7 Causes of Poor Accuracy in Drone Surveying

Drone surveying makes it easy to grasp large areas in a short time and is extremely effective as a preliminary step for site condition checks, as-built assessment, volume calculations, and drafting. On the other hand, on-site there are often complaints such as “the accuracy is not as good as expected,” “the point clouds or orthophotos are misaligned,” and “they don’t match when overlaid with previous data.” It is easy to assume that simply flying a drone will automatically produce high-accuracy results, but in reality the final accuracy is determined by a combination of factors such as flight planning, shooting conditions, how control points are established, processing methodology, and the site environment.

What is especially important for practitioners is not to look only at the conclusion that “the accuracy was poor,” but to break down and consider which stages of the workflow are prone to introducing errors. Accuracy degradation in drone surveying is not caused solely by the performance of the aircraft. Rather, there are many cases where accuracy suffers due to insufficient preparation, inadequate condition settings, or a lack of verification procedures. In other words, if the causes are correctly identified, there is ample room to improve accuracy even with the same equipment.

In this article, we summarize seven common causes of poor accuracy in drone surveying and, from a practical standpoint, explain why each leads to errors and how to address them in the field. This will be useful not only for those planning to put drone surveying into full operational use, but also for those who have already adopted it and are noticing variability in results. Rather than treating accuracy issues intuitively, read on to use this as decision-making material for improving accuracy in a reproducible way.

Table of Contents

• What to consider first when drone surveying isn't producing accurate results

• Cause 1: The flight plan does not match site conditions

• Cause 2: Insufficient design of calibration and validation points

• Cause 3: Camera settings are too lax, and image quality is unstable.

• Cause 4: Underestimating weather and lighting conditions

• Cause 5: Choosing a method that is not suitable for the target terrain or surface conditions

• Cause 6: The handling of coordinates and reference points is ambiguous.

• Cause 7 Analysis and verification processes are insufficient

• Summary

What to consider first when you can't achieve accuracy in drone surveying

When considering the accuracy of drone surveying, the first thing to clarify is which accuracy you are seeking. On site, "accuracy" can mean different things depending on what is being examined: horizontal position shifts, vertical height offsets, the fidelity of shape reproduction, the accuracy of overlaying datasets from multiple times, and so on. For example, the required level of accuracy changes depending on whether the goal is a rough current-condition overview, volume calculation, or construction management. If the purpose remains vague, you may end up being stricter than necessary in your evaluation, or mistakenly judge data that is actually insufficient to be adequate.

Also, the causes of poor accuracy do not necessarily occur in isolation. In practice, several minor unfavorable conditions—slightly high flight altitude, insufficient overlap rate, a biased distribution of ground control points, and overcast skies that make surface features hard to see—can combine to produce large deviations. For this reason, a single countermeasure is not enough; it is important to review the entire workflow from upstream to downstream processes.

A common occurrence in the field is to look only at the analysis results and conclude, "It's a software problem" or "the aircraft's accuracy is poor." However, in reality, pre-flight planning and post-flight verification often have a greater impact. When accuracy is lacking, it's all the more important to return to the basics: pre-flight conditions, on-site standards, image capture quality, and consistency checks during analysis.

From here, we will examine in detail seven causes for the lack of accuracy in drone surveying.

Cause 1: The flight plan does not match the site conditions

The primary cause is that the flight plan does not match the on-site conditions. In drone surveying, the way you fly directly affects the accuracy of the results. If flight altitude, flight speed, image capture interval, direction of travel, overlap rate, or the way the flight route is planned are inappropriate, no matter how much you refine post-processing, the accuracy will not be stable.

A particularly common case is raising the flight altitude too much to prioritize efficiency. Increasing altitude lets you capture a wide area at once, which makes the work appear faster, but fine ground features become harder to capture. As a result, feature points between images cannot be extracted sufficiently, and alignment becomes unstable. Even when you want to capture subtle surface undulations and boundaries, flying too high can leave you with an insufficient amount of necessary information, causing the resulting point cloud and orthophoto to appear blurred.

