drone-mapping emlid kb-autopilot orthomosaic

Q: Complete PPK Drone Mapping Workflow: From Field to Orthomosaic

Updated 12 min read

Quick Answer

PPK drone mapping uses a GNSS receiver on your drone to log positioning data during flight, then corrects it afterwards using base station data. The result is centimetre-accurate geotagged photos that produce survey-grade orthomosaics, digital surface models, and 3D point clouds.

What Is PPK and Why Use It for Drone Mapping?

PPK stands for Post-Processed Kinematic. Unlike RTK (Real-Time Kinematic), which applies corrections live during flight, PPK logs raw satellite observations on both the drone-mounted rover and a stationary base station, then processes them after the flight is complete.

This distinction matters for drone mapping because it removes the need for a stable radio link between base and rover during the mission. If you have ever flown an RTK survey and found patches of float solution in your trajectory because the radio link dropped behind a treeline or building, PPK eliminates that problem entirely. The raw data is all there in the logs, and post-processing fills in any gaps.

For mapping missions, this translates to more consistent fix solutions across the entire flight, and that means a more accurate orthomosaic at the end. PPK is the preferred method for most professional drone surveyors, particularly on sites with variable terrain, vegetation, or structures that can interrupt radio signals.

The trade-off is that you do not get real-time positioning feedback during the flight. You find out how good your data is after you land and run the processing. For most mapping missions, where the drone follows a pre-planned grid pattern, this is perfectly acceptable.

The Complete PPK Workflow at a Glance

A PPK drone mapping project follows a clear sequence of steps, each building on the previous one:

  1. Hardware setup: Configure your GNSS base station and rover, mount the receiver on the drone, and connect the camera trigger
  2. Ground control points: Place and survey GCPs across the site for independent accuracy verification
  3. Flight planning and execution: Plan the grid, set overlap parameters, fly the mission
  4. Data collection: Download raw logs from base and rover, collect the images
  5. Post-processing: Process the raw GNSS data to generate precise photo positions
  6. Geotagging: Apply corrected coordinates to the image EXIF data
  7. Photogrammetry: Process geotagged images into an orthomosaic, DSM, or 3D model
  8. Quality check: Verify accuracy against GCPs and project requirements

Each step is covered in detail below.

Hardware Setup: Building Your PPK Mapping Kit

At minimum, a PPK drone mapping system requires three GNSS components: a base station, a rover receiver mounted on the drone, and an antenna for the rover. You also need a method to synchronise the camera shutter with the GNSS position log.

The Base Station

The base station sits on a known point (or an unknown point that you will later determine through static processing) and logs raw GNSS observations throughout the flight. It does not need to communicate with the drone during the mission.

The Emlid Reach RS4 is a popular choice for the base role. It is compact, has multi-band support (GPS, GLONASS, Galileo, BeiDou), and logs for hours on its internal battery. The Reach RS3 also works well as a base if you already have one in your survey kit.

The Rover (Drone-Mounted Receiver)

The rover rides on the drone and logs raw satellite data during flight. For PPK, it records observations at a rate of 5 to 14 Hz (readings per second). A faster logging rate means more position fixes along the flight path, which helps match camera trigger events to precise coordinates.

The Emlid Reach M2 is a lightweight module designed for drone mounting. It connects to the drone's camera via a hot-shoe adapter, which detects the exact moment each photo is taken and logs a corresponding event marker in the GNSS data. This synchronisation between shutter and position is the most critical part of the hardware setup.

Camera Trigger Options

There are several ways to trigger the camera and synchronise it with the GNSS log:

  • Hot-shoe adapter: The most accurate method. The adapter sits between the camera and its flash shoe, sending a signal to the Reach module every time the shutter fires. This gives sub-millisecond timing accuracy.
  • Intervalometer: The camera takes photos at fixed intervals (for example, every 2 seconds). The Reach logs position at its update rate, and during post-processing the software interpolates the camera position from the nearest GNSS readings. Less precise than hot-shoe, but simpler to set up.
  • Software trigger via flight controller: The autopilot sends a trigger signal to both the camera and the Reach module over PWM or a digital protocol. Common on Pixhawk-based builds using ArduPilot or PX4.

