Sparnum · Moose survey

How a drone moose survey is flown

The mission file is calculated, not drawn by hand. Every number on this page is a parameter in our software, not an aspiration.

What a mission package contains

One package per drone model, as a zip. Inside are four .kmz files per sample plot in DJI's WPML format, one for each route, a map of the sample plots, a field sheet in Swedish as html and pdf, and a list of the files with checksums, so it shows if anything is missing or has been changed.

With the package comes the order of visits, how the visits fall across field days, and a suggested launch point for each sample plot. The suggestion is just that, and the final choice is made on site from access, visibility and a clear climb. If a plot cannot be reached it is not flown, and it is never swapped for another plot, because the hard-to-reach plots may well be the ones holding the most animals.

How the routes are built

The parameters follow SLU Rapport 2026.2 wherever the report gives them, with one exception: we photograph more densely along the line than SLU did, so that every point is seen in at least three photos. It costs no flight time, only more photos, and gives more chances to see an animal between the trees. Height, speed, overlap and plot size are the same in the calculation as in the routes below, so what is written here is what is flown. Each drone model gets its own mission file, calculated from that sensor's field of view, so the same sample plot is flown differently by different models.

Map of six squares in a real assignment. Two of them are drawn with a route, and where a flight line reaches water it ends and crosses without taking photos.
Six squares from a real assignment, with land cover and water beneath. The labels in the figures on this page are in Swedish. The routes are drawn on two of them. Where a flight line meets water it ends, and the crossing to the next shore carries no photography. Boundary: Länsstyrelserna, CC BY 4.0. Elevation data: Markhöjdmodell Nedladdning, ©Lantmäteriet, processed, CC BY 4.0.

The route does not stop without reason. The drone turns between flight lines without braking, and where a line crosses water or land outside the property it flies straight over instead of stopping at the edge. Every stop costs braking, turning and accelerating again. On a plot like the one bottom centre in the figure we expect the flight time to be 21–30 % shorter.

Each sample plot gets four routes, two directions and two starting sides. In calm weather the shortest is flown. If it is windy, the pilot chooses on the day the one that runs across the wind and starts downwind, so the wind helps the drone home when the battery is at its lowest.

A sample plot of 50 hectares, with 41.52 hectares of land and the rest water, built as a model of its own ground in contour lines every two metres, and the route of fifteen flight lines, 48.52 metres apart with a swath width of 64.69 metres, laid over the terrain.
A real sample plot, built as a model of the ground beneath it. The route is drawn from the same calculation as the mission file, so another drone model gives both a different line spacing and a different picture. The ground is exaggerated in height so the contours show, but the 100 m up to the route is not, so the route sits closer to the ground in the figure than in reality. Boundary: Länsstyrelserna, CC BY 4.0. Elevation data: Markhöjdmodell Nedladdning, ©Lantmäteriet, processed, CC BY 4.0.
Another sample plot of 50 hectares, with 39.69 hectares of land and the rest outside the area, built as a model of its own ground in contour lines every two metres, and the route of fifteen flight lines, 48.52 metres apart with a swath width of 64.69 metres, laid over the terrain. A power line crosses the plot, and the route flies over it.
A different sample plot from the six above, chosen because a power line crosses it. The line does not exclude the plot. It is flagged on the field sheet and the plot is flown as usual, which is why the route runs straight over the zone in the figure. How high the line hangs is not in the source data, so the zone's height in the figure is an assumption and not a figure for this particular line. The plot lies in Väddö älgskötselområde. Boundary: Länsstyrelserna, CC BY 4.0. Elevation data: Markhöjdmodell Nedladdning, ©Lantmäteriet, processed, CC BY 4.0. Power line: Lantmäteriet (Topografi 50), CC0.
  1. 100 m above the ground, not above the launch pointHeight is measured against the ground directly beneath the drone, not against the place it took off from, so it is the same down in a hollow as up on a hill. The route stays within five metres of 100 m above ground. Counted from the launch point it would follow the launch point instead: twenty metres too high and it hits the ceiling, twenty-five metres too low and gaps open between the photos. The drone flies at 9 m/s with the camera pointing straight down, 90 degrees, and every photo is taken with two cameras, a thermal camera and an ordinary one. The photos overlap 25 % sideways and 67 % forward, so that no ground falls between them. The spacing between flight lines is calculated from the camera that will actually fly, so another drone model gives a different line spacing and a different flight time.
  2. Sample plots of 50 haThe same size SLU calculated its spread on, so the measures can be compared. Edge squares are clipped to the area boundary, and a square with less than half land does not count as a sample plot.
  3. A quarter are flown againThe repeat flights are in the package as files of their own. They measure the detection rate, and without them the whole survey is a lower bound of unknown depth.
  4. Take-off and finishThe drone climbs straight up to 50 m before heading for the first waypoint, and flies home when the route is done. On loss of link the default is to return, which suits flying within visual line of sight. If the flight is beyond visual line of sight under a permit of its own, the procedure that permit requires is written into the file instead.
A sample plot in terrain near the mountains with a height difference of 233.7 metres, from 678.7 to 912.4 metres above sea level, where the route follows the ground up and down instead of holding a fixed height.
The same calculation on steep ground. The ground falls 233.7 m within this sample plot, from 912.4 to 678.7 m above sea level, and the route follows it all the way instead of holding a fixed height above the launch point. That is the difference between 100 m above the ground and 100 m above the point where the drone took off. The plot lies in Glöte älgskötselområde in Härjedalen. Boundary: Länsstyrelserna, CC BY 4.0. Elevation data: Markhöjdmodell Nedladdning, ©Lantmäteriet, processed, CC BY 4.0.

