New instruments for the Corvara Landslide: Permanent DSLR cameras and C-band corner reflectors

Recently, new instruments to improve the monitoring of the landslide were installed on the Corvara test site:
Six C-band corner reflectors were mounted in order to make use of ESA’s recently launched Sentinel 1 mission. In addition, three permanent DSLR cameras were installed on the Corvara study site. The cameras are programmed to shoot two pictures a day to have a continuous monitoring of the landslide. Two systems are powered by a solar panel and one is fed by household current.

UAV flight at the Corvara Landslide

After 2015 and 2016, on 21 and 22 June 2017 the third UAV flight on the Corvara landslide has been carried out successfully. The data is currently processed to generate a orthomosaic as well as a 3D point cloud and to compare it with the previous acquisitions.


The LEMONADE team contributed to EGU 2017 (Vienna Austria) with two presentations about landslide monitoring using close-range and remote sensing techniques… have a look:


Rutzinger, M., Zieher, T., Pfeiffer, J., Schlögel, R., Darvishi, M., Toschi, I., Remondino, F. (2017): Evaluating synergy effects of combined close-range and remote sensing techniques for the monitoring of a deep-seated landslide (Schmirn, Austria). In: Geophysical Abstracts. Vienna, Austria, Vol. 19 (EGU2017-6393-3). Download poster


Schlögel, R., Darvishi, M., Cuozzo, G., Kofler, C., Rutzinger, M., Zieher, T., Toschi, I., Remondino, F. (2017): Sentinel-1 and ground-based sensors for a continuous monitoring of the Corvara landslide kinematic (South Tirol, Italy). Vol. 19 (EGU2017-12913). Talk



Second survey campaign in Fortebuso

A second survey campaign in Fortebuso was realized in October 2016 over an over of approximately 150 x 100 m. The work included a UAV field campaign coupled with Total Station and GNSS survey in order to measure the position of several targets both inside and outside the landslide. The image block includes 232 images with a mean GSD of about 8 mm. Among the surveyed ground points, 6 points were included in the photogrammetric processing as GCPs whereas 6 points were used as cjeck points (CPs) to evaluate orientation accuracy. RMS errors on CPs resulted 8 mm (x), 11 mm (y) and 7 mm (z, vertical). The dense image matching extracted a point cloud with a mean spatial resolution at GSD level (around 8 mm). With the support of the Geological Service, the areas supposed to be “active” according to the periodic monitoring activities were identified. Then vegetation (trees) was filtered out and the cloud better aligned with the output of the previous campaign (June) by applying an ICP algorithm on the stable points of the area. Further analyses are now in progress to understand landslide movements.

Find more pictures from the Fortebuso test site in the gallery.


3D Reconstruction of a Large Landslide from UAV-Based Imagery

yprsThe LEMONADE project has been presented at the Young Professionals Conference on Remote Sensing hosted by the IEEE Geoscience and Remote Sensing Society (IEEE-GRSS) and by the German Aerospace Center (DLR) in Oberpfaffenhofen (Germany). Mehdi Darvishi talked about first project results giving a presentation about 3D Reconstruction of a Large Landslide from UAV-Based Imagery.


UAV flight at Corvara Landslide

In the frame of the LEMONADE project, EURAC – Applied for Remote Sensing Institute performed the UAV flight campaign on the 23.06.2016 aiming to produce high resolution orthomosaics, 3D point clouds as well as Digital Surface and Terrain Models.

The UAV was flying exactly the same area, considered as the most active part of the Corvara landslide (i.e.  13 hectares on air) covered in 2015. The original flight plan consists of five different flights, each of them 200 m long, 80 m wide and 60 m height.

Unfortunately, during the flight campaign of this year we faced technical difficulties because we were not able to transmit the flight plans into the automatic flight controller mounted on the UAV (control system that follows a pre-programmed GPS point route). Since one single field campaign takes a lot of effort, organization and time, the pilots decided to flight in manual mode, i.e. controlling elevation, orientation and speed during each flight. The general consideration that we agreed was to keep (i) the height of 60 m and (ii) the straight lines with the support of one observer standing in the opposite side of the landslide. The digital camera was configured to acquire one image every 2 seconds according to a flight speed of 5 m/s what gave us some images every 10 m approximately. In order to retrieve reliable results and decrease the effect of the manual flight mode, four flights were done using the same orientation and one in the orthogonal orientation.



We were able to collect 1190 images that were preliminary pre-processed to confirm the overlapping coverage. This task consisted on filtering images, converting images from raw to tiff (using Rawterapee software), assigning position to each image, and finally doing the first processing step on Pix4D software (e.g. keypoints and coarse cloud point generation and first derived digital surface model).