What is the accuracy of LiDAR measurements for cultural properties? 6 checkpoints to avoid failure

Interest in LiDAR surveying for the documentation, preservation, and maintenance of cultural properties has been increasing year by year. Even where records were traditionally compiled from drawings, photographs, manual measurements, and close visual inspection, LiDAR is increasingly being selected as a method that can quickly acquire three-dimensional shapes. In particular, for subjects in which complexity of form and spatial extent coexist—such as stone structures, buildings, archaeological remains, wall surfaces, staircases, and the terrain of historic sites—the mere ability to preserve them in three dimensions is of great value.

However, what practitioners really want to know is not whether LiDAR can take measurements, but how accurate it is and for which applications it can be relied upon. In cultural heritage contexts, it is not enough merely to generate a three-dimensional appearance. Questions are always raised about whether it can withstand future comparative surveys, whether it can serve as the basis for repair plans, whether it can be used to confirm changes in form, and whether it can be explained to third parties as a record. Therefore, if the notion of accuracy is left vague, the data painstakingly acquired may end up being usable only as a 3D model for viewing.

Furthermore, cultural heritage differs from common civil engineering structures and industrial products in that the material, degree of weathering, surface reflectance characteristics, conservation environment, and access conditions can vary greatly from one object to another. Some objects are stone that retains fine carvings and incised lines, others are dark-colored wooden elements, walls that include cracks, metallic decorations that readily reflect light, or intricate interior spaces. When the nature of the object and the measurement conditions change, the results obtained from the same LiDAR survey can vary significantly. For this reason, in LiDAR surveys of cultural heritage it is important not to discuss accuracy based solely on equipment specifications, but to confirm the objectives, standards, planning, on-site conditions, processing methods, and the intended use of the deliverables.

In this article, we organize the concepts of accuracy in LiDAR measurement for cultural heritage and explain six items that practitioners should check to avoid failure. It is compiled from a practical perspective so that those who are about to commission measurements, those considering measurement plans within their organization, and those who want to assess the validity of recorded results will not be confused in the field.

Table of Contents

• Why accuracy is important in LiDAR surveys of cultural heritage

• Check item 1: Calculate the required accuracy by working backward from the intended use

• Checklist item 2: Align the approach to reference points and coordinates

• Checklist item 3: Do not overlook the material of the object under inspection and the on-site environment

• Checklist Item 4: Finalize the measurement plan and measures to prevent missed measurements

• Check item 5: Confirm the flow of point cloud processing and error verification.

• Checklist item 6 Design the specifications of the deliverables and future operations

• To stabilize the recording accuracy of cultural properties.

Why Accuracy Is Important in LiDAR Surveys of Cultural Heritage

Accuracy in LiDAR measurements for cultural heritage is not simply a matter of smaller numbers being better. In practice, several perspectives come into play at once: the extent of absolute positional deviation, the reproducibility of relative geometry within the object, how well fine surface undulations can be represented, and whether the data were captured without gaps. For example, the type of accuracy to prioritize differs when you want to record the layout of an entire site versus when you want to compare the wear of inscriptions on a stone monument. In the former, coordinate consistency is important; in the latter, local shape reproducibility and resolution are more important.

When working with cultural heritage, the fact that it is difficult to redo work later also makes ensuring accuracy difficult. This is because constraints often overlap: restricted access; inability to touch or reinstall objects for conservation reasons; limited opportunities for re-measurement due to exhibition schedules; and sensitivity to weather and lighting conditions. Furthermore, it is not uncommon for recording purposes to span multiple uses beyond preservation, such as pre- and post-restoration comparison, condition monitoring, exhibition use, academic research, and educational use. Under these circumstances, data that appeared sufficient on site can become clearly inadequate as soon as one attempts to use it for a different purpose in later stages.

Another important point is that accuracy requirements for cultural heritage vary according to the value and risk of the subject. Important ornamental elements, areas where monitoring crack progression is necessary, and parts that serve as reference points for displacement monitoring tend to be inadequately handled if treated the same way as general areas. Conversely, in cases where breadth should be prioritized—such as confirming the terrain or layout of an entire site—pursuing unnecessary detail-level accuracy can cause operational breakdowns in terms of field time and data-processing load. In other words, accuracy in LiDAR surveying of cultural heritage should be understood as whether you can design reproducibility that is commensurate with the subject’s value, intended use, and constraints.

