6 Precautions When Conducting Continuous Observations with RTK

RTK is a convenient method for obtaining high-precision position information on site in real time. For a single observation, you can operate by confirming a Fix and recording the coordinates, but the situation changes somewhat for continuous observations. Sometimes you continue observing under the same conditions for a certain period, and other times you acquire coordinates continuously while moving, but in either case the idea that accuracy only needs to be good for an instant is not sufficient. If corrections become unstable during an observation, reception conditions change, or settings diverge midway, the overall reliability of the recorded data can be reduced without it being obvious at a glance.

Especially when using RTK for as-built verification, current condition assessment, construction management, displacement checks, or continuous position logging, maintaining quality during observations—not just at the start of observation—is important. Operating on the assumption that everything is fine because you had a Fix at the beginning, or because the receiver’s display looks normal, can lead to inconsistencies in the data when it is later processed. In continuous observations, a major practical point is how to capture and manage the changes that occur over time.

RTK is formed by a combination of multiple factors such as satellite geometry, the radio signal environment, correction data, installation conditions, and operational procedures. Small fluctuations that can be overlooked in a single observation can accumulate during continuous observations and affect the results. Therefore, the first thing to grasp is not RTK accuracy itself, but how to manage operations in a way that suits the continuous-observation operational mode.

This article organizes and explains six key precautions to keep in mind when conducting continuous observations with RTK. Rather than simply listing device configuration items, it explains step by step why each precaution is important and how to address it from the perspective of preventing the kinds of shifts and judgement errors that commonly occur in the field. These are items you should be sure to check before operation so that the results of continuous observations become data you can confidently use later.

Table of Contents

• Introduction

• Note 1 First clarify the observation objectives and the required accuracy

• Note 2: Do not underestimate the stability of reference stations and correction information

• Note 3: Monitor changes in the satellite reception environment even while observing

• Note 4 Do not change the antenna height and installation conditions midway

• Note 5 Unify coordinate systems, recording settings, and time management

• Precaution 6: Prepare verification points and procedures for responding to anomalies

• Summary

Introduction

The term "continuous RTK observation" has slightly different meanings depending on the site. In some cases it refers to continuously observing the same point for a certain period to assess the stability or change of coordinates, while in others it means continuously acquiring positions while moving along a route. Additionally, in a broader sense, operations that periodically re-measure the same point are sometimes called continuous observation. However, what is common to all forms is the need to keep the entire set of data obtained over time reliable, rather than relying on single instantaneous values.

In a single observation, if you have a Fix at that moment and it is consistent with a known point, it is relatively easy to assess the quality of that one point. However, in continuous observations, measurements may not remain at the same quality not only at the start but also 10 minutes, 30 minutes, or one hour later. Satellite visibility changes over time, and the communication state of correction services is not constant. Site conditions also change gradually, including changes in temperature and solar radiation, passing vehicles and pedestrians, and the influence of nearby obstacles.

Therefore, in continuous RTK observations, maintaining accuracy is more important than achieving it. Moreover, that maintenance cannot be left up to the receiver. Operators need to decide in advance what to monitor, which states to consider normal, and at what point to declare something abnormal. If this is ambiguous, even if a large volume of records is retained, the data may become unusable later.

For example, problems such as Float becoming mixed in from a certain point, corrections being temporarily lost so that recording continued in a state close to standalone positioning, the antenna height having changed partway through, or the coordinate system setting being switched midway are particularly critical for continuous observations. What makes them even more troublesome is that they often appear normal on site. Even if a position is displayed on the screen, its quality does not necessarily meet the operational requirements.

Because of this, when conducting continuous observations it is necessary to design the pre-observation preparation, monitoring during the observations, and post-observation checks as a single, integrated workflow. Below, we explain six points to watch out for—presented in order—that are particularly easy to overlook in practice.

Important Point 1: Clarify the observation objective and required accuracy first

Before beginning continuous observations with RTK, the first thing to confirm is the objective: why you are observing and what level of accuracy is required. This may seem obvious, but in practice observations sometimes start with surprisingly vague objectives. If you proceed on the assumption that RTK measurements will be highly accurate, you may impose unnecessarily strict operational procedures that reduce efficiency, or conversely omit necessary controls and end up with insufficient quality.

