Workflow and checkpoints for creating cross-sections from point clouds

Table of Contents

• Key concepts to understand before creating cross-sections from point clouds

• Things to check during preparation

• What to check when determining cut locations

• What to check when inspecting the point cloud

• What to check when drafting the drawings

• What to check during the final review

• Summary

Key concepts to understand before creating cross-sections from point clouds

The task of creating cross-sections from point clouds is not simply cutting three-dimensional data and replacing it with two-dimensional lines. It involves interpreting site topography and the condition of structures, organizing them into readable cross-sections tailored to the purpose, and presenting them in a form that makes it easy for stakeholders to make decisions. Therefore, the work does not end just because a shape becomes visible during the process; you must proceed while clearly deciding where to cut, how much to adopt as the cross-section, and how much information to retain.

Situations in which cross-sections are produced from point clouds in practice range widely: checking the topography of developed land, understanding the shape of roads and slopes, verifying the as-built condition of excavations and embankments, checking cross-sections of rivers and drainage channels, and整理ing the current state around existing structures. What these have in common is that the cross-section is not the final deliverable but a tool for decision-making and explanation. For example, whether it will be used to compare before and after construction, to check against the design, for internal review, or to explain to the client changes the required presentation and the way accuracy is handled. If you proceed with that purpose unclear from the start, it becomes easy to have rework later, such as lacking sufficient cross-section locations, being unable to read required elevation information, or inadvertently capturing noise as cross-section shape.

Also, because point clouds visually appear to contain a lot of information, they can give the impression that everything is captured correctly, but in reality there are issues to be aware of such as blind spots, occlusion, reflections, variations in measurement density, coordinate discrepancies, and the inclusion of extraneous objects. Cross-sections extract and use only a part of that point cloud, so the quirks of the original data tend to be revealed more strongly. Compared with planar drawings, even slight positional shifts or differences in extraction width can have a large impact on the results. That is why, when creating cross-sections, intermediate checks during the process are more important than the work itself.

In this article, we explain the workflow for creating cross-sectional drawings from point clouds by dividing it into five stages: preparation, determining the cutting position, point cloud inspection, drafting, and final verification. For each stage, we organize what to look at to reduce failures, clearly outlining the points that should be checked in practice. This is useful not only for those who are taking on cross-section creation for the first time, but also for those who want to review work they have been carrying out intuitively.

Things to Check During Preparation

When creating cross-section drawings from point clouds, the first thing to confirm is the premise of which data will be used and what you intend to show as the cross section. If you start work with this unclear, all subsequent decisions in later processes will waver. In practice, when a point cloud is prepared you naturally want to cut sections immediately, but it is important to first organize how the data were generated and what the intended use is.

First, what you need to check is how the point cloud was acquired. Whether it was obtained from the ground, from the air, or while moving changes the types of targets it is good at and the types it struggles with. Even if the top surface is well captured, the sides may be weak; conversely, slopes and walls may be captured while the ground surface can be difficult to read because of vegetation. How you use the point cloud depends on whether the cross-section requires the ground geometry, the surface of structures, or the position of the pavement edge. If you force everything to be completed with a single point cloud here, there is a risk of incorrectly filling in unseen parts.

Next, what you need to check is the coordinate system and the vertical datum. Point clouds may be managed in local/site coordinates or arbitrary coordinates, or they may be aligned to a public coordinate system. If you plan to overlay cross-sections with existing drawings or design data, you must verify not only the planimetric position but also that the vertical datum matches. Even if they appear to overlap in plan, differing elevation references can cause the cross-section assessment to be significantly off. In practice, even when cross-section drawings look plausibly aligned, differences in datum can lead to misinterpretation. During the preparation stage, it is essential to be certain which datum the data is based on.

