organize survey objectives and required accuracy first

• Item 2: Check site conditions and decide acquisition range and procedures

• Item 3: Stabilize point cloud quality through flight and shooting procedures

• Item 4: Perform accuracy checks and supplement missing parts after point cloud generation

• Item 5: Design operations with deliverable creation and continued use in mind

• Summary: Making the most of drone point clouds in cultural heritage surveys

Background: Why drone point clouds are sought in cultural heritage surveys

The reason drone point clouds are sought in cultural heritage surveys is that they allow efficient understanding of the overall picture of a target. Cultural properties often include not only individual structures but also surrounding topography, stone steps, stone walls, approach paths, gardens, precinct configurations, and the layout relationships of multiple buildings—features whose value is hard to interpret without a broad perspective. Being able to record relationships that cannot be fully captured from ground-level photographs and measurements, by including aerial viewpoints, is a major advantage.

In particular, for building-type cultural properties, it may be necessary to capture roof shapes, upper exteriors, roof ridgelines, around gables, and the composition of the exterior perimeter—areas that are difficult to see from the ground. For historic sites and ruins, the relationships with undulations and terrain, the spread of remains, and positional relationships with the surrounding environment can be important. In such cases, drone point clouds make it easier to record cultural properties not as surfaces but as spaces.

Furthermore, cultural heritage surveys often do not end with a single record. Comparisons before and after repairs, post-disaster checks, tracking chronological changes, and cross-checking with additional surveys may require multi-temporal comparisons. In those instances, a method that can record wide areas in a reasonably consistent way is effective. If viewpoints and plans are organized, drone point clouds can lead to reproducible records and are suitable for ongoing management of cultural properties.

However, it should be noted that drone point clouds do not replace all aspects of cultural heritage surveying. They are suited to recording places visible from the air, but there are limits for under-eaves areas, recessed parts, interior spaces, and fine ornaments or component condition checks. Because cultural properties often have complex shapes and unseen parts that directly affect their value, drone point clouds should be understood not as an omnipotent method but as one that excels at overall comprehension.

For that reason, before introduction, it is necessary to organize how much accuracy is required, what parts should be covered by drone point clouds, and which parts should be supplemented by other methods. The value of drone point clouds in cultural heritage surveys is determined not by whether they can be flown but by whether they are given an appropriate role within the entire survey.

Item 1: Organize survey objectives and required accuracy first

The first item for utilizing drone point clouds in cultural heritage surveys is to organize survey objectives and required accuracy in advance. If this remains vague, even if data are acquired, there is a high possibility that the data will be difficult to use in practice later. In point cloud use for cultural heritage, the required range and accuracy change significantly depending on what and how much you intend to record.

For example, if the objectives are to grasp the current condition of an entire site, confirm building arrangements, record roof shapes, or understand relationships with terrain, drone point clouds are very well suited. In this case, it is important to obtain comprehensive coverage over a wide area so that overall positional relationships are clear. On the other hand, if the objectives are to confirm precise component dimensions, capture the relief of ornamentation, or track very small defects, closer-range recording is necessary and drones alone may be insufficient.

What is important here is not a vague desire to increase accuracy, but to consider the level required according to the deliverables. In cultural heritage surveys, whether the data will be used as basic material for drawing production, as explanatory material for conservation plans, as three-dimensional records for research, or as comparison material for disasters will change the required reproducibility. There are situations in which wide coverage is prioritized and others in which reliable shape reproduction of specific parts is critical.

Also, “accuracy” is not a single concept. It includes multiple meanings in cultural heritage surveys: positional correctness, shape reproducibility, point density per part, lack of gaps, and ease of multi-temporal comparison, among others. A dense-looking point cloud does not necessarily meet required accuracy. Even if the point cloud appears dense visually, it may not be suitable for dimensional verification, while conversely it may be sufficient for overall comprehension. If you do not clarify what the accuracy is for, evaluation criteria will become inconsistent.