There are several main purposes for continuous observation. For example, what you need to monitor changes depending on whether you want to check the stability of the same point, record the trajectory of a moving object, continuously capture positions in line with construction progress, or confirm displacement trends within a given time period. Even with the same RTK, the required level of management can differ considerably.

If the objective is continuous recording of the current situation, it is important that there are few interruptions in position and that the data are recorded smoothly as a time series. On the other hand, if the observation requires strict alignment with reference points or control points, simply having continuous records is insufficient; during the observation you must check consistency with known points and decide whether to reinitialize. If the use is closer to displacement monitoring, even a few centimeters (a few in) of jitter can be meaningful, so you should not be reassured by maintaining a Fix only for a short time.

What is important here is to translate the required accuracy into operational conditions rather than verbal descriptions. For example, you should make it concrete in forms such as: the planar/horizontal direction to this level, the elevation direction to this level, performing consistency checks at regular intervals, not accepting records other than Fix, and returning to the main observation only after passing through a checkpoint following reinitialization. If the purpose and accuracy requirements are vague, on-site judgments will be vague as well.

Also, because RTK generally tends to exhibit greater variability in the vertical direction than in the horizontal plane, it’s important to be clear about what you primarily want to examine. If you intend to discuss height trends using continuous observation results but, in operation, only look at the horizontal components and judge them as normal, spurious oscillations can later be mixed into the height data. Conversely, when continuity of horizontal positions is important, relying only on the per-observation Fix indicator can cause you to miss unnatural jumps in the trajectory.

The length of the observation period must also be considered according to the objective. Short periods of continuous observation may be sufficient in some cases, while in others it is necessary to assess stability over a certain time window. Data that looks good only for a short time but exhibits drift or communication outages over longer periods will not be suitable for the intended purpose. In other words, the quality of continuous observations should be evaluated not only by instantaneous accuracy but also with respect to the time axis.

In practice, before observations begin, it is important for the site supervisor and the operator to agree on the adoption criteria for the continuous observation. For example, rules such as how many minutes apart to check the status, which indication should trigger an interruption, how many seconds of correction loss will render the data invalid, and which records to keep. If you rely on each worker’s experience without such criteria, quality will vary from day to day even at the same site.

Continuous RTK observations are not determined solely by equipment performance. By defining in advance what will be treated as correct, judgments made during observations become stable, and as a result the reliability of the entire dataset increases. Not leaving the initial setup ambiguous provides the foundation for all subsequent precautions.

Important Point 2: Do not underestimate the stability of reference stations and correction information

In continuous RTK observations, the next important factor is the stability of the reference station and the correction information. RTK does not achieve high precision by receiving satellites alone; it operates by receiving reference information and applying relative position corrections. Therefore, if the source of corrections is unstable, no matter how high-performance the receiver itself is, the quality of continuous observations will fluctuate.

When using your own reference station, you must first verify that the installation conditions of the reference station itself are stable. Whether the mounting surface is secure, whether there is any slight looseness in the tripod or fixtures, whether there is a risk of contact from people or vehicles during observations, and whether power and communications are stable — all of these factors directly affect continuous observations. If the reference station shifts even slightly or the radio signal is interrupted, that impact will extend to the entire record of the rover.

When using network-based correction information, the stability of the communication environment is especially important. Even if a connection is established at the start on site, communication quality can deteriorate over time or temporary reconnections may occur. For single observations you can simply redo them on the spot, but for continuous monitoring a momentary drop in data quality can introduce spurious oscillations or jumps into the entire time series.

What’s worrying here is that the problem isn’t only when corrections are completely lost. If communication degrades so that correction update intervals become irregular or delays increase, observation quality can gradually decline even though coordinates appear to continue being output. In continuous observations, it’s important not to overlook such intermediate unstable states. Rather than judging solely by connected or disconnected status, you need to consider correction reception, the continuity of the Fix, and the stability of the solution together.

