Geological Survey of Denmark and Greenland Bulletin 41, 2018, 79-82 79 The Greenland ice sheet has experienced an average mass loss of 142 ± 49 Gt/yr from 1992 to 2011 (Shepherd et al. 2012), making it a significant contributor to sea-level rise. Part of the ice- sheet mass loss is the result of increased dynamic response of outlet glaciers (Rignot et al. 2011). The ice dis- charge from outlet glaciers can be quantified by coincident measurements of ice velocity and ice thickness (Thomas et al. 2000; van den Broeke et al. 2016). As part of the Programme for monitoring of the Green- land Ice Sheet (PROMICE; Ahlstrøm et al. 2008), three air- borne surveys were carried out in 2007, 2011 and 2015, with the aim of measuring the changes in Greenland ice-sheet thicknesses. The purpose of the airborne surveys was to col- lect data to assess the dynamic mass loss of the Greenland ice sheet (Andersen et al. 2015). Here, we present these datasets of observations from ice-penetrating radar and airborne laser scanning, which, in combination, make us able to determine the ice thickness precisely. Surface-elevation changes be- tween surveys are also presented, although we do not provide an in-depth scientific interpretation of these. Instrumentation All three surveys were conducted using the same Air Green- land/Norlandair De Havilland DHC-6 Twin Otter aircraft, currently registered as TF-POF. This Twin Otter has been modified in such a way that part of the fuselage can be re- moved in the rear cargo hole providing an unobstructed view of the surface below the aircraft when airborne. The precise position of the aircraft (and instruments) is tracked by three geodetic dual-frequency GPS receivers each connected to one of two GPS antennas mounted on top of the aircraft. The orientation of the instruments is monitored by an iner- tial navigation system (INS). The primary INS is of the type Honeywell H-764G. During the last two f lights, we also in- stalled a back-up INS of the type OxTS Inertial+2. For measuring snow- or ice-surface elevations, a near in- frared, airborne laser scanner (ALS; Forsberg et al. 2001) was mounted in the rear cargo hole, alongside the INSs. The ALS f lown on the Twin Otter in 2007 was of the type Riegl LMS-Q140i-60, which was upgraded to a Riegl LMS-Q240i in 2011 and 2015. In 2007 and 2011, a 60 MHz coherent ice-penetrating radar, developed at the Technical University of Denmark (DTU), was also mounted to measure bedrock topography (Christensen et al. 2000). Survey design The survey f light path was designed as a polygon to encircle the entire Greenland ice sheet where the surface of the ice Circum-Greenland, ice-thickness measurements collected during PROMICE airborne surveys in 2007, 2011 and 2015 Louise Sandberg Sørensen, Sebastian B. Simonsen, René Forsberg, Lars Stenseng, Henriette Skourup, Steen Savstrup Kristensen and William Colgan 2 km B 800 m Ice surface Bedrock Transmit pulse 70.35 70.34 70.33 70.32 70.31 70.30 30.920 30.910 30.900 30.890 2400 2392 2384 2376 2368 2360 2352 2344 A La tit ud e (n or th ) Longitude (west) El ev at io n (m ) Fig. 1. A: Example of full-resolution versus reduced-resolution (circles) airborne laser scanner (ALS) data. B: Example of radargramme with a clear bedrock ref lector. © 2018 GEUS. Geological Survey of Denmark and Greenland Bulletin 41, 79–82 . Open access: www.geus.dk/bulletin http://www.geus.dk/bulletin 8080 is at an elevation of c. 1700 m above sea level, as well as to include survey lines over the centerline of several main outlet glaciers. The surveys have been carried out at four-year inter- vals (2007, 2011 and 2015). The planned f light path in 2007 left a data gap on the east coast (from c. 72N to c. 74N) which was bridged during the 2011 and 2015 surveys. All three surveys were planned to be carried out in Au- gust, as this timing represents the end of the melt season and ensures that the changes observed in surface elevations are not affected by individual accumulation events. Due to bad weather conditions in August 2015, half of the survey (the part from Constable Pynt in East Greenland clockwise to Kangerlussuaq in West Greenland) was carried out in