For the best accuracy, use the hot-shoe adapter. If your camera does not have a hot shoe (many integrated drone cameras do not), the intervalometer or software trigger methods still produce good results when combined with a fast GNSS logging rate.

Antenna Placement on the Drone

Mount the GNSS antenna on top of the drone, as high as possible and away from the propellers, motors, and any metal components that can interfere with satellite signals. A clear sky view is the goal. Use a short mast or bracket if needed. The antenna should be mounted rigidly so it does not vibrate or shift position during flight.

Ground Control Points: Your Accuracy Benchmark

Ground control points (GCPs) are visible markers placed on the ground at known coordinates. They serve two purposes: providing an independent accuracy check on the final map, and in some workflows, helping to georeference the orthomosaic.

With PPK, your geotagged photos already have centimetre-level coordinates baked into their EXIF data, so GCPs are not strictly required for georeferencing. However, they remain the best way to verify that your processed data meets the accuracy standard the project demands.

For a typical site survey, place 5 to 10 GCPs distributed across the area, with at least one near each corner and a few in the interior. Use checkerboard targets, cross-shaped markers, or any high-contrast pattern that will be clearly visible in the aerial photos at your flight altitude.

Survey each GCP with a survey-grade GNSS receiver (such as the Reach RS4 Pro) using RTK or static observation to get its precise coordinates. Record these coordinates in a CSV file. You will use this file later in the photogrammetry software.

Flight Planning and Execution

A PPK mapping flight is planned the same way as any photogrammetry mission. The key parameters are overlap, altitude, and ground sampling distance (GSD).

Overlap Settings

Overlap determines how many photos cover each point on the ground. For photogrammetry to work well, you need sufficient overlap for the software to match features between adjacent images.

  • Front overlap: The overlap between consecutive photos along the flight line. 75 to 80 percent is standard.
  • Side overlap: The overlap between adjacent flight lines. 65 to 70 percent is typical, though 75 percent or more is better for dense vegetation or complex terrain.

Lower overlap means fewer photos and faster processing, but risks gaps in the point cloud, especially over homogeneous surfaces like water, sand, or uniform grassland. If in doubt, increase the overlap.

Altitude and GSD

Flight altitude directly controls the ground sampling distance, which is the real-world size of each pixel in the imagery. Lower altitude gives finer GSD but requires more photos to cover the same area.

For most survey-grade mapping, aim for a GSD of 2 to 5 cm per pixel. At this resolution, you can reliably measure features down to about 2 to 3 times the GSD, so a 3 cm GSD gives you roughly 6 to 9 cm positional accuracy in the final orthomosaic.

Before Takeoff

Before launching, confirm the following:

  • The base station is powered on and logging raw data
  • The rover on the drone is powered and logging
  • The camera is set to take photos at the correct interval or is connected to the trigger
  • The drone has enough battery for the planned mission plus a safety margin
  • The weather is suitable: minimal wind, good visibility, no rain

Record the exact start and end times of the flight. This helps during post-processing if you need to trim the GNSS logs to the relevant period.

Post-Processing: Turning Raw Logs into Precise Positions

After landing, you have three sets of data: raw GNSS logs from the base, raw GNSS logs from the rover (including camera event markers), and the folder of aerial images. The post-processing step combines the base and rover logs to calculate centimetre-accurate positions for each camera event.

Emlid Studio Workflow

Emlid Studio is a free desktop application (Windows and macOS) that handles the entire post-processing pipeline. It works with logs from any GNSS receiver, not just Emlid hardware.