Insufficient overlap settings are also a typical problem. If forward and side overlaps are inadequate, the connections between images become weak, making local distortions and misalignments more likely during processing. Even if a model appears to have been created visually, errors can become amplified at some edges or where the terrain changes significantly. This is especially true on slopes and highly undulating development sites, where deciding flight conditions with the same mindset as for flat ground tends to result in insufficient coverage.

Furthermore, having a flight direction that is too monotonous is also a factor in reduced accuracy. If you capture images from only a single direction, features may be unevenly recorded depending on the shape of the subject, making three-dimensional reconstruction unstable. For long, narrow terrain or areas with elevation differences, adjusting the flight path or, when necessary, adding supplementary captures can improve the reproducibility of the shape.

When creating a flight plan, you need to design it not just based on area, but also taking into account terrain irregularities, whether structures are present, vegetation conditions, and the type of deliverable required. Prioritizing conditions that enable stable downstream analysis, rather than trying to capture an entire large site at once, will ultimately reduce rework. If you feel the accuracy is insufficient, the first step is to review whether this flight approach actually captured the information you needed.

Cause 2 Design of calibration and validation points is insufficient

The second cause is inadequate design of control and check points. Establishing ground reference points is crucial for stabilizing the accuracy of drone surveys. Although it is possible to align positions using only aerial images, depending on site conditions and the required accuracy, a lack of ground-based references may make the overall position and elevation unstable.

A common problem is focusing only on the number of control points without considering the balance of their placement. Even if there are enough points, if they are clustered on one side of the site or concentrated in the center, distortions tend to occur at the edges and along the perimeter. This is especially true for large sites or elongated areas: unless the layout holds the whole area evenly, the center may appear correct while errors increase toward the outside. You must also be mindful of vertical balance, not just the plan direction, on sites with elevation differences.

Also, there are cases where calibration points and validation points are not considered separately. Calibration points are used to stabilize the analysis, while validation points are used to confirm that the results are truly correct. If you use everything in the analysis, it becomes difficult to objectively assess the quality of the results. Even if things look consistent analytically, it is not uncommon for them to be off when checked against independent verification points. In practice, the idea of separating the creation and the verification is extremely important.

Furthermore, there can be problems with the way the control points themselves are installed. If they are hard to distinguish from the surroundings, cast in shadow, blended into ground patterns, or placed on an unstable surface, they cannot be read correctly on the image. Even if coordinates are obtained accurately, if the center is misread on the image, that error will propagate through the whole result. Measurement points need to be placed where they are easy to see, easy to read, and unlikely to be moved or deformed.

We must not overlook that the way control points are measured is itself often ambiguous. If ground-based observations are insufficient, it is meaningless to improve only the drone side. If the accuracy of the reference is low, results adjusted on that basis will naturally be unstable. It is necessary to recognize that the accuracy of drone surveying is not completed solely by aerial measurements, but is supported by the reliability of ground control.

When accuracy is insufficient, it's important not to judge based solely on the presence or absence of control points, but to review factors including placement bias, how you determine the number of points, the independence of validation points, visibility, and the observation method. Simply improving the ground-side design can greatly change the stability of the analysis results.

Cause 3: Lax shooting settings leading to unstable image quality

The third cause is inadequate capture settings and unstable image quality. In drone surveying, final deliverables are produced from a collection of images and measurement data. If the quality of that source data is low, there is a limit to the achievable accuracy no matter how carefully it is processed. Especially in methods that use photographs, the sharpness of each image and the stability of exposure directly affect how easily images can be aligned.

A common problem is motion blur. When the flight speed is too high, or the aircraft experiences small vibrations due to wind, the images become slightly blurred. Even if it is not very noticeable to the naked eye, it affects the accuracy of the feature-point extraction used for analysis. If the features become less distinct, the correspondences between images become unstable, and as a result the overall consistency of the model is compromised. This is especially true in areas with little surface texture, where even a small amount of blur can have a large impact.

