Thawing-induced cliff top retreat in permafrost landscapes is mainly due to thermo-erosion. Ground-ice-rich permafrost landscapes are specifically vulnerable to thermo-erosion and may show high degradation rates. Within the HGF Alliance Remote Sensing and the FP7 PAGE21 permafrost programs we investigated how SAR and optical remote sensing can contribute to the monitoring of erosion rates of ice-rich cliffs in Arctic Siberia (Lena Delta, Russia).We produced two different vector products:i) Intra-annual cliff top retreat based on TerraSAR-X (TSX) satellite data (2012-2014):High-temporal resolution time series of TSX satellite data allow the inter-annual and intra-annual monitoring of the upper cliff-line retreat also under bad weather conditions and continuous cloud coverage. This published SAR product contains the retreating upper cliff lines of a 1.5 km long part of eroding ice-rich coast of Kurungnakh Island in the central Lena Delta. The upper cliff line was mapped using a thresholding approach for images acquired in the years 2012, 2013 and 2014 for the months June (2013, 2014), July (2013, 2014), August (2012, 2013, 2014) and September (2013, 2014). The cliff top retreat vector product is called 'upper_cliff_TerraSAR-X'. While the 2014 cliff lines show a clear retreat of 2 to 3 m/month, the cliff top lines for 2012 and 2013 are not chronologically ordered. However, lines from the end of the season of a year are always close to the lines from the beginning of the next summer season, indicating low cliff retreat in winter.ii) 4-year cliff top retreat based on optical satellite data (2010-2014): Long-term cliff top retreat could be assessed with two high-spatial resolution optical satellite images (GeoEye-1, 2010-08-05 and Worldview-1, 2014-08-19). The cliff top retreat vector product is called 'upper_cliff_optical'. Results:The long-term cliff top retreat derived from optical satellite data are 35 m cliff retreat within 4 years. The higher-temporal resolution SAR data equivalently show long-term rates of 18 m within 2 years and nearly now degradation activities in winter but maximum erosion rates in summer months.The Intra-seasonal cliff top retreat lines from 2014 show a rate of 2 to 3 m per month.

Technical details for the TSX processing:For the TSX preprocessing steps, multilooking and terrain correction, the open source software NEST from ESA was used. To improve image quality a simple multilooking using two looks was performed for every image of the time series. Because the eroding coast is observed over a three-year time span and no Digital Elevation Modell is available for each of the acquisition dates, instead an ellipsoid-based correction was performed with a geolocation grid and nearest neighbor resampling with full spatial resolution of 2.5 m in the UTM Map projection WGS84 Zone 52 North.Technical details for the optical satellite data processing:Orthorectification of the 2 optical satellite images on full spatial resolution in the UTM Map projection WGS84 Zone 52 North was performed using the software module OrthoEngine of PCI Geomatica 2014. In the rational functions model in total 7 ground control points (GCPs) and 13 Tie Points were set with an overall GCP RMS of 0.33 meters. GCPs were collected during Expedition "Lena2013" in summer 2013 with a Leica Viva GNSS in real time kinematic - GPS mode.The satellite-derived cliff top retreat product contains the vector files of all discrete cliff top lines in UTM Map projection WGS84 Zone 52. While the 10 cliff top lines from TSX within the study area were digitized in 1:1000 scale, the 2 cliff top lines from the optical reference were digitized in 1:500 scale. There is a lateral shift between the cliff top lines derived from TSX versus the ortho-rectified optical product due to different terrain correction (about 57 meters lateral shift).