A simple wind measurement control loop was written for the Bebop2 quadcopter. It tries to keep the roll angle at zero to align with the wind. It works well if there is a certain wind speed and clear, non-turbulenced wind. The wind speed is going to be determined later from the pitch angle.

The SODAR is set up. An array of loudspeakers pushes a very loud, audible series of pulses into the air and receives the response from the atmosphere with microphones.

The wind direction and wind speed can be determined based on the reflections at the air layers (having a different temperature).

At noon a cloud layer starts to cover the sky. It starts to snow in the afternoon. During supper we see that the weather should stay like that the rest of the week. There is some chance that a time window might open again during the night.

The measurements require a stable boundary layer. Clear skies without clouds and cold air would be ideal. We prepare for flights in the night, but in the end we don’t. In between we watch the rocket launch.

Funny ice

It is hazy all day. The lighthouse disappears in the fog and you can not see the horizon.

Everything is covered with fine grown ice crystals. There is a jingling noise when running in it.

The 10m mast is put up and the instruments are connected and tested. The anchoring is done by wooden sticks that quickly freeze in drilled holes.

We fly some low circles with the SUMOs in the late afternoon. The moisture immediately settles as ice on the propeller and the leading edge of the wing.

The performance of the aircraft decreases slightly with the changed surface of the wings and the propeller. We land before it is critical. How nice that there is de-icing for man-carrying planes.


The last storm piled up the sea ice on the beach in up to 4m high walls, unfortunately. This bothers when starting and landing the unmanned aircraft and changes the meteorological conditions.

The measurement equipment has to be installed way out on the ice.

The mast will be erected tomorrow morning.


We are back in northern Finland (a report from the first trip will follow below). This is part of a measurement campaign over the sea ice of the Gulf of Bothnia (the northern continuation of the Baltic Sea), next to the island of Hailuoto.

Last year we could drive over the ice with the car. This year it is too weak (25cm instead of the required 40cm). We take the free ferry. 3MW of diesel-electric power push it through the broken ice. Only 30cm of water are under the keel at some points. You can see the lights from an icebreaker on the horizon. It is waiting to help ships loaded with wood from Oulu.

The unfriendly weather in Germany called for a de-icing of the aircraft. Here in the north it is crisp cold at -12°C. Just as winter must be.

One More

Leaving for the station to do some final experimental flights!


We do some tests with the quadcopter. There was not enough time (and daylight) besides the SUMO flights to try new stuff.


On our way home to UNIS we stop at the beach and make a final flight with four profiles over sea/land. There is a low hanging cloud. The humidity sensor is saturated during the entire flight. The PT1000 temperature sensor shows an erratic and significantly lower air temperature of -12°C instead of the -4°C the humidity sensor measures in the lower 150m altitude. We have a close look at the flight performance but there is nothing strange to see.


After landing we notice that the propellers leading edge and the wing tip is covered with clear ice.


There was a lot of wind last night. The snow in the surrounding mountains is blown away. There were heavy winds up to 16m/s in 100m altitude. The wind was lower at higher altitudes, only 10m/s in 300m. Nevertheless we stopped flying as no good scientific results were to be expected.


It is the last outdoor day. The groups go into the valleys to collect the weather stations.

Vimeo: Scientific unmanned aircraft over Svalbard

This is a video from spring 2014. It was three weeks later into the year and the sun was up higher. The SUMO flies figures-of-eight automatically at about 100m above the station through Adventdalen. Take-off and landing is manual.

In valleys

Adventdalen is the main valley, the northern lights station is located in the center and the airport at the end towards Isfjorden. Longyearbyen is located in a side valley (map).

Launch Endalen

The airport is closed on Saturday and we can go earlier. The first objective is to fly multiple profiles in the narrow Endalen. There are only 100m distance between the flight path and the valley at some points. We head to it with the rumbling belt wagon.


SUMO flies into the valley passing several waypoints and then does profiles on its way back from behind. The plane disappears behind laterally hillocks from time to time while flying low. Very unusual. Normally you would not fly a model airplane visually into terrain. It all works smoothly.

Adventdalen Fjord

We spend the rest of the evening at Adventalens beach for sea/land profiles. There is a weather change so it will be very interesting flights. There were westerly winds from the sea the last days. Now it switches to colder easterly winds from the land. The air masses interact at the water border, slide over each other creating a sharp boundary, mixing slowly.


The day ends with grilled sausages and salmon.


In the evening we use the belt wagon to drive to the brink of the transition between ice and water in Adventdalen to measure profiles between sea and land. All four tracks of that vehicle are driven. The steering happens through the hydraulic change of the angle between the front and rear cabin. It is uncomfortably loud inside during the ride, worse in the front as the engine is not really damped.

However, the cabin has the distinct advantage of an easy installation of the 5-hole probe and the fast temperature sensor during the flight preparation. Reading the logger and reprogramming after flight is also a lot easier then on snow scooters.


In general there is little precipitation on Svalbard, it is more of an Arctic desert. Today there are thick snowflakes. We have not yet flown under such conditions. We fear that the fine 12µm platinum wire of the rapid temperature sensor might brake, but we try it anyway.

Fine Wire

The wire actually breaks during the third profile just before the descent from 500m. It seems that there was no direct influence from the aircraft. The engine was constantly running at half throttle and there were no strong actuator movements.

The downward facing infrared radiation sensor measures a significantly lower temperature in high altitudes than the soil temperature measured at low altitude. It could be that the raw flakes are colder during the fall as they come from higher air layers.

First Flights

This time the focus of the measurements is on the heat exchange between the air, the land and the sea water. One group rides their sensor-equipped snow scooters through the valleys, one group goes out on the fjord with a small ship and gathers wind data, air and water temperatures, one operates the balloon at the station and another one flies with the SUMO. Every now and then data from the loggers of the various stations are read out.

Adventdalen Fjord

We received a permission to fly further out and higher up this time. Vertical flights are a lot more interesting than horizontal ones. We do profiles from 50m-500m in 1km distance in Adventdalen between the station and the shore line.