The geography department of the university of Innsbruck has a long tradition in research concerning soil erosion in the Alps. One test site is located in the "Schmirntal", a small and steep valley about 20km south of Innsbruck.
During the winter 2011/12, the snow conditions probably caused an increased number of so called "blaiken", small areas in which grass and underlaying soil are eroded. Since the latest ortho images were dating back to 2010, the project leader at the university approached GRID-IT with the idea to try a UAV flight, taking aerial images in order to produce an ortho image data set as basis for change detection.
Since GRID-IT is focussing exactly on these topics (UAV-based earth observation and change detection in the alpine terrain), and since this idea promised to be an interesting and challanging project, we soon agreed on a flight during the last week in July. On Thursday, the weather was unstable, so we had to postpone for one day. But Friday was very promising, and therefore we started early in the mornign at about 5:30am, to avoid the afternoon winds and thermals.
We could cover most of the altitude with a Jeep, only the last 200-300m had to be done by foot. The Quest200 was packed into a rucksack, the wings mounted separately. We also took a high precision GPS and colored markers to be used as ground reference points.
Close to the ridge, we found a relative flat area, where we could setup our equipment. The area still had a 5-10% slope was covered with a series of grass humps. The slope left us only one direction for takeoff and the opposite way for landing. The Quest200 was put together easily and within half an hour, we were ready to get airborne.
The QuestUAV launch system is very safe and almost nothing can go wrong with it. But still, knowing that due to topography, aborting the launch was not an option, added some stress to it. But as expected, the launch went well without any problems and soon enough we had the beautiful sight of a Quest200 flying in front of a lovely mountain scenery.
We had defined a flightplan with seven legs and a total length of about 8km. The flight path was covered exceptionally well, only some thermals and gusts pushed the UAV away from its track. We flew about 100m above launch, which gave sufficient height for the surrounding topography (we had a 1m digital elevation model and could check the height above ground for the flight plan beforehand).
Landing the UAV on that humpy upward slope was a bit tricky, but we managed to bring the bird down safely. Images and logger data were checked immediately and both showed good results. So we decided to try another flight, this time at lower altitude to get better ground sampling distance (the altitude difference from lowest to highest point was about 500m). Again, launch was easy - this time, we did some manual flying to get nice pictures before switching to AUTO for commencing the flight plan. However, the first approach towards the first leg was to steep and brought the UAV too close to the surface, so we aborted AUTO, flew the approach in ASSISTED and tried again. But again, we had the feeling that the UAV was getting to close to the ground, so we aborted the mission and went to landing the UAV as we did before. This time, we had some tail wind (which was getting stronger every minute) and during the approach, the motor block of the Quest200 was damaged due to the uneven ground. With the damage and the wind and thermals getting stronger, we decided to cancel the second flight, since we already got excellend data from our first flight.
After all, it was a very interesting day with lovely weather and we got some very nice data, which we are going to process during the following days. We hope to come up with a good ortho image and a detailled digital surface model, which we could compare with the LIDAR based elevation model. The main lesson learned was that much consideration needs to be given to the right choice of a landing site, and we will think of some other concepts to support landing in difficult terrain.