Orthoimage Creation Method to Improve Recording Accuracy of Archaeological Sites | 5 Fail-Safe Steps

In archaeological site investigations, orthophotos are not merely easy-to-view aerial or site photographs. They are fundamental records for preserving the condition of the ground surface and archaeological features that are lost as excavation progresses, with positional information and dimensional accuracy. In archaeological fieldwork, excavation itself is irreversible and the same state often cannot be reproduced for later verification, so the quality of the record directly determines its verifiability and reusability. In recent years, with the spread of photogrammetry, there has been an increase in cases where archaeologists themselves carry out continuous site recording, and orthophotos are positioned as central materials for site plans, feature interpretation, temporal comparison, and GIS integration. Furthermore, images that have only been rectified to a rectangle may look tidy but can be insufficient as measurement materials, so operation based on correct orthorectification and coordinate management is indispensable.

Table of Contents

• Why Orthophotos Are Important for Archaeological Site Recording

• Step 1 Define the recording objectives and deliverables first

• Step 2 Set up reference points and coordinate management

• Step 3 Finalize the shooting plan

• Step 4 Proceed with on-site shooting without disrupting the location

• Step 5 Perform post-generation accuracy verification

• Common mistakes and how to address them

• Summary

Why orthophotos are important in archaeological documentation

The value of orthoimages does not lie in producing attractive images of a site as seen directly from above, but in their ability to preserve records with minimized planimetric displacement. To turn the perimeter of an excavation area, stratigraphic contacts, the continuity of stone rows and trenches, surface color variations, and changes before and after excavation into materials that can be re-examined in later stages, it is important that the images be stable within a coordinate system. On site, details that were understood during the investigation can sometimes not be reconfirmed during the post-processing stage, but with high-quality orthoimages it becomes easier to reconfirm information directly from the images rather than relying solely on drawings and written descriptions. When following the same investigation area over a long period in large-scale excavations, if orthoimages created by a consistent procedure are accumulated, it becomes easier to track differences between stages and changes in interpretation. In practice, it has been reported that a continuous recording method incorporating photogrammetry has been well adapted to archaeological fieldwork and has helped improve documentation.

Orthophotos are not standalone materials; they truly demonstrate their strengths when linked with plan views, cross-sections, three-dimensional models, elevation information, and survey field notes. Tracing archaeological feature outlines, overlaying artifact find locations, integrating records by excavation unit, and later reinterpretations are all possible precisely because the original images are preserved in a measurable format. Even in older studies, correct orthophotos are an important complementary resource for recording features, and mere rectified photographs that are commonly used are generally regarded as unsuitable as measurement data. In other words, when creating orthophotos of archaeological sites, the objective is to produce a measurable record that can be reused later rather than to make a pretty image. Simply adopting this perspective will greatly change how you capture photographs, where you place control points, and how you verify results after processing.

Step 1 Define the recording purpose and deliverables first

The first step to avoid failure is to clarify what you want to record before shooting. On site, the required resolution and capture density vary depending on whether you want a single orthoimage of the entire survey area, high-resolution captures of individual features, images for daily progress comparisons, or figures for inclusion in a report. For example, the level of image detail needed to monitor the overall progress of an excavation area differs from that needed to read subtle soil color variations at the edges of features. If you start shooting while leaving this unclear, you are likely to end up with situations such as “the overall view is visible but the details cannot be read” or “the details are fine but the overall context is not connected.” Photogrammetry research in archaeology likewise emphasizes that preplanning based on the deliverable specifications and field conditions is important, and that field conditions and required standards determine the choice of imaging methods.

What matters here is not the size of the target but working backwards from the smallest unit you ultimately need to resolve. Whether you want to confirm only the outline of features within the excavation area, the boundaries of rammed-earth or burned soil, or the placement of stones and small irregularities will change the required ground resolution. Therefore, before deciding on flight altitude or capture distance, you need to determine what level of visibility is practically sufficient. At sites where orthophoto production fails, it is often the case that this specification of deliverables is missing even before image-processing settings are addressed. In particular, when operations involve multiple people, failing to share the same standards for each survey causes perceived accuracy and framing to vary day by day, reducing the value of the results as comparative material. Even in cases where archaeologists were able to independently continue photogrammetric recording, the standardization of procedures and consistency of operations proved to be highly significant.

