What does a heat map show? Four ways to read Ministry of Land, Infrastructure, Transport and Tourism standards

In construction sites and civil engineering work, the term "heat map" is not just about easy-to-see color coding. Especially in the context of the Ministry of Land, Infrastructure, Transport and Tourism (MLIT) as-built management, heat maps are treated as documents to understand the planar differences between design and measured data and to evaluate construction results against specification values. Therefore, reading them based on intuition—red means bad, blue means good—can easily lead to misjudgment, and discrepancies often arise between client-side checks and contractor-side internal reviews. As of the March 2025 revision, MLIT organizes supervision and inspection procedures using three-dimensional measurement techniques for each work type such as earthworks, paving, pavement milling, river dredging, and slope works, and reading is based on work-type-specific specification values and legends rather than a sensory interpretation like thermal images.

MLIT's civil engineering construction management standards also indicate that contractors should measure as-built conditions according to measurement standards, create as-built management charts by comparing design values and measured values, and that each measured value must satisfy the specification values. In other words, a heat map is not a "nicely colored diagram" but part of the as-built management charts that should ultimately be read in relation to specification values. This article organizes common confusions site personnel face when looking at heat maps and explains four practical ways to read MLIT standards.

Table of contents

• Background of why heat maps are used in practice

• Reading 1: Colors indicate differences from the design, not pass/fail

• Reading 2: Read the legend as a percentage relative to the specification value

• Reading 3: Look by component and don’t confuse points excluded from evaluation

• Reading 4: Confirm pass/fail by checking the numerical tables

• Practical tips to avoid misreading MLIT standards on site

• Summary

Background of why heat maps are used in practice

The reason MLIT uses heat maps in as-built management is that practice has expanded from checking representative cross-sections to surface management using three-dimensional measurement. In surface management, as-built evaluation surface data or point cloud data are overlaid with three-dimensional design data, differences at each point are calculated, and the surface is evaluated. MLIT explanatory materials for paving describe surface management as a method that converts data into a prescribed mesh for as-built evaluation, calculates elevation differences or horizontal differences from three-dimensional design data, and compares and judges them against specification values. In other words, a heat map is not a colored completion drawing but a management document that visualizes the results of difference calculations on a plane.

If you look at the diagram without keeping this premise in mind, two misunderstandings often occur on site. One is assuming that the stronger the color, the more certainly it is nonconforming. The other is judging that if the whole area is uniform in color tones, there is no problem. In reality, which surface is targeted, what difference is being examined, and which specification value is used for evaluation are decided first, and the color coding is applied on that basis. MLIT's supervision and inspection procedures also state that for as-built management charts, measurement items, measurement frequency, and whether specification values are satisfied should be confirmed, and that variations should be judged according to the legend of a distribution plot that shows each measured value’s deviation from the design as a ratio to the specification value. The important sequence is “before looking at the color, check what criterion the color was applied against.”

Reading 1: Colors indicate differences from the design, not pass/fail

The first point in reading a heat map is that the colors represent differences between the design surface and measured data, not temperature or danger level. MLIT-related as-built management documents present the idea that evaluation is performed based on the deviation between the three-dimensional design surface and as-built evaluation data at each point—specifically elevation difference or horizontal difference. Put another way, a heat map does not directly color-code “good or bad” for each location; it colors how much a location is above or below the design, or shifted sideways from it. Mistaking this can leave you unable to tell whether the same red indicates excess fill, over-cut, or simply an upper-range legend value.

For example, in surface management for paving or earthworks, the comparison target is the measured surface versus the design surface. The design surface is the reference, and the heat map shows differences from that benchmark. Therefore, when you receive a diagram on site, the first thing to confirm is “which layer, which surface, or which component’s difference does this map show?” Whether it is elevation difference of a lower subbase, horizontal difference of a slope surface, or elevation difference at the top surface, the same color can mean different things. MLIT materials also state that as-built management charts should be produced for each as-built confirmation location, and that areas such as flat areas, top surfaces, and slopes should be handled separately. Skipping the step of distinguishing the target component and only looking at colors is like judging distance on a map using only the scale without checking the legend—dangerous in practice.

