What do I need to bear in mind when flying over terrain?

Flying in hilly terrain is not always easy. As a rule, an altitude of 50 metres is sufficient to fly comfortably above the treetops. However, this is not the case in hilly terrain. When flying down a slope, it is easy to crash into a tree.

However, it is not only the risk of collision that makes it sensible to create a mission plan adapted to the terrain. Overlap and pixel ground resolution also change on slopes. These are parameters that are very important, for example, for reliably detecting wildlife.

Current drones can determine their own altitude using four sources:

• Barometric sensor
• GNSS receivers (GPS, Glonass, Galileo, etc., and also more accurately with RTK)
• Distance sensors (laser, radar, ultrasonic)
• mittels Kameras über einen photogrammetrischen Ansatz ( Stereo Vision, SfM)

The most reliable and one of the simplest sensors for measuring altitude is the barometric pressure sensor. Air pressure decreases constantly with altitude. The only problem with this sensor is when there are fluctuations in air pressure, for example when gusts of wind occur or the weather changes suddenly. Electronic barometers are very sensitive to temperature, which is why good barometers are usually temperature-stabilised. A huge advantage of barometers is their very large measuring range, which is usually from -600 to 9000 m.

Without RTK, GNSS receivers (GPS and similar) are very poor altitude measuring devices. Depending on the constellation of satellites, an absolute altitude measured with GPS can easily vary by 20 metres. With RTK, we are in the single-digit centimetre range.

Good laser distance sensors with sufficient range are quite expensive, which is why they are usually only installed in higher-priced drones. Ultrasonic sensors are only suitable for close range (centimetres to a few metres). Radar sensors are not yet widely used.

With normal CMOS cameras, height can only be measured in sufficient lighting conditions. This type of height measurement requires considerable computing power. Unlike laser distance sensors, cameras can measure not only a single distance, but also capture an entire topographical terrain in great detail with a single measurement.

There are four relevant altitude reference systems for drone pilots:

• Height above the official country-specific quasi-geoid (in Germany: DHHN2016)
• Height above the ellipsoid (e.g. WGS84)
• Height above the geoid (e.g. EGM96)
• Height above the drone’s take-off point

The altitude data in maps from European surveying authorities is always given in metres above the reference surface of the respective country-specific quasi-geoid. This reference surface is usually based on the average height above sea level in the immediate vicinity of the respective country. In Germany, this reference surface is called Normalhöhennull (NHN).

The GPS receiver in the drone usually indicates the altitude in terms of height above the ellipsoid. This is normally always the WGS84 ellipsoid. An ellipsoid is the three-dimensional equivalent of an ellipse, a flattened sphere, so to speak. The WGS84 ellipsoid is a mathematically very simple approximation of our Earth.

However, due to its inhomogeneous gravitational field, the Earth looks more like a potato than a sphere. Since the sea surface also aligns itself with the gravitational field, attempts are made to simulate this gravitational field more accurately using suitable geoid models. A frequently used geoid model is called EGM96.

By default, a drone flies at a constant altitude above the take-off point. However, the take-off point here does not refer to the manually set home point, but to the actual take-off point. This is where the motors were switched on and the drone actually took off. The drone usually uses its built-in barometric sensor to measure altitude. This is set to 0.0 m when the motor is switched on. The drone then automatically adjusts itself so that the set barometric target altitude is maintained constantly. In DJI’s Pilot2 app, this altitude mode is called ‘Altitude Relative to Take-off Point (Altitude)’. However, no altitude adjustment takes place.

DJI has named a mode in the Pilot2 app ‘ü. NHN (EGM96)’. This refers to the normal height zero surface of the EGM96 geoid. In other words, the altitude is not height above ground, but height above sea level. Unfortunately, the Pilot 2 app does not offer terrain adjustment in mapping mode (area route). Therefore, this is simply the constant height above normal zero elevation.

When transferring elevation data from official maps, it is important to note that German land surveying authorities currently use DHHN2016 instead of the EGM96 geoid. This could result in an elevation difference of up to one metre. However, an even greater error occurs if RTK is not used. Depending on the satellite position, this can result in a deviation of up to 20 metres above or below.

Now we come to the actual terrain mode. With the help of a digital elevation model (DEM), terrain variations at individual waypoints can be taken into account. However, the barometer is not taken into account here. For the drone, this information is always given as heights above the ellipsoid. If you now import a digital elevation model from surveying offices as a local file, you must add the geoid undulation at this point in addition to the conversion to WGS84. The use of AsterGdem data (when you click on DSM file->Download from the Internet in the Pilot2 app) is sometimes very inaccurate (during my last flight, the altitude was 26 m too low).

With RTK and a good DEM, everything is fine with this mode. However, if you don’t have RTK and/or a good DEM, terrain-adapted flying based on this principle is extremely inaccurate. If you plan to fly at an altitude of 50 metres above ground level but then actually only fly at 25 metres, this is very critical!

A much more accurate variant is the new ‘real-time tracking’ method. This is a photogrammetric process. The lower cameras for obstacle detection are used to calculate the terrain using a Structure from Motion (SfM) algorithm. However, the M3T requires a minimum illuminance of 15 lux.

This terrain mode works without a lot of sensor technology, but is still very accurate. Unfortunately, it is not available as standard with DJI. The altitude data is always given in relation to the starting point. Some flight planning programmes (e.g. QGroundControl or UAV Editor) use digital elevation models from the internet and calculate the respective height difference between the starting point and the corresponding waypoints, then simply add the desired flight altitude. Without RTK, laser or camera-based real-time altitude adjustment, this type of terrain flight is the most accurate compared to the others using elevation models.

However, it is important to bear in mind that a spontaneous change in the starting point automatically changes the height of the drone above the terrain.