Additionally, you need to check the extent and density of the point cloud. Verify whether there are enough points at the required section positions, whether the edges have been captured, and whether point density does not thin out around important change points. For cross-sections in particular, it is important to ensure that locations where the shape changes—such as the slope crest and toe, gutter edges, breaks in the shoulder, and corners of structures—are captured accurately. Even if a large-area point cloud has sufficient overall density, the specific section positions you need can be locally sparse. Because cross-sections are local readings, it is important to zoom in and inspect the area around the target section rather than judging by the overall appearance.

In the preparation stage, you should also anticipate the format of the deliverable in advance. Whether you will share cross-sectional drawings on paper or as a PDF, hand them to the next process as CAD data, or use them as part of a report will change the required cleanup of lines and the granularity of annotations. If it is for internal review, some of the roughness derived from the point cloud may still convey the information, but for external explanations you need to organize lines and symbols so the reader will not be confused. By assuming this output from the outset, you can reduce later rework to remake drawing representations.

The point to check here is not whether the point cloud is sufficient in itself, but whether it is sufficient for the intended purpose. Even a large, high-resolution point cloud is useless if the necessary areas are not visible, and conversely, a point cloud whose density is not that high may still be sufficient to determine the cross-sectional shape of the target. In preparation, organizing the acquisition method, coordinates and elevation, coverage and density, and how the deliverables will be used in advance—and clarifying what this cross-sectional drawing is intended to convey before proceeding to the next step—is the most important starting point in practical work.

Things to confirm when determining the cutting position

The biggest source of variation in creating cross-sections is the decision of where to cut—the cutting position. Even when using the same point cloud, simply adopting a different approach to choosing the cut location can greatly change the meaning of the resulting section. It is no exaggeration to say that whether a cross-section will be useful is largely determined at this stage.

The first thing to consider is the purpose of creating cross sections. For example, whether you want to check a road cross section, confirm the gradient of a slope, or verify the depth and width of a waterway will change how you choose the reference line. You need to decide whether to cut perpendicular to the main direction of the subject, cut so the sections pass through points of change, or cut continuously at regular intervals. If you create cross sections at equal intervals while the purpose is unclear, you may miss the locations you want to inspect and end up with many section drawings that are not useful. Conversely, concentrating sections on areas with large changes will produce materials that can be used for decision-making even with a small number of sections.

When deciding the cutting position, it is important to check not only the location in the plan view but also the direction in which you will cut. Even if the section line is only slightly oblique to the object's orientation, the apparent width and slope in the section can change. Especially for objects with a clear directionality—such as roads, waterways, retaining walls, and long slopes—the relationship between the reference line and the section line must be made explicit. In practice, something that looks fine on the plan can, when shown as a section, appear stretched on one side or unintentionally include points at the far edge. This is often caused by the angle setting of the cutting line.

Also, the way you think about cutting intervals is important. Cutting cross sections at regular intervals is easy to manage, but it can miss points of change. In stretches where the terrain or structure changes little, wider intervals are acceptable, but changes at the toe of slope, break points, connection joints, and near construction boundaries should, when necessary, have additional cross sections provided so that the actual conditions are easier to grasp. Having more section drawings is not inherently better; it is important to arrange them so that necessary changes are not overlooked. Because adding sections later makes organization complicated, identifying important locations during the plan-view stage first will stabilize the work.

You also need to decide here how to treat the cutting width. Ideally, a cross-section is the result of cutting along a single line, but in practice you extract points from a range that has some thickness to represent the section. If this width is too large, points at different depths mix and the cross-section becomes blurred. Conversely, if it is too narrow, there will be too few points and the shape will be incomplete. The thickness to adopt should be determined according to point cloud density, the shape of the object, and the required accuracy. If you mechanically fix this as a constant, it may be just right for one section but insufficient or excessive for another. In practice, it is safer to make a provisional setting on a representative section, check how the shape appears, and then apply it to the whole.

At the stage of determining cut locations, organizing section names and how measurement points are labeled will make downstream work easier. If it's not immediately clear which position a section refers to, it will take time to match it back to its original location after drafting the drawings. In projects that deal with multiple sections, it's essential to ensure the section lines on the plan correspond reliably to the names on the section drawings. If this is ambiguous, it becomes unclear which section is being viewed during explanations, undermining the credibility of the materials you've prepared.