Step 2: Set up reference points and coordinate management

The second step is to organize the coordinate framework first, rather than the visual appearance of the image. One of the major factors that determines the accuracy of an orthophoto is the number and arrangement of control points. Generally, placing control points only along the periphery tends to stabilize the planimetric direction, but to improve overall accuracy including elevation, it is important to place control points inside the survey area as well, without bias. Research also shows that, in addition to perimeter placement, distributing control points in layers toward the interior is effective in reducing overall error, and that the consideration of control point placement itself is central to ensuring accuracy. In other words, it is not enough to simply increase the number of points; how they are dispersed across the area is what matters.

At archaeological sites, as excavation progresses the ground surface changes, so temporary markers can disappear midway. Therefore, reference points should be chosen so they are unlikely to be lost during the investigation, are easily visible in images, and can be connected to other surveying results. It is possible to operate with a closed, local marker management limited to the excavation area, but if future reuse or connection with surveys in other years is anticipated, it is advantageous to link them to external coordinates and existing survey results whenever possible. Practical studies in close-range photogrammetry have also shown that relating reference points measured on site to other surveying results is effective for later-stage alignment. To make survey results cumulative data rather than records limited to a single year, these coordinate linkages cannot be neglected.

Furthermore, it is important to adopt the idea of providing points for accuracy checking that are separate from the reference points. If you judge that “the errors are small” by looking only at the points used in the processing, the model may simply be drawn toward those points. In photogrammetric accuracy assessment, not only the residuals after bundle adjustment but also comparisons with independent check points and outlier detection are considered important. The same applies to archaeological recording: by separating the points used as references from those used later for verification, you can evaluate the deliverables more objectively. Even if the result looks natural, distortions such as a slight stretching of part of a plane or a bulging of the central area can readily occur, so a system for evaluating processing results that does not rely on subjective judgment is necessary.

Step 3 Finalize the shooting plan

The third step is to work out the capture plan in detail before the actual photography. For creating orthophotos for archaeology, the basics are to ensure uniform coverage of the entire subject, secure sufficient overlap, and avoid creating blind spots. In practical examples of orthophoto generation from actual archaeological surveys, cases have been shown where about 80% overlap was planned in both the forward-backward and lateral directions, and securing overlap is a prerequisite for stable processing. If overlap is insufficient, the number of matching points between images decreases, connections become weak, and local distortions or gaps are more likely to occur. Conversely, once the necessary overlap is secured, shooting with a little margin beyond the target area makes post-processing trimming and accuracy checks easier.

However, photography from directly overhead is not always sufficient. When there are changes in elevation, deep cuttings, stonework, or wall-like exposed surfaces within an excavation area, there will inevitably be faces that cannot be seen in overhead images alone. Close-range photogrammetry research has shown that undulations and irregularities cause occlusion, increasing the editing workload and, under poor conditions, sometimes making adequate processing difficult. Other studies have also reported that in complex scenes, combining oblique images with overhead images reduces occlusion and blind spots and leads to more reliable reconstructions. The greater the height differences in archaeological features at a site, the more effective it is to secure planarity with overhead images while supplementing rising or vertical parts with oblique images as needed.

Assessing lighting conditions is also an important element of the image acquisition plan. At outdoor archaeological sites, strong shadows and reflections from direct sunlight and changes in the direction of light with time of day affect consistency between images. Studies addressing the effects of lighting conditions show that, while avoiding particularly unfavorable lighting keeps accuracy degradation within a certain range, deep shadows, uneven illumination, and low light levels tend to have adverse effects on image quality and reprojection error. Therefore, the time of capture should be chosen not as "a time when work is possible" but as "a time when shadows are least likely to be erratic." For sites where you want to read fine surface details, there are situations in which emphasizing micro-topography is desirable, but for basic capture intended for orthophoto production, prioritizing stable lighting over extreme contrast will reduce failures.