What we remove before anything takes off

Every sample plot is checked against five sources before it goes into a package. What is impossible to fly is removed as soon as the grid of plots is laid. What can be flown with a permit or with care goes along as a flag on the field sheet, with the name of the area, so no one has to look it up themselves.

  1. AirspaceRestricted areas and control zones (CTR) from LFV. An area whose remark bans drones, or refers to the Swedish Transport Agency's special permit, is excluded. An area that states who grants permission goes along as a flag with that information. An area that only requires a NOTAM is flagged as such.
  2. Protected areasNational parks, nature reserves and areas closed to access from Naturvårdsregistret. National parks are excluded. Reserves are flagged, because their regulations differ from one to the next.
  3. Wind powerBuilt turbines from Vindbrukskollen, with total height. A square with a turbine of 70 m or more within 150 m of the sample plot is excluded, because the route sits at 100 m. Turbines still under review are flagged instead.
  4. Tall obstaclesPower lines, masts, towers and chimneys within 150 m of the sample plot. They do not exclude a plot, they are flagged, because the route at 100 m clears pole heights of twenty to thirty-five metres. A launch point is never suggested closer than 50 m to a power line's centre line, though, or 150 m to a mast, a tower, a chimney or a wind turbine.
  5. WaterLakes, sea and watercourses from the map data. Water is subtracted from a square's land area, and a square left with less than half land does not become a sample plot.
A sample plot that is not flown. The ground is almost flat, but on it stands one built wind turbine of 210 metres, and the 150 metre zone around the turbine covers ground that would otherwise have been flown, so the square is drawn without a route.
A sample plot that never goes into a package. The ground falls 12.5 m across the whole plot, from 198.5 to 186.0 m above sea level, so it is not the terrain that stops the flight. The turbine is built and 210 m tall, and the route would have been at 100 m, so the square is removed when the grid is laid instead of being flagged. That is why it is drawn without a route. The ground is exaggerated 6.0 times in height for the contours to show at all, but the turbine's 210 m is not. The plot lies in Blädinge älgskötselområde in Kronoberg County. Boundary: Länsstyrelserna, CC BY 4.0. Elevation data: Markhöjdmodell Nedladdning, ©Lantmäteriet, processed, CC BY 4.0. Wind turbine: Länsstyrelserna (Vindbrukskollen), CC0.

This does not replace the check on the day of flying. UAS zones and temporary restrictions exist only in Drönarkartan, LFV's drone map, and it is read before every flight. Our source data is dated and kept with the assignment, so it is always possible to see how old it was when the route was calculated.

Every file is read back before it leaves us

Once a .kmz has been written, the software opens it again and checks the contents against DJI's format and against our own parameters. A file that cannot be read back is not sent.

The whole pass is then measured from the files themselves: ground heights, the deviation from 100 m, flight time, climb and descent, and how much of the flight lines lies over water. If a file is missing, has been changed, or the flight time does not match the list, the package does not go out.

The field sheet: the flags and the checklist the pilot carries

The field sheet carries the sample plot's flags and a checklist that belongs to that particular assignment. It says, among other things, that Drönarkartan is to be read before every flight, that a plot in a control zone needs air traffic control's permission every time, that the landowner is to be notified, that no flight happens over land where hunting is under way the same day, and that no one is ever told where the moose are while the flight is under way.

It also says that the flight is called off if the sun comes out. SLU flew in overcast weather to get the greatest possible thermal contrast between animal and ground, and that detail decides how much the survey actually sees.

Read on about moose surveys

Contact

hej@sparnum.com

Do you have a drone model we have not calculated for, or want to check the parameters against your own data? Get in touch.