The drone data processing workflow in Emlid Studio follows these steps:

  1. Convert logs to RINEX: If your raw files are in UBX or RTCM3 format, Emlid Studio converts them to the industry-standard RINEX format automatically. Drag and drop the files and press Convert.
  2. Add the rover RINEX file: Upload the RINEX observation file from the drone-mounted receiver. Set the antenna height to 0; the photogrammetry software will handle the offset between the antenna phase centre and the camera.
  3. Add the base RINEX file: Upload the base station observation file. If you placed the base over a known point, enter its coordinates manually. Otherwise, use the position from the RINEX header.
  4. Add the navigation file: Include the RINEX navigation file from either the base or rover. This contains satellite orbit data needed for processing.
  5. Process: Click Process and wait. The software calculates the precise trajectory of the rover and generates an events.pos file containing the exact coordinates for each camera trigger event.

The processing typically takes a few minutes for a standard flight. The results plot shows the rover trajectory colour-coded by solution quality. Green segments indicate a fixed solution (centimetre accuracy), while yellow or red segments indicate float or single solutions that may need attention.

What If You Do Not Have a Local Base Station?

If you do not have your own base station, you can use data from a nearby CORS (Continuously Operating Reference Station) instead. Download the RINEX observation and navigation files from a CORS within about 30 to 50 km of your site. Many national geodetic agencies provide this data free of charge. In the UK, the Ordnance Survey operates the OS Net CORS network.

Using a CORS extends the baseline between base and rover, which can reduce accuracy slightly, but within 20 km the effect is negligible for most mapping applications.

Geotagging: Applying Positions to Your Photos

Once you have the events.pos file from post-processing, the geotagging step writes the corrected coordinates into the EXIF metadata of each photo.

In Emlid Studio, this is straightforward:

  1. The events.pos file is automatically populated in the Geotagging section after processing
  2. Select the folder containing your drone images
  3. Click Tag Photos

Emlid Studio matches the number of events in the POS file to the number of images in the folder. If these counts do not match, the geotagging will not run. This mismatch usually indicates missing images, extra photos taken outside the mission, or a trigger synchronisation problem.

By default, Emlid Studio creates geotagged copies in a new folder, preserving your originals. If you prefer to update the originals in place, toggle the "Update original photos" option.

At this point, each photo now carries centimetre-accurate latitude, longitude, and altitude metadata. These geotagged images are the input for the photogrammetry software.

Photogrammetry: From Geotagged Images to Orthomosaic

The photogrammetry stage stitches the geotagged images into a seamless orthomosaic, generates a digital surface model (DSM), and optionally creates a 3D point cloud or mesh. Several software packages are commonly used:

  • Pix4Dmapper / PIX4Dmatic: Professional-grade, widely used in surveying and construction. PIX4Dmatic handles larger datasets faster.
  • Agisoft Metashape: Flexible and powerful, popular in research and academia.
  • WebODM: Open-source option, free to use, good for smaller projects or those on a budget.
  • Dronelink, DroneDeploy: Cloud-based platforms that bundle flight planning with processing. Convenient but require a subscription.

Processing Steps

Regardless of the software, the photogrammetry pipeline follows a similar sequence:

  1. Initial processing: The software analyses each image, detects feature points (keypoints), and matches them across overlapping photos. It builds a sparse point cloud and estimates camera positions and orientations.
  2. Point cloud densification: Using the matched features, the software generates a dense 3D point cloud representing the surveyed terrain.
  3. Mesh generation: The dense point cloud is used to create a triangulated mesh surface.
  4. Orthomosaic generation: The original photos are projected onto the mesh surface from above, correcting for perspective distortion. The result is a geometrically correct, georeferenced orthomosaic image.
  5. DSM/DTM generation: Elevation data from the point cloud is used to create digital surface and terrain models.

Adding GCPs for Accuracy Verification

Import your GCP coordinates (the CSV file from your field survey) and mark the corresponding GCP positions in the images. The software uses these as check points or control points. If the GCP positions in the orthomosaic match their surveyed coordinates within the expected tolerance (typically 1 to 3 times the GSD), your map passes quality control.

If accuracy is outside tolerance, common causes include incorrect antenna height settings, poor satellite geometry during the flight, or GNSS signal interference from nearby structures.

