How to create a heat map from point clouds? Five steps for MLIT-compliant procedures

When you consider creating a heat map from point clouds, the first thing to grasp is that being MLIT-compliant is not simply about producing a color-coded, easy-to-read diagram. Under the current MLIT arrangements, when using 3D data for as-built management in ICT construction, it is premised that you follow the "Guidelines for As-Built Management Using 3D Measurement Technologies (draft)," and the as-built management guidelines were revised in March, R8, while the supervision and inspection guidelines were revised in March, R7 for many types of work. In other words, before worrying about colorful visuals, you need to determine which work categories will use which management methods, at what measurement accuracy, and which deliverables must be prepared.

What the field commonly calls a "heat map" is, in practice, often treated as a kind of as-built management report that visualizes the area difference between the design surface and measured values. However, MLIT’s Q&A clarifies that a heat map is not mandatory as an as-built management report and requires consultation with the client. Moreover, for some work categories, instead of comparing the entire area against 3D design data, dimensional control or cross-section control using point clouds may be the primary approach. Therefore, producing a heat map from point clouds cannot be done mechanically with the same procedure at every site. First determine whether surface-based management is appropriate, and then evaluate the differences in line with MLIT’s approach.

Table of contents

• Understand the basics of point cloud heat maps and MLIT compliance

• Step 1: Clarify management purpose and applicable work categories

• Step 2: Prepare 3D design data and coordinate conditions

• Step 3: Measure point clouds to meet accuracy and density requirements

• Step 4: Calculate differences from the design surface and create the heat map

• Step 5: Compile reports and electronic deliverables

• Common pitfalls and avoidance strategies

• Summary

Understand the basics of point cloud heat maps and MLIT compliance

At first glance, creating a heat map from point clouds looks straightforward: measure the site, process the point cloud, overlay it with design data, calculate the differences, color-code them, and compile the report. However, when MLIT compliance is required, more rigorous preparation is actually necessary. That is because what supervisors and inspectors look at is not “whether the colors are neatly applied,” but “by which standard were the differences evaluated,” “are the source data used to calculate those differences reliable,” and “were the results handled appropriately against the standards for each work category?” MLIT’s supervision and inspection guidelines state that for multi-point measurement management in road earthworks and river earthworks, the elevation difference or horizontal difference between the 3D design surface and the measured values should be checked. In other words, the essence of a heat map is not the colors but the differences from the design surface.

Starting work without understanding this premise leads to common mistakes. For example, you might prioritize aesthetics and set very fine color bands without clarifying which surface the underlying differences reference, or you might forcibly interpolate missing point cloud areas to create a smooth surface despite many missing measurements, or you might use in-house thresholds for red/blue judgments instead of the work-category standard values. Such materials may be useful for internal briefings but are precarious as MLIT-compliant as-built management documents. A heat map is the final visualization result; it only makes sense when the preceding measurement conditions, design conditions, and evaluation conditions are all in place.

Also, even within MLIT compliance, you cannot process all work categories with a single template. The Q&A indicates there are work categories—such as retaining wall-related works—where evaluation is not based on 3D design-data full-surface comparison but rather on dimensional control using point clouds. Further, for natural slopes, MLIT sometimes does not require creating 3D design data at all. Therefore, this article focuses on situations where surface management with point clouds is relatively straightforward—particularly earthworks and development surfaces where the difference from the design surface is easy to present as an area—and, while touching on exceptions for certain work categories, translates the approach into practical decision criteria you can use in the field.

Step 1: Clarify management purpose and applicable work categories

The first step is to be clear about why you are creating the heat map. If this is vague, everything that follows will wobble. The required granularity and the way you design color bands differ depending on whether you’re creating it for as-built management, for checking intermediate construction quality, for prioritizing internal corrective actions, or as supporting material for client explanations. MLIT’s Q&A also states that a heat map is not mandatory and requires consultation with the client. In short, if you don’t decide the role of the deliverable in advance, it will easily become unclear whether what you produce is “reference material” or “as-built management documentation.” It’s important to organize whether it will be used as an official management document or as an explanatory aid during project meetings.

Next, confirm whether the target work category is actually suitable for area-based comparison. For work types like road earthworks and river earthworks, where it is meaningful to capture differences between the design surface and the actual surface as an area, heat maps are highly effective. On the other hand, for some work categories it is more appropriate to evaluate using conventional dimensional controls—width, thickness, length, cross-sectional shape—often combined with other checks. The fact that some slope works are not premised on full-surface comparison using 3D design data is a typical example. As a field engineer, you should not immediately convert point clouds into heat maps simply because you obtained them; first determine whether the work’s as-built management is surface-based, cross-section-based, or dimensional. Getting this wrong risks producing difference outputs that cannot be used in inspections or briefings.