Variations in exposure are also problematic. Overly bright images suffer from blown highlights, while overly dark images lose fine detail. Furthermore, if brightness changes significantly during flight, the same subject may appear differently in different images, which can make analysis unstable. When surface texture and boundaries of the ground are not clearly visible, the accuracy of both planimetric position and elevation tends to degrade.

Lack of sharp focus is another factor that is often overlooked. Even if an image is slightly blurred overall, it can appear to be adequately captured at first glance, so it may be processed as is. However, for surveying purposes, what matters is not whether an image looks clean but whether the features required for analysis can be reliably extracted. Sometimes only the edges are blurred, or only certain contiguous sections suffer a drop in quality, and this can cause local distortions.

Another issue is an inappropriate image capture interval. If the capture interval is too long relative to the aircraft's speed, the expected overlap rate will not be achieved. Even if sufficient overlap is assumed in the plan, actual flight speed or deviations in capture timing can lead to insufficient overlap. This is one of those problems that are hard to detect afterward, and it is not uncommon to only notice it once the results have degraded.

Camera settings are not something you decide once before arriving on site and forget; they should be adjusted according to the weather and the subject. The conditions required differ between flat ground and slopes, bright daylight and thin cloud cover, and paved surfaces and grass. If you want to stabilize accuracy, you need to prioritize not just whether you can fly, but whether the images are captured at analysis-grade quality. Even checking a few shots after capture can help you detect blur or exposure problems early.

Cause 4 Underestimating Weather and Lighting Conditions

The fourth cause is underestimating weather and lighting conditions. While drone surveying often focuses attention on the aircraft and data-processing technologies, in practice it is strongly affected by natural conditions. On site, because of the desire to fly as scheduled, operations are often carried out in light wind or cloudy conditions, but such decisions can lead to reduced accuracy.

First, wind affects both the aircraft's attitude stability and image quality. Not only during strong winds, but even light winds with intermittent shaking can reduce the repeatability of the flight path and subtly disturb the attitude during shooting. As a result, images that should have been taken under the same conditions may not look consistent, increasing errors during analysis. In particular, the higher the altitude, the more susceptible you are to wind effects, so conditions that seem fine on the ground can be poor aloft.

Lighting conditions are also important. For example, during periods of strong direct sunlight, shadows become pronounced. If shadows from structures, slopes, or trees cover parts of an image, surface features become difficult to see. Conversely, uniform overcast skies can weaken shadows and sometimes make features easier to observe, but if the light level is too low the image becomes dark and features again become hard to discern. In short, it’s not simply a matter of brighter being better; what matters is whether the lighting conditions allow the target features on the ground to be stably recognized.

Reflections are another element that cannot be overlooked. Puddles, wet ground, metal surfaces, and glass surfaces can look very different depending on the time of day and the viewing angle. If they appear overexposed in one image and underexposed in another, it becomes difficult to match images. At land development or construction sites, effects from the previous day’s rain or from watering may remain, so you must take into account the ground conditions on the day itself.

Furthermore, differences in solar altitude due to season and time also affect the results. Light at a low angle produces long shadows, and on undulating terrain some parts are easily obscured. Conversely, even light close to overhead can produce weak contrast on pale surfaces, making features difficult to capture. Depending on site conditions, it is necessary to consider which time of day will yield the most stable images.

In drone surveying, it's essential not only to choose a day when flights are possible but also to select days and time windows that make it easier to achieve accurate results. Weather conditions cannot be controlled, but by not overlooking them you can prevent major failures. At sites where accuracy is lacking, you should review not only the aircraft and processing settings but also whether the timing of the operation itself was appropriate.

Cause 5: Selecting a method that is unsuitable for the target terrain or ground surface conditions

The fifth cause is choosing a method that does not suit the site terrain or ground surface conditions. Even in drone surveying, the same approach is not applicable to every site. The appropriate measurement techniques and flight plans differ between flat, open areas; tree‑covered areas; locations with continuous slopes; and sites with densely clustered structures. If this is misjudged, it can be difficult to achieve accurate results even when equipment and procedures are fine.