Common Issues and Troubleshooting

Fewer Fix Solutions Than Expected

If large sections of your trajectory show float or single solutions after post-processing, check these factors:

  • Satellite visibility during the flight. Heavy canopy, urban canyons, and low-elevation satellites reduce the number of usable signals.
  • Base station sky view. The base needs a clear view of the sky throughout the observation period.
  • Raw data quality. Review the observation files for gaps, cycle slips, or low signal-to-noise ratios.
  • Baseline length. If using a CORS, check that it is within a reasonable distance of your site.

Mismatched Event and Photo Counts

If the number of camera events in the POS file does not equal the number of images, check that:

  • No photos were taken before or after the GNSS logging period
  • The camera trigger connection is secure and did not miss any shutter events
  • Extra images (like handheld shots of GCPs) are separated from the mission photos

Poor Orthomosaic Quality

Blurry patches, gaps in the point cloud, or misaligned areas in the orthomosaic usually stem from insufficient overlap, motion blur from high flight speed, or lighting changes during the flight. Increase overlap and reduce flight speed to resolve these issues on the next flight.

Accuracy Expectations

With a properly configured PPK workflow using multi-band GNSS receivers, you can expect the following accuracy:

  • Horizontal (X/Y): 1 to 3 cm relative accuracy
  • Vertical (Z): 2 to 5 cm relative accuracy

Absolute accuracy depends on the accuracy of your base station position. If you place the base over a known survey point with accurate coordinates, absolute accuracy approaches relative accuracy. If you use a CORS or estimate the base position through static processing, absolute accuracy may be slightly lower.

For comparison, standard GPS (without any differential correction) typically achieves 2 to 5 metre accuracy. The improvement from PPK is roughly two orders of magnitude.

Who Should Use PPK Drone Mapping?

PPK drone mapping is used across a range of industries:

  • Construction: Progress monitoring, earthwork volume calculations, as-built surveys
  • Land surveying: Topographic surveys, boundary mapping, site planning
  • Mining and aggregates: Stockpile volume measurement, pit progression tracking
  • Agriculture: Crop health analysis, field boundary mapping, drainage planning
  • Environmental monitoring: Coastal erosion tracking, flood assessment, habitat mapping

If your project needs maps accurate enough for measurement, volume calculation, or regulatory compliance, PPK provides the positional precision those tasks require. For a deeper comparison of when to choose PPK over RTK, see our RTK vs PPK comparison guide.

Summary

The PPK drone mapping workflow transforms raw GNSS observations and aerial photos into centimetre-accurate maps. The key steps are setting up your hardware correctly (base, rover, camera trigger), flying with adequate overlap, post-processing the GNSS data in software like Emlid Studio, geotagging your photos, and running photogrammetry to produce the final orthomosaic.

Each step builds on the previous one, and quality at each stage directly affects the final result. Invest time in proper hardware setup and GCP placement, and the downstream processing will reward you with survey-grade outputs.

FAQ

Do I need a base station for PPK?

Not necessarily. You can use data from a nearby CORS network instead of your own base station. However, having your own base gives you more control over baseline length and data quality.

What is the minimum overlap for PPK mapping?

75% front overlap and 65% side overlap is the practical minimum. For complex terrain, vegetation, or water surfaces, increase both to 80% or more.

Can I use PPK with any drone?

Yes, as long as you can mount a GNSS receiver and connect a camera trigger. Many pilots use PPK with DJI, Autel, and custom-built drones. The Emlid Reach RX2 and M2 are lightweight enough for most platforms.

How long does post-processing take?

GNSS post-processing in Emlid Studio typically takes 2 to 5 minutes for a single flight. Photogrammetry processing varies widely depending on the number of images, your hardware, and the output formats required. A flight with 200 to 400 images might take 30 minutes to a few hours.

Is PPK more accurate than RTK?

In practice, PPK often produces slightly better results because it can use forward and backward processing of the GNSS data, filling gaps that would cause float solutions in real-time RTK. However, both methods achieve centimetre-level accuracy when set up correctly. For a full comparison, see our RTK vs PPK drone mapping guide.