More importantly, because heat maps show differences from the design, do not jump from color shading to conclusions about construction stability. Whether the whole area is slightly shifted in one direction, only part is significantly out, or only the edges are disturbed changes the likely causes—construction machine operation, how the reference surface was established, handling of design data, localized finishing deficiencies, etc. Heat maps are an entry point for finding anomalies and trends and a quick way to grasp surface differences from the design. The correct use is to use the colors as a starting point and then investigate causes, not stop at just looking at colors.

Reading 2: Read the legend as a percentage relative to the specification value

The second point is that MLIT-standard heat maps should fundamentally be read not by absolute values alone but as a ratio to the specification value. In the earthworks and paving supervision and inspection guidelines, variations are judged according to the legend of a distribution plot showing each measured value’s deviation from the design as a percentage of the specification value. Heat maps are required to color-code results in a range from −100 percent to +100 percent, explicitly show the legend, and indicate distinct regions around about ±50 percent, around 80 percent, and outside the specification range. Thus, how you read the colors is not only “how many millimeters” but “how close it is to the allowable range for that work type.”

Without this understanding, you may not see why a location appears yellow on one site and green on another. In practice, different work types and measurement items have different specification values, so the same measured difference can occupy different positions on the legend. For example, a 10 mm difference might still be comfortably within tolerance at one location but much closer to the specification limit at another. MLIT’s management standards themselves allow application according to the type, scale, and construction conditions of the work, and where specification values are not defined, they instruct contractors to consult with supervising staff for construction management. Therefore, when reading a heat map, you must first confirm the specification values for the target work type and component, then read the legend—do not reverse the order.

MLIT-related materials also recommend that even for work types where specification values are set only on one sign (positive or negative), the display should ideally present a symmetric specification on the opposite sign as well. This improves visual balance and prevents viewers on site from mistakenly thinking there is a one-sided specification. In other words, the legend is not mere decoration but the rulebook for reading the heat map. Before being swayed by impressions of red or blue, confirm where the legend’s center is and where the ±50 percent region, ±80 percent region, and out-of-specification region are—this is the most important aspect of reading MLIT standards.

Reading 3: Look by component and don’t confuse points excluded from evaluation

The third point is to understand that heat maps are not an overview of the entire site to be glanced at in a single image; they are created by component and may include points that are excluded from evaluation. MLIT-related as-built management documents indicate that as-built management charts should be produced for each as-built confirmation location such as flat areas, tops, and slopes, and handled separately for components with differing specification values. In other words, do not read multiple conditions mixed in one diagram; separate and understand which component was evaluated under which specification. When you feel “only the edge color is strange” or “there is a blank band here,” first confirm whether that location is on the same evaluation surface or belongs to a different component.

Moreover, there are measured points near slope shoulders, slope toes, and cross-sectional change points that are excluded from evaluation. MLIT explanatory materials state that for elevation-difference evaluation, measured points within ±5 cm (±2.0 in) horizontally from cross-sectional change points such as design slope shoulder or slope toe are excluded, and for horizontal-difference evaluation, measured points within ±5 cm (±2.0 in) vertically are excluded. This is because boundary areas where shapes change are prone to influence from point placement and interpolation, making simple comparisons disadvantageous. Therefore, seeing blanks or color breaks near boundaries and immediately concluding measurement omissions or construction defects is premature.

Also, in preparing as-built management documents, it is required to use three-dimensional design data and as-built evaluation data and to ensure a data density of at least one point per 1 square meter (10.8 sq ft). Other documents indicate an approach of acquiring at least one as-built coordinate value per 10 cm (3.9 in) mesh across the entire range from the start to the end of the control section. From this we can see that what matters for heat maps is not visual smoothness but which area and what density were evaluated. Even if colors look uniform and aesthetically pleasing, if the target-area partitioning or density requirements are not met, the reading itself becomes unreliable. Conversely, edges looking a bit rough may be acceptable once you account for excluded evaluation points or component partitioning.