Also decide at the outset whether the differences you want to evaluate are elevation differences or horizontal differences. The supervision and inspection guidelines state that in multi-point measurement management for road earthworks and river earthworks, you should confirm either the elevation difference or the horizontal difference between the 3D design surface and the measured values. This implies that a heat map must be a difference map corresponding to the management item you intend to verify, not just a vaguely color-shaded elevation map. For development surfaces, elevation differences tend to be central, but if you want to check slope toes, edge fit, positional shifts, or alignment interactions, horizontal differences may be more useful. Deciding which difference to output at the start clarifies the later data processing considerably.

In practice, it is recommended to think of management purposes in three categories. First, conformity checking against standard values — this is close to formal as-built management and requires careful threshold settings and report structuring. Second, tracking trends during construction — at this stage, the focus is on whether there is a uniform deviation across the entire surface, localized sinking in parts, or bias at the edges. Third, use as explanatory materials — in this case, color widths and legend design must be easy for non-specialists to understand. Even for the same heat map, the optimal creation method differs depending on what decision it is intended to inform. To stay MLIT-compliant, start with this purpose clarification.

Step 2: Prepare 3D design data and coordinate conditions

The second step is to prepare the 3D design data that will serve as the comparison baseline. Since a heat map colors the difference between the design surface and the measured surface, ambiguous design data will invalidate the comparison. The Kanto Regional Development Bureau’s usage manual organizes 3D design data as outputs of the shapes of project objects specified in the design documents—such as road centerline alignment or normals, as-built cross-section shapes, construction control point information, and the coordinate system to be used—exported as surface data like TIN. In short, simply making a 2D drawing look 3D is insufficient; coordinates, surfaces, and control point information must be included so the data are comparable.

A common mistake here is starting processing before the coordinate conditions of the design drawings and point cloud are matched. If the plane rectangular coordinate system zone number, the handling of elevation, the positions of construction control points, or the presence of site origin shifts do not align, the heat map will appear uniformly shifted. Worse, it often looks like a relatively smooth difference map visually, so detection can be delayed. In actual sites, time spent reprocessing due to coordinate mismatches can exceed time spent on the measurements themselves. Before creating the heat map, always confirm on a separate screen that the point cloud and design data overlap on the same basis, and verify how they match at edges and around control points. This is a low-profile but highly effective check.

Also, not all work categories require the same level of detail in 3D design data. MLIT’s Q&A indicates that for natural slopes with irregularities and curvature, creation of 3D design data may not be mandatory, and when created it may be sufficient to indicate only the direction of the as-built cross-sections. Conversely, when you want to evaluate the difference from the design surface as an area—such as in earthworks and development surfaces—preparing a design surface suitable for comparison is indispensable. Thus, the right perspective is not simply “make or not make” design data, but “prepare the design data to the level required for as-built management of that work category.” When creating heat maps from point clouds, the latter typically applies, so the quality of the design surface directly determines the heat map quality.

MLIT’s supervision and inspection guidelines require a 3D design data check sheet for certain work categories and measurement methods. This reflects the idea that creating design surfaces is not the end; you must confirm whether the design data are in a state usable for design checking, as-built management, and quantity calculation. In practice, upon receiving the design data you should confirm the target range, management items, coordinate system, reflection of design changes, and approach to splitting surfaces, and prepare the data into a form usable on-site before measurement. Sites where heat map outputs collapse in later stages typically lack this preparation. Before producing difference maps, cultivate the habit of questioning the comparison foundation.

Step 3: Measure point clouds to meet accuracy and density requirements

The third step is to acquire point clouds correctly. There is a tendency to think that any captured point cloud can be turned into a heat map, but MLIT’s approach stipulates that measurement accuracy and point cloud density appropriate to each measurement method must be met before they can be used for as-built evaluation. The usage manual states that applying 3D measurement technologies presumes conducting initial surveys and as-built measurements in accordance with measurement accuracy that secures the machine accuracy required by specified standard values and point cloud measurement density, and it provides guidance on required accuracy and density by device type. The important point is not the device type but whether the final point cloud achieves the quality required for the work category and management item.

At this stage, focus not on making the whole surface look uniformly captured but on ensuring the necessary accuracy and density across the evaluation target. MLIT’s 2025 Q&A assumes measurements meet the specified point cloud density across the construction area, while allowing that locations where measurement or evaluation is inherently difficult can be excluded from the evaluation, and in such cases missing-data interpolation using prescribed measurement technologies is permissible. The key is the order: “as a principle, secure the required density, and only then treat genuinely difficult areas under the guidelines,” not “leave gaps and fill them later at will.” If you plan measurements assuming missing data, the heat map’s persuasiveness drops sharply.