Step 4 Carry out on-site shooting without causing disruption

The fourth step is not to shoot strictly according to plan, but to complete the photography without disruption while taking changes at the site into account. An excavation site is not a stationary subject. The movements of workers, the placement of tools, excavated soil, protective coverings, footprints, puddles, and so on can change conditions in a short time. Field studies of close-range photogrammetry also note that although archaeological data acquisition tends to shorten the time spent on site, because excavation is a continuous and dynamic process it is important to reliably meet the necessary conditions within a short measurement time. When photographing, it is essential to organize the area so that moving objects other than the target do not remain in the images and to complete the series of shots before conditions change between images.

In practice, simply pausing activity in the excavation area before photographing to perform surface cleaning, confirm markers, remove unwanted items, and make a final check of the shooting area greatly increases the success rate. Loose soil piled around stones and artifacts, reflections from wet areas, puddles, and sheets moved by wind can cause noise or erroneous responses during the processing stage. Because the condition of the excavation surface changes daily, standardizing the pre-shoot checklist for each session is effective. In long-term, large-scale excavation operations, the reason archaeologists themselves were able to continue photogrammetry was the clarification of the workflow and its integration into field operations. Standardizing pre-photography procedures, rather than relying on the individual intuition of the person in charge, leads to more stable final accuracy.

Also, on site you need to focus not on whether a photo was merely taken, but on whether it can be processed later. It is important to check a few images on the spot to confirm that reference points are properly captured, there is no motion blur, the field of view is not clipped, and sufficient overlap is ensured. Edges of features and the corners of the excavation area are especially prone to being missed, and it is not uncommon to discover deficiencies only after processing. Even earlier studies of close-range photogrammetry report that when shooting conditions are poor or terrain is highly undulating, manual corrections increase and some areas cannot be fully recovered. Spending a few minutes to check on site reduces the risk of problems that cannot be redone later. Because excavation is an irreversible process, immediate on-site checks are crucial.

Step 5: Perform post-generation accuracy verification

The fifth step is not to stop at generating the orthophoto but to always perform accuracy verification. In the field, people tend to judge a job as "well done" if the images are sharp and the seams are not noticeable, but from the standpoint of improving recording accuracy that alone is insufficient. In photogrammetry research, it is standard to assess the quality of deliverables through the root mean square of control point residuals, comparison with independent points, checking for outliers, and so on. For orthophotos of archaeological sites, you should at minimum check the trends of control point residuals, deviations from check points, the presence or absence of local deformations, and the consistency between the edges and the center. If residuals are large only in specific locations, you need to isolate causes such as insufficient image overlap, misreading of control points, or the inclusion of ground-surface changes.

In evaluation, you should look not only at numerical values but also at practical usability for interpretation. For example, check whether feature lines are unnaturally wavy, whether linear structures appear bent, whether there is shrinkage or stretching at survey area boundaries, or whether overlays with past datasets feel inconsistent. In archaeology, because materials are ultimately used by people to read and make judgments, it is necessary to consider both visual consistency and measurement consistency. Especially when comparing orthophotos from multiple occasions, even small errors in each capture can make it hard to tell whether differences are real phenomena or simply errors if imaging conditions or control point configurations have different biases. If the data are to be used for ongoing monitoring or process comparisons, operating with the same verification criteria each time is effective. Because reproducibility can be affected even by differences in lighting conditions, keeping a record of the imaging conditions helps stabilize subsequent processes.

Furthermore, precision management should be considered to include how deliverables are stored. Even if you retain only the orthophoto as an image file, its value as documentation drops significantly if, when reviewed later, it is unclear which coordinate system was used, which reference points were used, or which processing stage the surface corresponds to. By managing the survey date, area covered, reference points used, the results of checkpoints, generation conditions, and related drawing numbers or survey-area numbers as a set, you create a record that can be reused. Even in cases where photogrammetry is well established in research settings, a common feature is that it is operated as a standardized recording system rather than as one-off image generation. Improving accuracy means not only reducing the error of the images themselves but also preserving them as a re-readable information system.