For example, in areas with grassland or dense vegetation, the information captured in photographs can often correspond to the top of the vegetation rather than the ground surface itself. Even if it appears to capture the ground, the height of the vegetation may be mixed in, and using such data for soil volume calculations or site-development comparisons can produce large errors. In particular, when comparing before and after construction, differences in grass height due to seasonal variation can appear directly as changes, so caution is required.

Problems are also likely to occur on slope faces and cut areas. Aerial photography alone may not adequately capture the shape of a slope. The steeper the slope, the more likely the apparent area and angles in the images will be biased, reducing reproducibility. As a result, flat sections may be correct while only the slopes become distorted, and the heights at boundaries may look unnatural. At such sites, acquisition methods must be adapted to the terrain.

In areas with many structures, roofs, walls, scaffolding, heavy equipment, and temporary installations become intricately intertwined, increasing shadows and occlusions. Photographic methods tend to produce blind spots, and target surfaces may be processed without being adequately captured. In such cases, missing geometry and unnatural surfaces can occur, reducing the reliability not only of planar positions but also of three-dimensional shapes. In situations that require a three-dimensional understanding, such as as-built management or clash/interference checks, simply photographing the whole site from above may not be sufficient.

Also, monotonous ground surfaces can be surprisingly difficult conditions. On paved plazas, uniform soil surfaces, or development sites with little patterning, there are few features to match between images. From an analysis standpoint, the lack of landmarks makes image alignment prone to instability and can cause local stretching and shrinking. Even if it looks like a plain flat surface to the human eye, it is important to understand that it is a challenging subject computationally.

In other words, the accuracy of drone surveying is influenced not only by the aircraft’s performance but also by its compatibility with the target. If you apply your usual method without checking site conditions, you tend to end up with results where, for some reason, only this time the accuracy is poor. To ensure accuracy, it is essential to clarify in advance whether the area is flat or highly undulating, whether there is vegetation, whether there are many structures, or whether surface features are sparse, and to choose an acquisition method that matches those conditions.

Cause 6 Handling of coordinates and reference frames is ambiguous

The sixth cause is ambiguity in how coordinates and reference frames are handled. When people say that positions don't match in drone surveying, it may not be an issue with image capture or analysis, but rather that the references being compared are different to begin with. This is a very practical pitfall and is more likely to occur when assumptions are not aligned among site staff, surveyors, and designers.

A typical example is a case where the reference systems for horizontal coordinates and elevations are not unified. If the current survey data use one reference, the design drawings another, and past survey results are treated differently, misalignment will occur when they are overlaid. To the eye, this may look like poor accuracy in drone surveying, but in reality it is a mismatch in the comparison conditions. If you evaluate it as "off by tens of centimeters (several to a few dozen in)" without checking differences in the references, you will misdiagnose the cause.

The same applies in the vertical direction. Height is particularly prone to misunderstanding, and if you confuse values such as ellipsoidal height and the elevation used in practice, large discrepancies can arise. If the horizontal plane matches reasonably well but only the heights are unnaturally offset, you need to doubt not only the measurement methods and analysis conditions but the height datum itself. On site, people may believe the height datum is verbally shared, yet in actual calculations and output it may be handled differently.

Coordinate units and the handling of the origin are also points to watch. When importing external drawings or existing data, differences in how units or axis directions are interpreted can cause positions to be slightly misaligned or appear rotated. This is often mistaken for an issue with capture accuracy, but the root cause is the alignment settings between datasets. Especially when multiple software packages or multiple people are involved, a tiny discrepancy can be introduced during conversion, and on site it may be perceived as “the drone’s accuracy is poor.”

What’s important on-site is to clarify before the flight “what will ultimately be overlaid on what.” If you know whether you’ll be comparing against the existing condition, the design, the as-built result, past data, or external drawings, the required criteria and inspection items will become clear. Conversely, if you decide how the measurements will be used only after measuring, rework due to mismatched criteria is likely to occur.