Therefore, what practitioners should check is not the instrument’s advertised accuracy specifications themselves. It is important to clarify under which conditions those figures hold, what error sources increase when the target or field environment changes, and what you want to guarantee as a deliverable, and to determine whether the measurement plan is aligned with those assumptions. From here, we will examine six specific checkpoints for that purpose.

Checklist Item 1: Derive the required accuracy from the intended use

The first thing to confirm is whether you can articulate which cultural property, for what purpose, and to what level of accuracy is required. In consultations about LiDAR surveys, the desire to capture data with high accuracy often takes precedence, and discussions may proceed while the definition of the intended use remains ambiguous. However, if the purpose is vague, neither the acquisition density, the on-site observation positions, nor the criteria for post-processing will be determined. As a result, the deliverable tends to be inefficient, with necessary areas captured too coarsely and only unnecessary areas captured excessively finely.

For example, if the goal is to capture the current plan and cross-sections of an entire historic site, what matters is the overall spatial connectivity and positional relationships. In that case, priorities are minimizing blind spots, maintaining the continuity of terrain and structures, and ensuring consistency with reference points rather than local fine detail. On the other hand, if you want to record the condition of carvings, incisions, weathering, flaking, or losses, you must place greater emphasis on the faithful reproduction of surface geometry. Furthermore, if you are planning for future comparisons over time, you need to preserve conditions that allow data to be reacquired under the same approach in subsequent surveys, not just data that looks good this time.

The important point here is not to express required accuracy with a single number. In LiDAR surveying of cultural properties, it is necessary to separate and organize multiple aspects such as absolute coordinate accuracy, relative accuracy between components, the ability to represent surface detail, the low incidence of missing data, and reproducibility upon remeasurement. For example, it is more practical to divide the subject and set requirements accordingly—for the whole site prioritize positional accuracy, while for important parts prioritize density and local accuracy. If this organization is not done, differences in understanding about accuracy among field personnel, the client, and analysts can arise, leading to problems where the delivered data cannot be used for the intended purposes.

Because non-contact methods are the rule for cultural heritage, you need to clarify to what extent information that cannot be verified by touch should be represented as 3D data. Whether you want 3D records as a supplement to photographs, as foundational material for drafting, or for quantitative comparison in condition assessments will change the required level. LiDAR is not a panacea; subtle tonal differences, nearly transparent parts, and assessing the condition of different materials are sometimes more effectively addressed by combining other recording methods. Start by working backwards from the intended use and clearly delineate what LiDAR will cover and what should be left to other recording methods; this is the starting point to avoid failures in accuracy.

In practice, during meetings it's useful not only to confirm the final appearance but also to anticipate which cross-sections will be taken, which parts will serve as comparison targets, and in what units changes will be tracked in the future, as doing so helps specify the required measurement conditions. In LiDAR surveying of cultural heritage, what truly matters is not aiming for the highest possible precision but ensuring a level of accuracy that can be explained for the intended use.

Verification Item 2: Align the concepts of reference points and coordinates

Another important point is whether the concepts of reference points, coordinate systems, and alignment have been clearly defined. In LiDAR surveys of cultural properties, even if the geometry looks clean, ambiguous handling of coordinates can cause major problems later. In particular, when overlaying with site-wide plans, comparing measurements from different times, merging point clouds acquired in separate parts, or integrating with external survey results or photographic outputs, it is extremely important to know which reference was used to determine positions.

A common failure is that, although things are consistent relative to each other on-site, they shift as soon as they are placed into the overall coordinate system. This tends to happen when the placement of control points is unstable, observation conditions are insufficient, or the handover of references between measurements taken on different days is unclear. For cultural properties, conservation requirements often prevent freely placing markers, so securing control points itself can become a difficult problem. For that reason, it is necessary to carefully consider what will be regarded as an immovable reference, how temporary references will be managed, and whether records can be created that allow later reproduction.