Also, the distance from the reference station and the conditions for using network corrections cannot be overlooked during long-duration observations. In general, the stricter the correction requirements, the more susceptible they are to environmental changes. If the site is large and sky visibility or communication conditions vary by location, measurements may be good at the starting point but the quality can suddenly change after moving elsewhere. For continuous observations, you should plan for the entire work area and anticipate in advance where instability is likely to occur.

Power management is also part of correction stability. If the battery levels of not only the mobile station but also the reference station, communication equipment, and related terminals become unstable during operation, communication interruptions or output shutdowns can occur, interrupting continuous observations. Especially for long-duration observations, it is important not to rely solely on the battery level display but to incorporate battery replacement timing and spare preparation into operational procedures. If batteries are hastily replaced mid-operation, resuming work tends to be prioritized over observation continuity, and verification steps are easily omitted.

Furthermore, it is important not to change the correction method or connection settings midway. If communications become unstable on site and you switch to a different connection method or change the correction source, you may be able to continue acquiring coordinates, but the conditions will differ between the first and second halves. If these are treated as the same time series without recording the change, consistency will be lost when performing comparisons or analyses later. If it is operationally unavoidable to change the conditions, you must always record the time and the reason.

Correction information for continuous observations is not something you can take for granted like air. The ability to receive it continuously and stably is itself a quality requirement. Don’t be reassured by connection checks before starting alone; continuously monitor the reference station and the status of corrections during observations, and if you see the slightest sign of abnormality, promptly isolate the issue.

Note 3 Monitor changes in the satellite reception environment even during observations

When conducting continuous observations with RTK, people sometimes feel reassured after confirming the sky is clear at the start. However, what really matters for continuous observations is not only the environment at the start but how the reception environment changes during the observation. Satellite geometry changes over time, and the surroundings at the site are not always constant. Just because there is no problem at the start, it’s safer not to assume you can observe under the same conditions until the end.

For example, beside buildings, at the base of slopes, close to trees, or on sites with many heavy machines and vehicles, even a slight change in position can alter reception. This applies when a worker moves during continuous monitoring, of course, and even if you believe the equipment is installed in exactly the same spot, parked vehicles, mobile equipment, or workers’ movement paths nearby can affect satellite reception. The more temporary these changes are, the harder they are to notice, and when you review the data later they often explain disturbances confined to specific time periods.

Moreover, the effects of multipath cannot be ignored in continuous observations. Reflected signals are often difficult to detect at the moment you are on site, but as satellite geometry changes over time, errors can increase during specific periods. In environments with metal surfaces, water surfaces, glass, or nearby walls in particular, subtle fluctuations can occur that are hard to capture from the receiver's status display alone. While a single occurrence may be overlooked without causing significant problems, in continuous observations it tends to remain as an unnatural oscillation in the time series.

Therefore, during continuous observations, you need to get into the habit of monitoring not only whether you have a Fix but also the number of satellites, changes in reception conditions, solution stability, and trends in position variability. The information displayed on the receiver's screen varies by model, but at minimum you should be able to compare the status at the start, during, and at the end. If there are signs such as a large change in the number of satellites midway, frequent switching of Fix, or the track becoming unusually jittery, that segment must be treated with caution.

Even when repeatedly observing the same point, changes in sky view can occur. At sites where surrounding activity differs between morning and afternoon, or in environments where sunlight changes the conditions around equipment, measurements may be stable immediately after starting but the solution may become unstable as time passes. This is even more true for mobile surveys, so it is important to anticipate in advance which parts of the route might become weak points. Within the overall site, areas beside buildings, under trees, along retaining walls, and narrow passages should be treated as caution zones from the outset.

A preliminary site survey before observations is also effective. Even a short check of reception conditions along the planned route and at the intended installation positions will give you an idea of where quality tends to drop. During the actual continuous observation you are often pressed for time, but if you know weak spots in advance, it becomes easier to respond by increasing checks there or by adjusting to a different position.

It is even more important to record when environmental changes occur. For example, events such as a large vehicle pulling up alongside, a crane coming in nearby, workers clustering close to a survey point, or rain or wind picking up are important clues when reviewing the data later. In continuous monitoring, recording only coordinates is not sufficient; keeping surrounding information related to data quality as well makes it easier to identify the causes of anomalies.