If a terrace or small bench has another structure installed and the earthwork surface is not exposed, management of that bench itself may be omitted. In this way, parts of a diagram lacking color or continuity may be due to operational reasons such as being excluded from evaluation, treated as a different work type, boundary exclusion, or because a structure has already been installed—not solely construction defects. To read a heat map correctly, you must check not only the presence or absence of color but also “why that area was set as the evaluation range” along with drawing conditions.

Reading 4: Confirm pass/fail by checking the numerical tables

The fourth point is that final judgment based on a heat map should not be completed with the colored surface alone; you must always check the numerical tables. MLIT-related as-built management documents indicate that as-built management charts should present mean, maximum, minimum, data count, evaluated area, number of rejected points, and other items in tabular form. In other words, heat maps are for visually grasping variation and bias, while pass/fail is determined together with numerical management items. Looking at the diagram and concluding “it’s mostly green so it should be fine” is like looking at only half the report.

This point is clearly illustrated in MLIT paving explanatory materials, which show examples where even if the average is within specification, the work is nonconforming if the maximum value is outside the specification. That is, a good overall average does not constitute passing if there are large local deviations. Conversely, even if some colors are concerning, the number of rejected points and relationship to specification values may allow management to resolve the issue. MLIT’s civil engineering construction management standards also require that each measured value must satisfy the specification values, so the final check requires reading not only averages but individual measured values, maximums and minimums, and point-count conditions.

In practice, when opening a heat map your eye is first drawn to color bias, but the correct order of checks is the reverse. First confirm the target work type, component, measurement item, and specification values in the report; next check the mean, maximum, minimum, data count, evaluated area, and number of rejected points; and then look at the heat map to see where anomalies are concentrated and whether there are construction biases. Think of the heat map as “the eyes for finding anomalies” and the numerical tables as “the basis for determining pass/fail,” and handling on site will be more stable.

Practical tips to avoid misreading MLIT standards on site

Applying the four points above in practice makes tips for preventing misreading very clear. First, when you open a heat map, confirm which work type, which component, and what difference the map shows. Next, on the assumption that the legend is created as a ratio to the specification value, check the color coding for the ±50 percent region, ±80 percent region, and out-of-specification ranges. Then check whether slope shoulders, slope toes, boundary areas, excluded ranges, or mixed different work types exist. Finally, determine pass/fail by comparing with the report’s numerical table. Simply unifying this order for internal reviews and site checks will greatly reduce variations in how different personnel read the charts.

Recently, deliverables have been expanded not only to submitting as-built management charts as PDFs but also to datasets that include viewer-equipped 3D data or projection-capable data sets for projecting heat maps on site using digital technology. In pilot supervision and inspection practices that utilize data, AR and other technologies are suggested to confirm as-built conditions directly on site, improving the efficiency of construction management and supervision/inspection. However, even if the display method changes from paper to a 3D viewer or AR, the basics of reading do not change. Read differences from the design in relation to specification values, and confirm the target component and numerical tables together. If this principle holds, judgment will be less susceptible to changes in display medium.

Mastering heat maps on site is not getting used to flashy colors but being able to quickly retrieve the standards behind the colors. When you can mentally overlay the four layers—work-type procedures, target component, excluded-evaluation areas, and numerical tables—your accuracy in checking drawings improves and explanations to clients or internal approvals become easier. MLIT-standard heat maps are documents to be read by standards, not by intuition. Sharing this recognition makes the biggest difference in practice.

Summary

Heat maps are not merely color-enhanced diagrams. In MLIT as-built management contexts, they visualize the planar differences between three-dimensional design data and measured data and are management documents to check variations and deviations in relation to specification values. To read them correctly, do not directly equate colors with pass/fail; read the legend as a ratio to the specification values; avoid confusing component-by-component evaluation and excluded points; and finally, check the numerical tables. Grasping these four points will greatly change how you interpret heat maps.

If you want to make as-built confirmation, simple surveying, and alignment with design easier on site, it is important to make the flow from measurement to confirmation as consistent as possible. LRTK, as an iPhone-mounted GNSS high-precision positioning device, is a means to facilitate a smooth workflow of measuring, recording, and sharing while keeping position on site. In an era when reading heat maps, point clouds, and three-dimensional data becomes important, reducing uncertainty in the initial site measurements greatly affects the accuracy and speed of downstream decisions.