Also, in cultural heritage surveys, situations where absolute coordinates are essential and situations where internal consistency is prioritized coexist. When dealing with the layout of large-scale remains and their relationship to the terrain, consistency with external coordinates is important, but for indoor or limited-design areas, priority may be given first to the stable reproduction of local shapes. However, if this distinction is not made explicit, misunderstandings can arise after delivery—for example, that the results were thought to include georeferenced position data but were actually in local coordinates, or conversely that aligning to absolute coordinates had degraded local accuracy. The handling of coordinates is not only a matter of accuracy but also a matter of how the deliverable is interpreted.

For cultural heritage, the same location is often measured again in the future for long-term comparison. If the coordinate reference used this time is ambiguous, it becomes impossible to determine whether an observed difference is a real change or simply a positional shift. This ambiguity is fatal when you want to track subtle differences such as crack progression or displacement of components. Therefore it is essential to record the positions of reference points, the observation date and time, the type of coordinates, the integration method, the relationship to known points, and the procedures to reproduce the measurements. By preserving not only the resulting data but also the assumptions under which you judged that the shape was in that position, the reliability of the documentation as a cultural heritage record is increased.

Furthermore, for cultural properties with extensive grounds, there are cases where stabilizing the overall position using LiDAR alone is difficult. In such situations, it is effective to supplement with high‑precision positional references using other methods and to tie the geometry acquired by LiDAR to reliable coordinates. Treating shape acquisition and securing positional references as separate tasks makes it easier to stabilize the accuracy of the entire measurement.

Checklist Item 3 Do not overlook the material of the object and the on-site environment

The accuracy of LiDAR measurements is not determined solely by the shape of the object. Materials and conservation environments specific to cultural heritage can greatly affect the quality of point cloud acquisition. If these factors are overlooked, the actual data quality can fall far below the theoretical accuracy. What practitioners in charge should check in advance is how the object's surface is likely to respond and what kinds of disturbances may occur on site.

For example, weathered stone may appear to be a stable surface at first glance, but it can have fine irregularities and chips that cause acquisition density to vary with angle. Wood can reflect differently depending on the part, and darker areas or the condition of old paint films can lead to differences in how the shape is captured. Metal parts are highly affected by reflectance characteristics, and strong reflections can cause noise. Wet wall surfaces and moisture‑laden stone faces can be more difficult to capture stably compared with when they are dry. Components such as glass or those that are nearly transparent, extremely dark surfaces, and mirror‑like faces are often problematic for LiDAR alone, so it is safer to assume the use of complementary methods.

Environmental conditions are also important. For outdoor cultural properties, the intensity of direct sunlight, the way shadows fall, wind, dampness after rain, surrounding vegetation, visitor circulation routes, and the presence or absence of temporary scaffolding all affect the quality of data acquisition. For indoor cultural properties, narrow passageways, ceiling height, lighting conditions, restrictions on access time, and the presence of exhibits that cannot be moved influence measurement planning. Cultural properties cannot always be approached freely like subjects at general measurement sites, and limited standing positions tend to bias observation angles. As a result, unseen backs, recesses, overlapping parts, and upper decorations are prone to missing measurements.

Also, for cultural properties, considerations for preservation can make it difficult to extend measurement times. You may need to finish quickly during opening hours, the attendance of curatorial or management staff may be required so flexibility is limited, and the installation of temporary fixtures may be restricted; for these reasons, planning has a more direct impact on results than at ordinary sites. That's why, rather than making decisions after arriving on site, it's important during the preliminary survey to identify potential blind spots, materials likely to be affected by reflections or shielding, and how conditions change with the time of day.

When it comes to ensuring accuracy, people tend to imagine high-performance instruments, but in LiDAR surveying of cultural heritage, setting conditions appropriate to the subject and the environment often has a greater effect on the results. Even for the same subject, the stability of data acquisition can change significantly between dry and wet conditions, sunny and cloudy weather, and quiet and crowded times. Not overlooking on-site conditions and identifying in advance which factors can lead to errors or missing data are fundamental to accuracy control.

Checklist Item 4: Finalize the Measurement Plan and Prevent Oversights

To stabilize accuracy in LiDAR surveys of cultural heritage, a very important factor is planning how measurements will be taken on site. By “planning” I do not mean simply how many units to bring or how many hours it will take. It refers to the design that includes from which positions, in what order, at what density to aim, which parts to acquire redundantly, and which locations will serve as on-site verification points. If the measurement plan is inadequate, individual scan results may look fine, but distortions and missing data can become apparent during overall integration, necessitating a return visit.