To correctly evaluate the accuracy of drone surveying, it is important to consider measurement accuracy and data consistency accuracy separately. Even if the measurements themselves are good, if the conditions for comparison are misaligned, the results can become difficult to use. When you suspect poor accuracy, you should first carefully check whether the reference standards are aligned and whether assumptions about the coordinate system and heights are consistent.

Cause 7: The analysis and verification processes are insufficient

The seventh cause is insufficient analysis and verification processes. In drone surveying, the job is not finished the moment you fly and collect data. Rather, how carefully you check the data afterward, how quickly you detect anomalies, and whether you can decide on necessary corrections or re-acquisition greatly affects the final accuracy. If this step is omitted, deliverables will be produced while still containing problems.

A common mistake is to treat the fact that a model could be generated as evidence of success. Just because processing ran to completion does not mean the results are sufficiently correct. Even if there are forced connections between images, the whole can still take shape. However, hidden within it may be local distortions, edge undulations, and positional shifts in specific areas. Unless you check not only the appearance but also discrepancies from validation points, overlap with time-series data, and consistency with known points, you cannot know the true accuracy.

Also, when outliers are found, one may not trace back and consider their causes. For example, if only some validation points show large discrepancies, rather than simply excluding those points and finishing, it is necessary to consider whether it is due to a bias in control point placement, problems with image capture quality, or the challenging nature of the target terrain. Repeating ad hoc corrections without identifying the cause will lead to operations that are not reproducible.

Insufficient verification also appears as a lack of comparison targets. If there is no step to cross-check against previous deliverables or ground survey results, you are left to judge only by apparent naturalness. This is especially inadequate in situations that require numerical judgment, such as earthwork volume calculations or as-built verification. To produce results that can be used on site, it is necessary to define in advance what degree of difference is acceptable and to establish a system to evaluate against that standard.

Furthermore, delays in deciding whether to re-fly are also a problem. If image quality and coverage are checked immediately after capture, missing areas can sometimes be addressed on the spot. However, if problems are noticed only after bringing the data back for processing, it becomes necessary to return to the site, increasing man-hours. Sites that require stable accuracy place importance on an immediate primary check right after acquisition. In other words, accuracy is not something to be hoped for later; it is something you must lock down on the spot.

Accuracy management in drone surveying is not achieved by flight skills alone. Only by operating data acquisition, analysis, validation, and reassessment as a single continuous workflow can stable results be obtained. When accuracy is lacking, it is as important to reflect on whether the verification process itself was sufficient as it is to review the capture conditions.

Summary

The causes of poor accuracy in drone surveying are not simply a matter of aircraft performance. Multiple factors are involved: flight plans that do not match the site, insufficient design of control points and verification points, variability in image capture quality, overlooking weather and lighting conditions, incompatibility with the target terrain, mismatches in coordinate and elevation reference systems, and insufficient analysis and validation processes. When accuracy problems occur on site, what’s important is not to judge based on the results alone, but to sequentially isolate which stage was most likely to have introduced the errors.

In practice, the most effective routine is to clarify the purpose and required accuracy before flight, develop a flight plan suited to site conditions, carefully establish ground control, ensure imaging quality, and always verify the results after analysis from an independent perspective. Rather than flashy measures, aligning standards and not skipping verification procedures leads to more consistent outcomes. Drone surveying is a convenient method, but achieving accuracy requires the accumulation of careful preparation and validation.

Furthermore, to make data acquired by drones truly usable on site, it is important not to rely solely on aerial measurements but to combine them with ground-based position checks and reference control. In particular, if you are aiming for construction management, as‑built verification, and overlaying with design data, you need to consider both a system for quickly capturing the current site conditions and an on‑site capability to handle high‑precision reference control.

In that regard, at sites where you want to make ground control acquisition and position checks more convenient, the idea of combining an iPhone-mounted GNSS high-precision positioning device such as LRTK is effective. If it becomes easier to acquire control points and check points before and after drone surveys, matching them with aerial data becomes easier and the reliability of the results can be improved. When you are struggling with the accuracy of drone surveys, it is especially important to review the entire operation—not only flight conditions but also on-site control management.