Because RTK relies on satellites, it cannot be immune to sky conditions. In continuous observations, those effects will manifest over time. Don’t be satisfied with checking only at the start; continue monitoring on the assumption that reception conditions may change during the observation, as this leads to more stable results.

Precaution 4: Do not change antenna height or installation conditions midway.

In continuous RTK observations, it is extremely important to keep the antenna height and installation conditions consistent. However, on site this aspect is sometimes treated rather lightly. In single observations, it may be sufficient to check and record the antenna height each time, but in continuous observations small changes during the session can have cascading effects on the entire dataset. What makes it particularly troublesome is that changes in antenna height and installation conditions are often not noticeable from the equipment display alone.

For example, when continuously surveying while moving with a pole, the effective observation conditions can change depending on how the operator holds it, how they walk, and how the pole tilts. Even if you keep the same height setting, if the pole is not kept perfectly vertical at all times, the coordinates obtained will not be stable. Likewise, even in fixed observations, if changes occur such as a tripod leg sinking slightly, the mounting surface vibrating, or the fixation loosening, then although the continuity of the observations may appear to be maintained, in practice data from different conditions will be mixed.

Antenna height directly affects the vertical direction, so careful management is required, especially for observations that include elevation. Even differences of only a few millimeters (a few hundredths of an in) to a few centimeters (a few tenths of an in) can be significant when checking displacements or sorting out height differences. If, during continuous observations, you change the pole length, alter the grounding condition of the spike, or transfer the instrument to a different fixture, you should consider that the observation conditions switched at that point. If you treat them as the same data series, it will later appear as an unexplained step.

In practice, you should establish the reference for antenna height before observations and, as a rule, not change it during the session. If a change becomes unavoidable, make a clear break at that point, separate the records before and after the change, and, if necessary, re-establish consistency using known points or check points. If you change it but simply continue recording, you may not be able to identify that boundary later.

Consistency of installation conditions is equally important. For example, differences such as using tilt compensation in one section but not in another, a change in how the pole tip contacts the ground partway through, or a change in the installation surface from pavement to soil will affect the quality of continuous observations. Even receiver features that seem convenient will change the meaning of the coordinates obtained if operating conditions change along the way. Especially when multiple people hand over observations, it is important to standardize operations as much as possible, including how equipment is held, how the pole is set up, and even habitual actions during checks.

For continuous monitoring with a fixed installation, the stability of the mounting platform must not be overlooked. In locations with soft ground, it can settle slightly over time, and exposure to sunlight or temperature fluctuations can change the condition of the mounting materials. Outdoors, wind can also cause minor, sustained shaking that introduces variability into the measurements. Changes that are undetectable in single measurements can appear as gradual trends when plotted from continuous monitoring.

To prevent such issues, taking installation photos and recording antenna heights before observation, performing interim inspections, and conducting checks at the end are effective. Confirming the same conditions at the end as well as at the start makes it easier to determine whether anything changed during the observation. Especially for long-duration observations, simply confirming that the antenna height and installation condition at the start and end match can provide significant reassurance.

Because RTK deals with numerical coordinates, it's easy to focus only on the display on the screen, but those numbers are supported by the physical installation conditions on site. In continuous observations, even when equipment appears to be in the same location, results change if the installation conditions vary even slightly. That's why it is essential to rigorously follow the basic rule of not changing the antenna height or installation conditions partway through.

Point 5: Standardize coordinate systems, recording settings, and time management

In continuous RTK observations, attention tends to focus on reception conditions and installation conditions in the field, but inconsistencies in settings are actually also a major source of problems. In particular, if the coordinate system, recording format, output interval, or time management shift partway through or are not aligned between devices, the observations themselves may appear to have gone well, yet the data can become unusable during post-processing or comparison. For continuous observations, consistency of these settings is more important than for single observations.

First, about coordinate systems. In RTK, positions may be handled in latitude and longitude or in planar coordinates such as a plane rectangular coordinate system. Also, regarding heights, the meaning changes depending on whether you use ellipsoidal height as-is or treat it as elevation. If these settings change during continuous observations, even though the data are from the same site you will not be able to directly compare the values from before and after. Moreover, because the numbers alone can appear plausible at first glance, it is easy to be slow to notice.