Especially for cultural properties, there tend to be many unseen parts such as complex reliefs, overlapping pillars and beams, deep carvings, narrow gaps, upper ornamentation, and backside space. These missed areas are difficult to notice by only looking at the screen on site and only become apparent during processing as holes or unnatural joins. To avoid failure, rather than tracing the visible surfaces of the object, you need to anticipate in advance where unseen parts will arise. In other words, acquisition plans should not be devised from the front of the object, but should be worked backward from blind spots.

Also, if there is insufficient overlap between measurement positions, alignment in post-processing becomes unstable. For subjects with repeated similar shapes, such as cultural heritage, everything can look alike locally, so processing tends to become unstable when overlap is small. For example, long corridors, repeating rows of columns, stone masonry, and uniform wall surfaces can produce recurring patterns that may lead to incorrect alignments. For such subjects, simply walking around once is not enough; it is important to ensure overlap from multiple directions while being mindful of positions where feature points are likely to appear.

Moreover, higher measurement density is not always better. There are reasons to increase density where detailed tracking is required, but aiming for excessive density across the entire area not only increases on-site time but also significantly raises the burden of post-processing and storage. In cultural heritage practice, it is more realistic to vary density according to the hierarchical levels of the subject — for example, broad-area recording, high-density capture of decorative elements, and focused capture of areas showing deterioration. Without the mindset of assigning the necessary conditions to the necessary places, the result will be that everything ends up half-measured.

Methods for on-site verification should also be included in the plan. Decide which cross-sections should be checked immediately after acquisition, which important areas require enlarged inspection, and how to determine whether to reacquire data at candidate missing spots—otherwise you may pack up based only on a superficial sense of reassurance. Precisely because cultural properties are difficult to revisit, it is important not to leave on-site verification criteria vague. You must be aware that accuracy is not produced in the processing room but is largely determined at the on-site planning stage.

Checklist item 5: Confirm the workflow of point cloud processing and error verification

LiDAR measurement does not end the moment points are collected on site. A major factor that determines the accuracy of cultural heritage is the post-acquisition point cloud processing and error verification. If this is treated as a black box, the deliverables can be less reliable than their clean appearance suggests. In particular, when integrating multiple surveys, converting to an external coordinate system, or performing noise removal or thinning, the interpretation of shapes can change depending on what processing was applied.

In point clouds of cultural heritage objects, it is often difficult to distinguish between noise and the true boundaries of surface relief. Roughness from weathering, tiny chips, surface degradation, moss and dirt, and deeply shadowed carvings can all be mistaken for noise, and removing them by simple criteria risks losing information that should be preserved. Conversely, leaving too much noise makes errors appear large when interpreting cross-sections or performing difference comparisons. Therefore, in point cloud processing for cultural heritage, rather than mechanically smoothing the data, it is necessary to establish, according to the intended use, consistent criteria for what to keep and what to remove.

Checking alignment accuracy is also essential. Even if the result of merging multiple scans looks natural at first glance, there can be duplication in critical areas, softened edges, or slight steps. These issues are hard to notice in an overall view and are often revealed only by zooming in or inspecting cross-sections. For cultural heritage, those slight steps can be mistaken for deterioration, so local checks of important areas after merging are necessary. Don’t rely solely on the overall error figures; it’s important to see where errors are concentrated.

Also, validation should capture not only the average but also the variability. Even if the average looks good, there may be areas where quality drops, such as shadowed regions or spots with strong reflections. In practical cultural heritage work, if the quality at the locations you most want to inspect is low, the value of the deliverables is greatly reduced. Therefore, validation should not be applied uniformly across the whole object; instead, it should focus on locations that are important for the intended use—critical parts, joints, areas around occlusions or blind spots, and areas showing deterioration. If necessary, you should also compare with on-site photographs and existing drawings to determine whether any unnaturalness in shape is due to processing or is inherent to the object.

Furthermore, if you plan to perform change comparisons or create drawings in later processes, you must also check whether the data are in a state suitable for those processes. Excessive decimation or smoothing may improve visual appearance but can be disadvantageous for change detection. Conversely, leaving the data close to its raw state makes it easier to re-evaluate under different analysis conditions in the future. In cultural heritage documentation, not only the visual clarity of the deliverables created once but also the potential for future reanalysis is part of quality. Whether the point cloud processing workflow and the approach to error verification are shared is a major point in judging the reliability of the results.