Be cautious about careless interpolation. The Kinki Regional Development Bureau’s Q&A addresses a case where mobile-device slope measurements missed the central area, stating that interpolation points in the central area were not accepted and that, if you want to perform surface management, you should carry out additional interpolation measurements using other 3D measurement technologies. This means that simply filling missing areas by software because it looks natural is not necessarily acceptable for as-built management. Therefore, do not use “interpolation” as a single catch-all term; distinguish whether it is additional on-site measurement, guideline-authorized missing-data interpolation, or mere cosmetic correction. If you are aiming for MLIT compliance, treat interpolation as a controlled management action under specified conditions, not as a last-minute fix.

Also consider ease of post-processing when measuring. Excessive inclusion of objects like vegetation, heavy machinery, temporary materials, or people increases time for removing unwanted points and makes it harder to extract management surfaces. Conversely, overzealous removal of inconvenient points after the fact can distort the original shape. In practice, identify before measurement the areas to evaluate, potential blind spots, and areas like edges and slope toes where differences tend to appear, and revise measurement routes or flight plans if necessary—this usually saves time in the end. Although heat maps appear to be completed in post-processing software, most of their quality is actually determined at the site measurement stage. Color schemes can be changed later, but missing point clouds or misaligned coordinates are not easily recoverable.

Step 4: Calculate differences from the design surface and create the heat map

The fourth step is to compare the acquired point cloud with the design surface, calculate the differences, and render them as a heat map. The core here is not simply coloring the point cloud itself, but creating the evaluation surfaces or evaluation points from the point cloud, defining the differences relative to the design surface, and then visualizing them. MLIT’s supervision and inspection guidelines also specify that the confirmation content is “the elevation difference or horizontal difference between the 3D design surface and the measured values,” and the axis of evaluation is always the difference from the design. Therefore, heat map creation is a phase in which, before choosing colors, you decide which surfaces to compare, how to extract evaluation points from the point cloud, and how to aggregate differences.

In practice, first extract only the management target surface, then decide whether to compare the point cloud directly or to convert it into an evaluation surface such as a TIN for comparison. For relatively smooth surfaces like development areas, it is easy to establish evaluation points at regular intervals or densities and compute elevation differences against the design surface. Conversely, change points such as edges, break points, slope toes, and slope heels can be obscured by averaging, so treating them separately can make management easier. Heat maps are good for viewing overall trends, but localized changes are sometimes better communicated with cross-sections or enlarged views. For MLIT compliance, do not try to conclude everything with a heat map alone; combine it with cross-section checks and dimensional checks as needed. This is a practical operation that centers on area-based management while fulfilling explanatory responsibilities.

Color scheme design is actually important. For example, assigning green around zero, warm colors to raised areas, and cool colors to excavated areas makes the map intuitive. However, making color bands too fine can emphasize small fluctuations within the standard values and mislead field judgment; conversely, bands that are too coarse can miss local anomalies. A recommended approach is to use two variants: one that displays standard-value-based classifications for inspections and explanations, and another finer display for trend analysis during construction. Color appearance can be changed later, but for MLIT compliance you must clearly indicate in the legend which thresholds were used for color-coding. Even if difference calculation methods are the same, ambiguous legends lead to divergent interpretations.

Another important point is not to treat a heat map as an assurance of everything. An area-average might indicate no problem while large deviations are concentrated at the edges. Particularly for development surfaces and slopes, differences often concentrate at the start of water slopes, transition zones, and interfaces with structures. While heat maps are powerful for showing the overall picture at a glance, they can also fall into the averaging trap. Therefore, after looking at the overall color distribution, analyze the positions of extremes, bands where abnormal colors continue, construction lot boundaries, and relations to machine travel direction. For MLIT-compliant documentation, treat the heat map not as a “pretty image” but as an analytical diagram that explains where, to what extent, and with what bias differences occurred.

Step 5: Compile reports and electronic deliverables

The fifth step is to compile the created heat maps into a form that constitutes as-built management documentation. The important point here is not to consider the heat map alone as the deliverable. The supervision and inspection guidelines list that electronic deliverables should include as-built measurement data, 3D design data, measured point cloud data, construction control points, and, where applicable, benchmark point data and as-built management documents. In other words, a heat map is only part of the as-built management materials, and it only makes sense when supported by the underlying data and reference information. Submitting only a color-coded diagram without the measurement and design conditions that underpin it lacks explanatory power and reproducibility.

The guidelines also state that electronic deliverables should be stored in the ICON folder specified by the "Guidelines for Electronic Delivery of Construction Completion Documents," so you must be aware not only of the data contents but also of the delivery format. This means you cannot simply submit an internally produced image file. A common on-site mistake is saving only the heat map image in a separate folder and losing the linkage to the original point cloud and design data; then it cannot be traced later. On MLIT-compliant sites, organizing file naming, version control, and recording which design data were used as the basis is practically very important.