What should be practically checked at this stage is whether the section location is representative of the object, whether the section orientation is appropriate, whether the cut width suits the purpose of understanding the shape, and whether the correspondence with the plan view is clear. Section drawings tend to be judged by how they look after creation, but fundamentally the validity of the chosen cutting positions is what matters most. If you decide these carefully, subsequent point-cloud checks and drafting will become significantly easier.

What to check during point cloud verification

Once the cut position is decided, the next step is to check the condition of the point cloud that will be extracted there. This is not merely about whether the section is displayed; you need to assess whether that section is reliable. The quality of the sectional drawing depends heavily on how the original point cloud was acquired, so it is risky to skip this step and proceed to drafting.

The first thing to check is the inclusion of unwanted points. Point clouds from the site may contain people, vehicles, heavy equipment, temporary materials, vegetation, and disturbed points in shadowed areas. Even if these are not conspicuous in plan views or 3D displays, some unwanted points can appear to protrude in cross-sections and be mistaken for the actual shape of terrain or structures. Especially in cross-sections of ground surfaces and slopes, the presence of vegetation or floating points can make the surface look rougher than it actually is and lead to incorrect judgments about slope conditions or excavation depth. If you see a sequence of points that feels off when viewing a cross-section, rather than simply connecting the line as is, you should return to the original three-dimensional positions and check the cause.

The next important point is whether there are any missing data. Point clouds are not omnipotent, and points do not appear in places that cannot be seen. The bottom of waterways, the backs of structures, deep shadows, and surfaces that are difficult to reflect can all have locally missing points. When converted to a cross-sectional view, if part of a shape appears to be interrupted, you need to determine whether that is an actual missing piece or simply unacquired data. If you interpolate with lines without checking this, you may end up drawing a shape that does not exist. Conversely, if you smooth over areas that should be treated as missing, the cross-section becomes a document that only looks neat.

Variations in point locations are also an important item to check. Even on the same target surface, the surface can appear to have thickness depending on the measurement conditions. When an area that should be smooth, such as a pavement surface, appears spread out in bands, whether to simply take the average position or to consider a representative line that avoids the upper and lower edges must be decided according to the intended use. Because a cross section is represented by a single line, the operator must understand which points that line is representing. The position of the line will change depending on whether the center of the shape is taken or a boundary appropriate for design verification is chosen.

Furthermore, it is necessary to reassess how sections appear relative to the extraction width. Even if the width set in the previous process is appropriate, points from deeper structures or adjacent surfaces can become mixed in depending on the location. If a section appears doubled, crease points increase unnaturally, or a flat surface appears thick, the width should be readjusted. When creating section drawings, adjusting the extraction parameters while checking the appearance on representative sections, rather than applying the parameters set once to all sections, will ultimately lead to more consistent quality.

One thing that is easy to overlook here is the alignment between the plan position and elevation. Even if the cross-section itself looks clean, if its position on the plan deviates from what was intended, it loses its meaning as a cross-section. Especially when comparing with existing drawings or design lines, you must check whether the section line passes through the intended location and whether there is any discrepancy between the point-cloud reference and the drawing reference. Don’t be reassured by the shape of the section alone; it is important to perform back-and-forth checks with the plan.

When inspecting point clouds in practical work, it is important not to accept what you see at face value, but to consider why it appears that way. Being able to distinguish whether something is an unwanted point, the actual shape, missing data, or an effect of extraction conditions greatly increases the reliability of cross-sectional drawings. If this process is performed carefully, you can reduce forced corrections during subsequent drafting and be able to present cross sections with a solid justification during explanations.