Checklist Item 6 Design the deliverable specifications and plan for future operations

Finally, what I want to confirm is whether you have designed what will be left as deliverables and how they will be used in the future. In LiDAR surveys of cultural heritage, the measurement itself can easily become the objective, but what is truly important is how the data are used after acquisition. Point cloud data, mesh data, cross-sectional drawings, plan views, orthorectified images, annotated comparison materials — the required accuracy and processing approach will vary depending on which form you choose to preserve.

For example, if you are going to keep records for long-term archival storage, not only easily visualized outputs but also the raw data and processing history that make reanalysis easy later are important. Conversely, if the main purpose is field presentations or internal agency sharing, readability and lightness should be prioritized. However, in the case of cultural heritage, uses often increase later on, and something intended initially for publicity or documentation may be wanted years later for restoration comparisons or condition assessments. If only outputs that prioritized lightweighting are preserved, the scope for reuse becomes greatly limited. If you want to keep future usability in mind, it is safer to treat viewing copies and archival copies separately.

Also, in recording cultural properties, metadata is as important as the data itself. If it is not clear when, where, over what extent, according to what standards, and under what conditions the data were acquired, it becomes difficult to compare with future measurements or to explain to third parties. In particular, if comparisons over time are intended, information such as the measurement range, the rationale for reference points, processing conditions, the treatment of removed noise, and the acquisition density of important parts needs to be preserved so that a future person in charge can reproduce them. Because the value of cultural properties endures over the long term, it is important to keep records in a state where their meaning is not lost even when personnel change.

Furthermore, if the deliverable specifications are vague, a gap arises between visual satisfaction and practical usability. The ability to render a 3D visualization is not the same as being able to use it as a basis for cross-sections and dimensions. Even with clean data, if the origin and scale, section locations, coordinate information, and comparison criteria are not organized, the usability of the survey outputs will decline. Therefore, you should make explicit not only the types of deliverables but also the intended use of the outputs, and, if necessary, prepare different versions tailored to each use.

LiDAR surveys of cultural heritage should not be one-off tasks completed on site; they should be situated within the continuum of conservation, research, and maintenance management. To produce deliverables that can serve future repair decisions, longitudinal comparisons, additional investigations, and public use, it is essential to plan for their operational use before measurement. Accuracy should be judged not only by numerical quality but also by whether the records are established in a way that will remain usable over time.

To Stabilize the Accuracy of Cultural Property Documentation

When considering the accuracy of LiDAR surveys of cultural heritage, looking only at the instrument's nominal performance is not enough. Only when you include the purpose of the survey, which coordinate reference will be used, what difficulties the object's materials and the environment present, what plan is in place to prevent missing data, and how the results will be validated and delivered does an accuracy that is usable in practice become clear. Because cultural heritage is difficult to redo and must be considered with future comparisons in mind, the success or failure of a survey is largely determined by the preparation carried out before going to the site.

What's particularly important is not to understand accuracy as a single number. In documenting cultural heritage, positional correctness, reproducibility of form, fidelity of fine detail, low levels of missing data, and preservation of reproducible procedures all come into play. Even if one of these is excellent, the practical value diminishes if others are lacking. Conversely, if you organize the types of accuracy required according to the intended use and adopt measurement plans and verification methods suited to them, you can more easily obtain reliable results while avoiding excessive investment.

Furthermore, at cultural heritage sites, the accuracy of location information—where measurements were taken—cannot be ignored in addition to recording the shape itself. For large historic sites, outdoor remains, or surveys spanning multiple points, linking not only the shape data but also each measurement point, camera position, and inspection point to stable coordinates greatly facilitates later comparison and sharing. In such situations, combining LRTK, a smartphone-mounted GNSS high-precision positioning device, can make it easier to streamline benchmark checks around the heritage and the recording of positions for related locations. By preserving three-dimensional shape with LiDAR and supporting on-site coordinate understanding with LRTK, records of cultural properties become not merely three-dimensional representations but organized spatial information that is easier to utilize in the future. If you truly want to improve the accuracy of cultural heritage surveys, it is important to adopt a perspective that designs shape acquisition and position management together rather than separating them.