Why Is My RTK Accuracy Dropping? Complete Diagnosis Guide

Coordinates look plausible but are consistently 30–80 cm off control points. The NTRIP connection is active. The receiver appears to be working normally.

The solution never resolved to Fixed, or dropped from Fixed to Float mid‑survey without a visible alarm.

1. Check solution status explicitly before every recording session. In ApekSurv, the solution type is displayed as text in the status bar. Do not rely on the network connection indicator alone.
2. Set a minimum accuracy threshold alert. Most RTK survey software allows you to configure an alarm if horizontal RMS exceeds a defined threshold. Set this to 30 mm for precision survey work.
3. If the solution is Float, check differential age. If above 3 seconds, the correction data stream has been interrupted. Reconnect the NTRIP client and wait for Fixed before resuming.
4. If Fixed was achieved but then dropped, move to an area with better sky visibility and allow 2–3 minutes for reinitialisation.

Fixed solution is achieved, but recorded coordinates jump by 10–50 mm between adjacent measurements at the same physical point.

Satellite signals bounce off nearby reflective surfaces (metal roofing, chain‑link fences, glass facades, parked vehicles, water bodies) before reaching the antenna, introducing systematic timing errors.

1. Identify and move away from multipath sources. A 3–5 metre repositioning can dramatically change multipath geometry.
2. Set elevation mask to 10–15°. Satellites near the horizon transmit through the thickest atmospheric cross‑section and are most prone to multipath. Cutting low‑horizon signals improves solution quality even if it slightly reduces satellite count.
3. Enable the APEKS 120° IMU tilt compensation only after achieving a clean Fixed solution. IMU tilt works correctly in Fixed mode; in Float mode it applies accurate tilt correction to an already‑degraded position.
4. Re‑observe known control points. If your final survey RMS on control points exceeds 20–30 mm horizontal after repositioning, the environment may be too compromised for single‑session RTK — consider static GNSS observation for that location.

Fixed is achieved near populated areas but drops to Float in the outer parts of the survey block, or on project sites more than 50 km from the city.

Most CORS networks guarantee reliable Fixed solutions within 50–70 km of the nearest physical reference station. Beyond this range, atmospheric decorrelation between the rover and reference station degrades the correction accuracy.

1. Check the coverage map for your CORS provider. Most national networks publish interactive maps showing reference station locations and guaranteed coverage polygons. Confirm your project site falls within the polygon.
2. Select the closest available Mountpoint. If multiple mountpoints are listed, always select the one geographically nearest to your project site — even a 10 km difference in baseline length can make the difference between Fixed and Float in marginal coverage zones.
3. If no CORS mountpoint is within 50 km, deploy a self‑deployed base station. Use any APEKS RTK receiver as a base with 2W UHF radio for sites within 15 km, or the APEKS MAX5 with 5W LoRa for sites up to 25 km.

Accuracy is good in the morning but degrades to Float in the afternoon, or suddenly worsens on days following solar activity.

The ionosphere and troposphere introduce atmospheric delays that GNSS receivers must model and correct. During periods of high ionospheric activity — particularly near the equator during solar maximum — residual atmospheric errors can exceed the corrections being applied.

1. Plan precision surveys for early morning hours (06:00–10:00 local time) when ionospheric activity is typically lowest. Avoid post‑noon sessions for highest‑precision work in equatorial survey regions including Indonesia, Malaysia, and East Africa.
2. Check GNSS space weather status. Services such as the US Space Weather Prediction Center (swpc.noaa.gov) publish Kp indices — values above 5 indicate geomagnetic storm conditions that significantly degrade RTK performance globally.
3. Shorten the baseline. Atmospheric decorrelation scales with distance. Moving the base station 10 km closer to the survey area reduces atmospheric error contribution proportionally.

Horizontal coordinates check correctly against control, but all elevation values are consistently wrong by a fixed offset — for example, every point is 0.15 m too low or 0.23 m too high.

The antenna height was entered incorrectly, or the wrong measurement reference was used (slant height vs vertical height).

1. Confirm the measurement reference. APEKS receivers require the vertical height from the ground mark to the antenna reference point (ARP), which is marked on the receiver body. If your pole has a built‑in height of 2.0 m and you add a tribrach and tribrach adapter, the total vertical height must account for all components.
2. Re‑measure all heights on site if there is any doubt. A systematic elevation error that is consistent across all points is almost always an antenna height entry problem, not a satellite geometry problem.
3. Use the APEKS APS1 handheld RTK or a known benchmark to verify elevation independently before starting a precision levelling survey.

RTK shows Fixed with excellent RMS, but coordinates do not match existing survey data or national grid marks by metres.

The project is configured with the wrong geodetic datum or projection, or the geoid model applied to convert ellipsoidal heights to orthometric heights is incorrect for the country.

1. Confirm the datum in ApekSurv project settings matches the CORS broadcast datum and your existing survey data. Country reference: Indonesia = DGN95/WGS84; Saudi Arabia = GCS‑80; South Africa = Hartebeesthoek94; Brazil = SIRGAS2000; Turkey = TUREF/ITRF96.
2. Apply the correct geoid model. Ellipsoidal height and orthometric (sea‑level referenced) height can differ by 30–60 metres in some regions. Without the correct national geoid model applied in ApekSurv, all elevation data will carry a systematic error.
3. Verify against a known control point before beginning productive survey. Checking coordinates against one national grid monument at the start of each session takes 5 minutes and prevents hours of office rework.