Things to confirm when creating drawings

After the point cloud inspection is complete, you organize the cross-sections into a form that can be read as drawings. In this process, rather than simply laying out the point cloud data as-is, you retain only the necessary information and translate it into section drawings that will not confuse the reader. It may look like a task to tidy up the appearance, but in reality it is also a process of reselecting information.

First, what needs to be confirmed is which feature to adopt as the line. Cross-sections obtained from point clouds include fine irregularities and variations, but if you convert all of them into lines the drawings become difficult to read. On the other hand, over-simplifying will lose the original shape variations. The important thing is to organize the data in a way that conveys the meaning of the subject. For example, for a ground cross-section it is preferable that shape changes such as the slope shoulder, slope toe, bottom, and crest are readable, rather than preserving minute roughness as-is. For a structure cross-section, you need to choose lines that convey corners, thickness, and connection relationships. In other words, instead of drawing a single line directly from the point cloud, it is important to consider which shape should serve as the representative line when creating the drawing.

Next to check are the scale and aspect ratio. In cross-sectional drawings the vertical dimension is sometimes exaggerated to emphasize height, but this can significantly change how slopes and irregularities appear. To avoid misleading the reader, be conscious of the ratio used and choose a presentation appropriate to the purpose. What is appropriate may differ between wanting to make changes easy to see during internal review and wanting to accurately convey conditions externally. When producing drawings, maintaining a balance between readability and accuracy is essential.

Reference information attached to cross-section drawings is also important. If it is not clear where the section is located, which direction is being viewed, or which reference elevation it is based on, the shape of the section alone cannot be judged. Ensure the relationship between the plan view and the cross-section is clear, display section names and station points consistently, and, where necessary, organize key dimensions and reference lines. However, packing too much information makes the drawing harder to read, so you need to prioritize the organization according to the intended use. Creating drawings is not about increasing the amount of information, but about arranging the necessary information in a way that can be communicated.

Also, the continuity of the section line should be checked. In sections derived from point clouds, there may be areas where points are sparse or missing, and connecting the lines as-is can produce unnatural kinks or jumps. In such cases, a judgment is required on how far to correct. It is effective to tidy parts that are clearly distorted by unnecessary points, but you should avoid connecting parts that were not actually captured without justification. Prioritizing the appearance of the drawing too much will undermine the reliability of the section drawing. Always remember that the整理 is based on the original data, and it is important not to confuse correction with estimation.

At the drafting stage, you should also check for consistency among multiple cross-sections. If the way each section is represented or the way reference lines are taken differs, comparison becomes difficult. When line types, text placement, handling of reference lines, or the naming of sections vary, readers can become confused about how to read the drawings before they even get to the content. This is especially true when dealing with consecutive cross-sections; by standardizing the rules of representation, it becomes easier for readers to focus on differences in shape.

In practice, the key point to check is not to simply trace cross-sections derived from point clouds, but to always be aware of what the cross-section is intended to convey. By carefully organizing the selection of representative lines, the granularity of representation, how reference information is attached, and the consistency among multiple cross-sections, the cross-section drawings become deliverables that can explain site conditions rather than merely drawing outputs.

Items to confirm during final verification

Once the cross-sectional drawings have been put together, the last thing to do is to check whether they hold up as drawings and whether they can be used for their intended purpose. It is important here to review them from the reader’s perspective, not the drafter’s. A cross-section may be easy for its creator to understand, but its intent might not be conveyed to a third party. In the final check, aim to minimize that discrepancy as much as possible.

First, check the correspondence with the plan. Even if a cross-section looks well formed on its own, it becomes a difficult-to-use document if it’s unclear where on the plan the cut was made. Verify there are no discrepancies in the position of the section line, the section name, the direction indicator, and the handling of survey points, and check consistency by moving back and forth between the plan and the cross-section. In projects where multiple people are working, the mapping between the cross-section and the source data can shift partway through, so be sure to review the entire set at the end.