For example, in practice it can happen that data recorded in the morning using a plane rectangular coordinate system are resumed on another terminal in the afternoon and appear in latitude/longitude format, or that the height reference has changed. In continuous observations, comparisons between intervals and understanding temporal changes are important, so such discrepancies in settings can be critical. As a rule, before starting observations you should clearly specify which coordinate system, which vertical datum (height reference), and which units will be used for recording, and not change them midway.

Next are the recording settings. In continuous monitoring, the recording interval, whether to save only during certain states, and how much attribute information to retain directly affect quality. If the recording interval changes midway, the density of the time series will change and make analysis difficult, and if the system is set to automatically save states other than Fix, lower-quality data will be mixed in. Conversely, if logs of state changes are not retained, it becomes difficult to identify abnormal periods afterward.

In continuous observations, it is important to record not only the coordinates but also the state at that time. At minimum, keeping track of the observation time, the solution status, and, when necessary, satellite status and the status of corrections will make later quality assessment easier. If the recorded data are only neat sequences of numbers, you may not be able to trace the cause of anomalies and thus may be unable to determine whether to accept them.

Time management is another point that is easy to overlook. In continuous observations, you need to track exactly when events occurred. If the clocks of the receiver unit, the recording terminal, auxiliary logs, and field notes are out of sync, it becomes difficult to associate correction outages or environmental changes with data anomalies. For example, even if a field note says communications were interrupted at around 10:15, if the equipment clocks are off by a few minutes it becomes unclear which segment of the data was affected.

When performing simultaneous continuous observations with multiple units, extra caution is required. If the settings or time differ even slightly between receivers, you may think you are observing the same phenomenon but end up with results that cannot be compared. In particular, at sites where multiple people operate in shifts, settings can revert to their defaults when switching devices, or input formats can vary from person to person. Align the list of settings before observation and make sure all operators follow the same operating rules to reduce such mistakes.

Standardizing file names and recording rules is also practical for operations. In continuous observations the amount of recorded data increases, so unless you make it immediately clear which time period and which conditions each dataset corresponds to, you will become confused during the organization stage. Recording the start time, observation location, instrument name, whether conditions changed, and so on according to a consistent rule makes it easier to verify data quality later. This is not mere paperwork but quality control to protect the reliability of the results.

Coordinate systems and recording settings may seem more mundane than the observations themselves. However, in continuous observations, even small inconsistencies in these settings can greatly reduce the overall value. Standardize them before measuring, do not change them while measuring, and if you do change them, always record the change. By rigorously following this basic rule, you will obtain continuous observation data that can be handled with confidence later.

Precaution 6: Prepare verification points and procedures for responding to anomalies

In continuous RTK observations, preparing before starting and monitoring during observations are important, but that alone is not sufficient. What ultimately matters is being able to judge how much you can trust the data. To do that, you need to have verification points to check quality along the way and procedures in place for responding if anomalies occur. If you carry out continuous observations without these, you may be unable to stop when a problem arises and end up having to question all of the data afterward.

A verification point is a reference point used to confirm the correctness of observations. It can be a known point or a control point that can be stably and repeatedly checked on site. The important thing is to check that point before, after, and during continuous observations so you can assess the stability of the solution and whether any drift has occurred. Because data continue to stream during continuous observations, small drifts can develop without being noticed. By inserting verification points at regular intervals, it becomes easier to determine whether the overall observation remains within a normal range.

For example, if you establish a procedure of checking verification points before the observation starts, rechecking them at checkpoints during the observation, and measuring the same points again at the end, it becomes easier to determine whether conditions changed at any point. If there is a discrepancy between the verification results at the start and at the end, a problem may have occurred somewhere in between. In continuous observations, this kind of before-and-after comparison approach is effective for detecting anomalies.

Procedures for handling abnormal conditions should also be decided in advance. For example, decision criteria such as what to do if the Fix is lost, whether it is acceptable to resume immediately when corrections are reconnected, whether to perform a known-point check, or after how many minutes of communication outage a segment should be invalidated. Without such criteria, field crews tend to just continue working, which can result in a mix of low-quality data.