Furthermore, it is not uncommon for the client and the implementer to have different images of the required accuracy. If the client expects detailed three-dimensional records while the implementer assumes overall recording, recognition gaps will arise after delivery. In cultural heritage surveys, such discrepancies can greatly affect subsequent stages, so it is extremely important to verbalize and share the concepts of objectives and accuracy at the earliest stage.

In short, before deciding whether to use drones, you need to organize what the survey is for, what range of records is necessary, and the level of accuracy required for each use. Once this organization is done, it becomes clear when drone point clouds should take the lead and when to combine ground surveying or close-range measurement. In cultural heritage surveys, working backward from the desired deliverables rather than starting from the method helps avoid failure.

Item 2: Check site conditions and decide acquisition range and procedures

The second item is to confirm site conditions and then concretely decide the acquisition range and procedures. In cultural heritage surveys, plans that seem viable on paper often cannot be used as-is in the field. The sites for cultural properties are not open vacant lots or simple plots; they often involve a mixture of trees, stone structures, steps, approach routes, visitor circulation, and neighboring buildings.

The first thing to confirm is which faces of the target you want to acquire. The necessary flight area changes depending on whether you intend to capture roofs and the entire exterior, include the terrain of the precinct, or also record surrounding stone walls and slopes. Because cultural properties are often recorded including their relationship to the surroundings rather than as isolated objects, thinking too narrowly about the recording range can make later utilization difficult. Conversely, expanding the range beyond what is necessary increases processing load and prolongs fieldwork. It is important to decide what constitutes the survey target this time, taking into account how it appears on site.

Next, you need to identify areas that are hard to see. For example, buildings with deep eaves, convoluted elevations, stone structures surrounded by trees, and remains under slopes often have many parts invisible from above. Such unseen parts are prone to gaps even if you generate point clouds, and may be impossible to補て later. Identify which areas on site are blind spots and from which directions they can be seen, and plan procedures on the premise of sharing acquisition with ground-based capture as necessary.

You must also confirm takeoff and landing locations and safe waiting positions on site. Cultural heritage sites often lack flat areas, have lots of gravel or grass, have little space available for use other than approach routes, or limit where equipment can be placed for preservation reasons. Even if flight itself is possible, the plan does not work if safe and practical operations cannot be ensured. Cultural properties also often have general users and visitors present, so you need to consider not only flying but also directing people around the site.

Moreover, selecting the time of day is important. At cultural properties, the flow of people and lighting conditions can change greatly depending on the time. Strong backlighting, deep shadows, dappled sunlight, and crowded times affect both point cloud quality and site operations. You should also consider windy periods and times that need to account for ceremonies or public schedules. In cultural heritage surveys, procedural design must consider site operation conditions as well as measurement techniques.

Thus, at the stage of confirming site conditions, it is important not simply to check whether flight is possible but to concretely specify what to acquire, in what order, to what extent, and how. Understanding the site conditions becomes the foundation of quality for successful drone point clouds in cultural heritage surveys.

Item 3: Stabilize point cloud quality through flight and shooting procedures

The third item is to stabilize point cloud quality through the way you conduct flights and photography. In cultural heritage surveys, drone point cloud quality is greatly affected not only by post-processing but also by how you fly and how you shoot in the field. Cultural properties, in particular, often have complex surface materials and shapes and are more susceptible to shooting condition influences than general buildings.

First, be conscious of acquiring the target from multiple viewpoints rather than a single direction. A purely nadir (straight-down) approach may capture roof and ground information well, but it will not sufficiently reproduce elevations, overhangs, or complex exterior outlines. For cultural heritage buildings, information from the top, sides, and oblique directions is important, so a combination of overhead and oblique viewpoints is essential. By securing an overall bird’s-eye view and then planning shoots to fill in required surfaces, you will obtain point clouds with fewer gaps.