Next, what we want to verify is the validity of the cross-sectional shape. Even if the line looks clean, we check whether any unreasonable corrections have been introduced compared with the original point cloud or whether important shape changes have been omitted. We confirm that key locations used for assessment—such as slope breaks, gutter shoulders, changes in the road shoulder, and corners of structures—are correctly represented. In practice, it is common during the process of tidying up the appearance to remove even necessary features. In the final review, it is important to prioritize whether the meaning is conveyed correctly over whether it merely looks neat.

This is also the point where you want to detect any oddities in heights and dimensions. Because cross-sections are local views, even a small shift in the reference can significantly change the results. When comparing with existing drawings or design values, it is useful to check whether the heights and widths on the cross-section are within a reasonable range and whether there are any unnatural jumps compared with other cross-sections. If a single cross-section is extremely different, you should reconfirm whether it reflects an actual change in terrain or whether it is caused by the cut position or extraction conditions. Viewing all cross-sections side by side makes it easier to spot inconsistencies you might not have noticed when looking at them individually.

A final check of the presentation is essential. Confirm that text does not overlap, that reference lines are legible, that the handling of scale and aspect ratio is reasonable, and that no extraneous information remains. When using the material as explanatory documentation, consider where the reader will look first and organize the priority of section names, primary lines, reference elevations, and annotations. Abbreviations and conventions that are understood among on-site staff may not be easily conveyed to other departments or external parties. Therefore, in the final check it is better to prioritize clarity from the reader’s perspective over the convenience of the workers.

Finally, it is important to organize deliverables so that you can return to the original data. If only the cross-section drawings remain as standalone items, you may not be able to trace under what conditions they were created when corrections or additional requests arise later. If it is clear which point cloud they were based on, where the cut was made, and which reference or standard was adopted, creating additional cross-sections and responding to inquiries becomes easier. Cross-section drawings are not something you make once and finish; because they are often used later for comparisons and updates, organizing them with an awareness of reproducibility is important in practice.

The purpose of the final check is not merely to find errors. It is to determine whether this cross-section correctly conveys the on-site conditions and can be used for decision-making. Only after reviewing consistency with the plan view, the plausibility of the shapes, any incongruities in heights and dimensions, the clarity of the representation, and the assurance of reproducibility will a cross-section created from point clouds become a deliverable that can be used in practice.

Summary

The workflow for creating cross-sectional drawings from point clouds proceeds in the order of preparation, determination of cutting position, point cloud verification, drafting, and final review, but what actually affects quality is the checks performed at each stage. In preparation, organizing the point cloud acquisition conditions, coordinate reference, and intended use, and clarifying what the cross-section is intended to show, is the starting point. In determining the cutting position, it is important to choose the position and orientation that appropriately represent the subject, and to consider the extraction width and the placement of the section according to the purpose. In point cloud verification, identify unnecessary points, missing data, variability, and the effects of extraction conditions, and judge whether the shape you see can truly be trusted. In drafting, instead of simply listing the point cloud information, organize it into a cross-section that communicates to the reader. In the final review, recheck consistency with the plan view, the validity of the shape, and the clarity of presentation, and finish the work so it is usable as a deliverable.

A common way to fail at creating cross-sectional drawings is to rush the work and only be careful in the latter half. However, in practice it is the decisions made in the early stages that are harder to correct later. If which point cloud to use, where to cut, or the width for extraction remain ambiguous, even neatly drafted drawings will result in cross-sections with weak justification. Conversely, if the early-stage checks are thorough, a drawing that uses the minimum necessary representation can still effectively convey the results.

As the use of point clouds expands further, creating cross-sections will require not only drafting skills but also the ability to interpret onsite information and organize it appropriately. Producing cross-sections with an awareness of site positioning and connections to measurement results makes it easier to reduce rework and misunderstandings. When considering such workflows, mechanisms that make high-precision positional information easy to handle in the field are also important. For example, having means like LRTK that facilitate linking field position checks and measurements with data utilization makes it easier to organize the whole process before and after generating cross-sections from point clouds, leading to improvements in overall accuracy and efficiency in practice.