One point to be especially careful about is not to conclude that the original quality has immediately been restored just because an anomaly appears to have been resolved. For example, even if a correction was temporarily lost and then reconnected, returning to continuous observation without reinitialization or rechecking can mean the conditions for the solution before and after are not aligned. Even if coordinates appear to be available, there are cases where, from a quality-assurance perspective, the data should be segmented. You need to understand that in continuous observation there are situations where you should prioritize consistency over continuity.

How to take field notes is also important. Concisely recording information such as what happened at what time, whether reinitialization occurred, whether the observation position was changed, whether communication degraded, or whether there were changes in the surrounding environment improves the accuracy of judgments during post-processing. Vibrations whose cause cannot be determined from the data alone may reveal their reason when cross-checked with the field notes. In continuous monitoring, the perspective of evaluating quality by combining numerical data with field records is indispensable.

It is also important not to rely too much on individual experience when handling anomalies. Even in situations where a veteran would notice something is off, decisions can become inconsistent when the person in charge changes. Therefore, for example, if a loss of 'Fix' persists for a certain period, suspend operations; before resuming, verify checkpoints; when the antenna height is changed, resume with a new file. By establishing such procedures so that anyone can make the same judgment, quality stabilizes. Since continuous observations can last for long periods, this kind of standardization is particularly effective.

Post-observation inspections are also part of anomaly response. Simply glancing over the recorded trajectories and time series to check for unnatural jumps, steps, or sudden increases in scatter can lead to early detection of problems. If you perform a quick check before packing up at the site, you can reobserve or carry out additional checks if necessary. Noticing inconsistencies later can be problematic, since site conditions may have changed and rechecking may become difficult.

When continuous RTK observations are going well, people tend to forget to plan for anomalies. However, because something will inevitably change in the field, incorporating validation points and response procedures from the outset will minimize damage if problems occur. For successful continuous observation, it is important not simply to keep measuring without interruption, but to detect anomalies and correctly isolate them.

Summary

When conducting continuous observations with RTK, quality must be managed from a different perspective than for single observations. Crucially, you should not focus only on the Fix display at a given moment, but be mindful of whether you can continue to acquire data under the same conditions and with the same meaning from the start to the end of the observation. In continuous observations, the passage of time itself becomes a risk factor. You must design the entire operation to cover correction stability, the satellite reception environment, installation conditions, unification of settings, and responses to anomalies.

The six cautionary points discussed here may seem independent, but in reality they are deeply connected. If the observation objectives and required accuracy are clear, it becomes easier to determine how much correction stability is necessary. By monitoring corrections and the reception environment, you can detect signs of anomalies earlier. If antenna height and installation conditions are kept consistent, and coordinate systems and recording settings are standardized, the meaning of time-series data will remain consistent. Furthermore, if validation points and procedures for responding to anomalies are prepared, even if problems occur along the way it will be easier to judge which portions of the data can be trusted.

A common practice in the field is to begin continuous observations with the attitude that “RTK is high-precision, so it’s fine” or “it’s Fix, so there’s no problem.” However, what truly matters is not using high-precision equipment, but maintaining a high-precision state and being able to explain that that state was maintained. In continuous observations, the greater the volume of data, the more differences in quality control are reflected in the results. Carefully managed short datasets can even be more valuable than long datasets with lax management.

If you are going to conduct continuous observations with RTK from now on, I recommend first translating these six points into on-site procedures. Decide the objectives and criteria before observation, prepare the pre-start check items, establish how to monitor and take notes during observation, and decide how to respond to anomalies. Simply having these preparations in place can greatly change the reliability of the results, even when using the same equipment.

Continuous RTK observations are an extremely powerful tool when operated correctly. However, if operations remain vague, the data can look orderly yet end up being difficult to use in practice. That is why it is important not to leave everything to the equipment but to continuously manage the observation conditions. Rather than simply prolonging continuous observations, be mindful of linking them with the same quality from start to finish. This is the basic principle for achieving reliable continuous observations with RTK.