Next, avoid unduly close flights. In cultural properties there is a tendency to want to capture fine details, but plans that approach the target too closely introduce not only safety risks but also instability in shooting. Cultural sites often require attention to wind turbulence, proximity to obstacles, and consideration of surrounding sightlines, so it is more realistic to collect required information within flight paths that allow margins. To improve quality, accumulating stable viewpoints is more important than aggressive flying.

Pay attention to lighting conditions. Cultural properties are often located among groves, at the foot of mountains, under deep eaves, or have reflective white walls and tiles—conditions with large changes in light. Strong backlighting or extreme shadows can make surface recognition unstable. If the quality of the acquired images is inconsistent, distortions and gaps are more likely to occur during point cloud generation. Therefore, in cultural heritage surveys you should not rely solely on equipment performance but deliberately choose stable lighting conditions for shooting.

Point cloud quality is also affected by the materials of the target. Cultural properties commonly include mixed materials such as wood, stone, tiles, earthen walls, metal, and vegetation, and the ease of reproduction varies by part. Surfaces with many distinguishing features are relatively easy to capture, while homogeneous surfaces, thin components, and areas with overlapping branches and leaves tend to be unstable. In cultural heritage surveys, proceed with shooting based on an understanding of which parts are easy or difficult to capture, not treating the whole target uniformly.

Additionally, an important aspect during flight and shooting is judging on site whether the necessary data have been obtained. The idea that processing later will somehow fix everything is dangerous for cultural properties. Missing acquisitions and large blind spots are hard to補て after leaving the site. Confirm on site for overall gaps, shortages in specific parts, and viewpoint bias, and be prepared to acquire additional data then and there if necessary. For drone point clouds in cultural heritage surveys, completion depends not on leaving everything to post-processing but on accumulating sound on-site judgments.

Item 4: Perform accuracy checks and supplement missing parts after point cloud generation

The fourth item is to perform accuracy checks and supplement missing parts after generating point clouds from the acquired data. In cultural heritage surveys it is easy to regard the mere generation of a point cloud as a success, but what really matters is whether that point cloud has sufficient quality for the survey objectives. You must not be satisfied with having acquired data; an evaluation process to make the data usable is indispensable.

First, check whether the overall shapes connect naturally. Examine roofs, terrain, building perimeters, stone walls, and slopes for unnatural distortions or steps. Cultural properties often mix straight and curved lines, planes and undulations, and man-made and natural terrain, and local inconsistencies can affect later drawing production and comparative work. You need a perspective that confirms not only neat appearance but also that the target’s shape is correctly reflected.

Next, identify where gaps exist. In point clouds of cultural properties, roofs are often well captured while under-eaves areas, lower elevations of facades, behind trees, and complex overlapping areas are prone to missing data. If these deficiencies do not affect deliverables, there may be no problem, but if they occur in necessary parts,補て must be considered. In practice, it is more pragmatic in cultural heritage surveys to determine whether deficiencies affect the objectives rather than aiming for perfect full coverage.

At the accuracy-check stage, it is also effective to compare with other materials or on-site records as necessary. By comparing with existing drawings, measured values, site photos, or data acquired by other methods, it becomes easier to judge the reliability of the point cloud. Because required accuracy varies by part in cultural properties, you need to examine both overall consistency and reproducibility of important parts. Thorough checking here stabilizes the quality of subsequent deliverable creation.

How you consider補て of missing parts is also important. Areas that remain insufficient with drone point clouds can be supplemented by additional ground-based photography, close-range measurements, or by combining existing materials. In cultural heritage work, it is often more practical to combine the strengths of multiple methods rather than rely on a single method to meet all requirements. For example, use drone point clouds to capture overall layout and roof shapes, and補て under-eaves and blind spots from the ground. Whether such補て planning exists makes a big difference to the final deliverable’s completeness.

Moreover, accuracy checks are not only for the present survey but also relate to future uses. Cultural heritage surveys are often intended for future comparisons, additional surveys, and coordination with conservation and repair, so reuse is assumed. Therefore, it is also important to record where and under what conditions you acquired data and what quality you obtained. Creating a point cloud and managing it as a usable record are different tasks, and in cultural heritage surveys you need to be aware of this distinction.