Common Mistakes and How to Address Them

One common mistake in creating orthophotos of archaeological sites is not adequately covering the outer perimeter of the capture area, which leads to noticeable distortion at the edges after processing. Shooting exactly to the target may seem efficient, but it weakens image-to-image connectivity and coordinate constraints, compromising the stability of peripheral areas. As a countermeasure, it is effective to include a bit of the area outside the target in the photographs and to ensure that there are control points or surfaces with distinctive features along the outer perimeter. Research on control point placement also shows that positioning around the outer perimeter is important for planar accuracy, and neglecting the edges directly leads to reduced quality.

The second case is when accuracy cannot be achieved despite having reference points in place. The cause is often not a lack of reference points but biased placement, poor visibility, and difficulty reading them during capture. If points are clustered on one side of the survey area, absent from the central area, or the shapes around points change as excavation progresses, processing may complete but the results will be unstable. As countermeasures, it is important to reconsider the layout to combine the perimeter and interior, to check the relationship between the recorded surface and the reference points each time, and to secure independent points for verification. Simply increasing the number of points will not improve matters if their placement is poor.

The third is underestimating lighting conditions and shooting during times with strong shadows. Even if the relief of the archaeological surface appears more pronounced and easier to see, deep shadows and reflections interfere with image matching. Studies on illumination changes indicate that, while reproducibility is relatively easy to maintain if extremely unfavorable conditions are avoided, non-uniform lighting and low illumination tend to increase error and instability. As countermeasures, fix the time of shooting, record the direction of shadows, and, if necessary, use simple shading or adjust surface conditions to make acquisition conditions consistent each time. In archaeological recording, images taken under different conditions on different days are often compared later, so variations in lighting have a greater effect than one might imagine.

The fourth issue is relying solely on overhead images and failing to notice gaps around steps and wall faces. Even if the base of the excavation unit looks fine, cuttings or stonework can create occluded areas, leaving poorly connected sections. As a countermeasure, while centering on planimetric recording, add oblique and close-up images where necessary, assuming they will cover blind spots. Multiple studies have shown that combining nadir and oblique images in complex cultural-heritage and archaeological settings helps improve reconstruction reliability. Even when an excavation surface appears flat, it often has small elevation differences that will have an effect later, so caution is required.

The fifth point is treating generated orthoimages solely as illustrations for reports and not retaining verification values or capture conditions. This makes them difficult to reuse for next year’s surveys or for additional interpretation. As a remedy, attach at least minimal metadata to each deliverable so that you can trace later under what conditions it was produced. For long-term operation of excavation records, whether the recording process itself can be formalized into a procedure and maintained continuously determines whether the system will be usable on site. Prioritize creating a record system that can be continuously used over merely producing images.

Summary

When creating orthophotos to improve the recording accuracy of archaeological sites, it is more important to first decide what to preserve, establish control points and coordinate management, develop an acquisition plan that accounts for overlap and blind spots, collect data on site without disturbing the site's conditions, and finally verify accuracy both numerically and visually, than to rely on newer equipment or processing environments. Because excavations cannot be undone once progressed, orthophotos must remain not as temporary illustrations but as measurement records that can be referred to later. For this reason, rather than producing images that merely look good, a shortcut to improving the quality of archaeological surveys is to reliably build up georeferenced, reusable materials.

In practice, the quality of orthophotos is not completed by photography alone; it is largely supported by control point surveying and the accuracy of on-site coordinate verification. In situations where you want to quickly check the survey area's control points, reliably secure the positions of targets to be photographed, or link records from multiple days under a consistent coordinate framework, whether there is a system that allows coordinate management to be handled on-site without undue burden will determine work efficiency. By utilizing an iPhone-mounted high-precision GNSS device like LRTK, you can streamline control point checks and position recording in archaeological surveys while making it easier to maintain the coordinate consistency that underlies orthophoto production. The more a site aims to build orthophotos into more reliable records, the more important it is to review not only the imaging process but also the efficiency of the preceding coordinate acquisition.