When compiling reports, at a minimum include the work category, target range, measurement method used, measurement date, control point information, design data version, evaluation items, color legend, relation to standard values, and an explanation of areas where differences concentrate. Additionally, include cross-sections, enlarged views, quantity calculation results, and photo correspondences as needed to facilitate understanding on-site and during inspection. MLIT’s supervision and inspection guidelines emphasize confirmation of the 3D design data check sheet and the accuracy verification test report, so aim for a state where not only the person who made the heat map but also the measurement personnel, design data preparer, and construction management staff can share the same documentation structure. As-built management is not about making a single diagram but about creating traceable evidence.

Operationally within the organization, it is effective to keep a lightweight heat map for on-site correction separate from the formal submission report. The formal version should have strict legends and standards and be used for explanations; the lightweight version should be used frequently to check trends during regular as-built confirmations. Mixing the two can lead to unnecessary embellishment in the formal version or an on-site version that is too heavy to use. Designing submission materials and routine operations separately makes it easier to implement MLIT compliance without slowing on-site workflows.

Common pitfalls and avoidance strategies

The most common failure in creating point cloud heat maps is treating the task as an image creation problem. In reality, most failures occur before the image stage. Accumulated issues such as ambiguous management purposes, unclear applicability by work category, inconsistent coordinate conditions, insufficient point cloud density, sloppy handling of missing data, inadequate version control of design data, and legends not tied to standard values can together reduce the heat map’s persuasiveness. The countermeasure is simple: before coloring, be able to state in one line “what is being compared, what is being shown, and what is the basis for judgment.” A heat map that cannot be described in one line usually fails to satisfy on-site or inspection requirements.

Another common mistake is relying solely on full-surface comparisons for judgments. Even if the area color distribution looks good, differences concentrated at edges or interfaces can be masked. Especially at construction lot boundaries or where machine travel direction changes, characteristic deviations are likely. Therefore, after viewing the overall trend with the heat map, follow up on extremes with cross-sections or detail drawings. MLIT emphasizes checking differences not because overall proximity is sufficient, but to ensure required checks are performed for each management item. On site, treat a good overall average and the absence of issues at critical locations as separate matters.

Handling of interpolation is another typical pitfall. MLIT’s 2025 Q&A assumes that measurements satisfying the specified density across the construction area are the premise, while allowing conditional missing-data interpolation when fundamentally difficult to measure. Meanwhile, the 2023 regional bureau Q&A disallowed interpolation points in the central slope from mobile-device measurements, requiring interpolation measurement using other 3D measurement technologies. Reading these together shows that you should avoid freely filling gaps based on on-site judgment and instead process them according to the guidelines, Q&A, and conditions for each work category. In short, the right question is not whether you can interpolate, but under what conditions interpolation can be justified.

Also, do not assume heat maps are mandatory. MLIT’s Q&A states that heat maps as as-built management reports are not mandatory and require consultation with the client. This means you must determine whether the recipient expects formal as-built management materials or merely reference area-understanding documents. Effectiveness and obligation are different matters; ignoring this distinction risks spending effort on materials that will not be evaluated. Confirm the deliverable’s role before creating it—this small step can save significant cost.

Finally, do not try to complete the heat map perfectly in one go. In practice, the first map is for checking, the second for refining color bands and legends, and the third for final reporting. Rather than aiming for a perfect initial appearance, it is faster and more stable to refine in the order of coordinates, differences, extremes, legend, and submission format. Treat point cloud processing and heat map creation as iterative parts of the overall as-built management workflow; this attitude will ultimately align best with MLIT requirements.

Summary

When organizing the steps to create a heat map from point clouds under MLIT compliance, the flow is clear. First, clarify why you are creating the heat map and whether the target work category is suitable for surface-based management. Next, prepare the 3D design data and coordinate conditions that form the comparison basis. Then acquire point clouds that meet the required accuracy and density, compute differences from the design surface as elevation or horizontal differences, and finally compile reports and electronic deliverables. The key is to treat heat maps not as images but as documentation that makes the basis of as-built management traceable.

On site, while you want to speed up as-built confirmations, including checks of measurement conditions, control points, coordinate management, and photo correspondence can make the task heavy. That is why aligning comparison criteria, measurement quality, and report roles with MLIT’s approach from the start is effective. Instead of worrying after taking point clouds, reverse-engineer the final heat map before measurement—this is the shortest practical route.

If you want to minimize the effort from control point confirmation to as-built verification, it is also essential to consider ways to simplify the measurement itself. LRTK, as a smartphone-mounted high-precision GNSS positioning device, is strong in enabling quick, easy positional capture on site. The effectiveness of point clouds and heat maps is not determined solely by post-processing performance, but significantly by how reliably you can collect correct position information on site without hesitation. Those who wish to run MLIT-compliant as-built management more pragmatically may find it beneficial to review the entire site workflow, including simple surveys using LRTK.