Item 5: Design operations with deliverable creation and continued use in mind

The fifth item is to design operations so that point cloud acquisition does not end as a one-off task but is integrated with deliverable creation and continued use. In cultural heritage surveys, the real value of acquired point clouds is determined by how they are used afterward. Ideally, they are not only preserved as an overall three-dimensional record but are organized in a form that supports drawing production, repair planning, post-disaster comparison, explanatory materials, research, and comparison with future re-surveys.

First, clarify who will use the data and for what purpose. The way data should be organized differs depending on whether survey staff only need temporary confirmation, a municipality or manager will store it long-term, or it will be used for design and repairs. A point cloud that only its acquirer can understand is not an asset. It is necessary to manage data so that later users can understand what extent was captured, the time of recording, and the purpose of the data.

In cultural properties, the value of continuous comparative records can be greater than single-instance records. For tracking aging, post-disaster checks, and before-and-after repair comparisons, it is important to handle multi-temporal data under the same standards. If positional concepts are vague or acquisition conditions are not recorded, the reliability of comparisons decreases. Therefore, think not only about how to store the current point cloud but also about how to link it to the next survey.

Furthermore, in outdoor cultural heritage surveys the relationship between point clouds and positional information is especially important. Being able to confirm on site which positions within a site were captured and where components are located improves the efficiency of additional surveys and maintenance. Holding an overall point cloud as a record is useful, but connecting it with on-site position management makes the practical work of cultural heritage surveys much more effective.

To successfully introduce drone point clouds, do not view point cloud creation as an isolated task. Consider the whole flow: organizing survey objectives, setting acquisition ranges, confirming accuracy,補て of deficiencies, deliverable creation, long-term storage, and position management. Cultural properties are subjects to be preserved for future use, not things that are recorded once and forgotten. Designing operations on that premise makes it more likely that drone point clouds will become an established part of the survey system rather than a passing trend.

Summary: Making the most of drone point clouds in cultural heritage surveys

To make use of drone point clouds in cultural heritage surveys, do not base decisions solely on the advantage of efficiently acquiring wide areas. Drone point clouds are powerful where aerial perspectives are effective: roofs and overall exteriors, site composition, and relationships with terrain. At the same time, they have limits for fine shapes, hard-to-see surfaces, interior spaces, and parts requiring close inspection, and cannot record every aspect of cultural properties on their own. That is why it is necessary to understand accuracy and procedures and consider how drone point clouds fit into the overall survey.

Looking back at the five items presented here, the first thing to do is organize survey objectives and required accuracy. Next, decide the acquisition range and procedures based on site conditions, and stabilize point cloud quality through flight and shooting practices. After that, perform accuracy checks and補て missing parts post point-cloud generation, and finally design operations with deliverable creation and continued use in mind. Following this flow clarifies when to use drone point clouds in cultural heritage surveys and makes realistic introduction decisions easier.

What is truly important in cultural heritage recording is not that data were acquired, but that they are preserved as records usable into the future. It is not enough to produce visually attractive three-dimensional data at the time; the data must be organized to withstand preservation, repair, comparison, explanation, and transmission. Drone point clouds are a powerful means for that purpose, but they are not universally sufficient. Understanding their strength in overall recording and combining them with ground-based and close-range measurements as needed is practical for cultural heritage surveys.

Finally, to proceed more practically in cultural heritage surveys, consider not only point cloud acquisition but also high-precision on-site position management. On wide sites or sites with multiple targets, being able to accurately identify what was recorded at which positions enhances the precision of continued surveys and maintenance. If you want to operate not only wide-area drone point cloud recording but also on-site position confirmation and multi-temporal data alignment, incorporating iPhone-mounted GNSS high-precision positioning devices such as LRTK into your approach is also effective. If you aim to develop a continuous recording system linked to positional information rather than a one-off measurement, considering such means in the overall design will lead to more practical and reusable operations.