Octo- ber. The late acquisition of these data thus results in a poten- tial bias of individual accumulation events due to snowfall in this dataset compared to the surveys in 2007 and 2011. As the f light path from 2011 was repeated in 2015, and since the bedrock elevation is not expected to change within this time frame, it was decided not to utilise the ice-penetrating radar on the last survey in 2015. Surface-elevation data The ALS operates in the near-infrared wavelength band, which is ref lected from the snow or ice surface. This means that data can only be acquired during periods without clouds or fog below the aircraft. The sampling frequency of the ALS instrument is 10 kHz, resulting in 40 across-track scan lines per second. Each of these scan lines consists of 250 individual elevation measurements on-ground. The scan angle of 60° and the typical f light height of c. 300 m result in a swath width on ground of c. 300 m with c. 1 m resolution. The processing of the ALS data combines the raw ALS data with the positioning data from the GPS and altitude data from the INS. Post-processing of the data includes visual inspec- tion to filter out laser ref lections from clouds. The positional uncertainty in both latitude and longitude is estimated to be ± 1 m, while the elevation uncertainty is estimated from track cross-over differences to be ± 0.05 to 0.1 m over f lat surfaces. To reduce the file size and to create a dataset which is more comparable to the resolution of the bedrock data, the full-resolution ALS data have been reduced to a spatial reso- lution of c. 100 m. This has been done through simple averag- 30°W50°W 60°N 65°N 70°N 75°N 80°N 80°N 30°W50°W 60°N 65°N 70°N 75°N 80°N 80°N N –500 –250 0 250 500 750 1000 1250 1500 Bedrock elevation (m) 3000 2500 2000 1500 1000 500 Surface elevation (m) 2011 500 km 2011 500 km Fig. 2. A: Surface elevations along the PROMICE circum-Greenland f lights in 2011. B: Bedrock elevations along the PROMICE circum-Greenland f lights in 2011. 81 ing of available height measurements along and across track. An example of full-resolution versus reduced-resolution data is shown in Fig. 1A. The data are compiled in one file per year (ALS_ yyyy.ave) and can be downloaded from http:// promice.dk/DownloadAirborne.html. As an example, the el- evations from 2011 are shown in Fig. 2A. Bedrock-elevation data The ice-penetrating data acquisition consists of transmitting pulses at a pulse repetition frequency of 10 kHz (i.e. sampling in the f light direction) and sampling the returned echo at 75 MHz, which results in 4096 samples per transmitted pulse. While internal scattering masks the desired echo, ref lection and absorption within the ice sheet reduce the strength of the returned echo. Substantial processing is therefore carried out to produce a radargramme that enhances the detection of the echo from the bottom of the ice-sheet. A semi-automatic layer detection program is used to digital- ise the surface and bedrock layers individually. In some areas, primarily near the ice margin in South Greenland, the radar was not able to detect the bedrock due to heavily crevassed ice or water present within the ice. Figure 1B shows a good exam- ple of a radargramme where a bottom echo was obtained. Based on radar system setup, vertical uncertainty in radar-derived ice- sheet bed elevation is estimated to be ± 35 m, which is con- firmed by the cross-over differences between the two surveys. The data are compiled in one file per year (ARS_ yyyy. ave), which is also available for download from http://pro- mice.dk/DownloadAirborne.html. As an example, the bed- rock elevations from 2011 are shown in Fig. 2B. Surface-elevation changes Having three surveys of surface elevations spanning eight years enables us to derive and analyse surface elevation changes along the f light lines. In Fig. 3, we show the mean annual surface-elevation changes between August 2007 and August/October 2015. The map was generated by comput- ing height differences between any points in the two (re- duced resolution) datasets for the two years. Height differ- ences are computed only if the points are located not more than 200 m apart. By knowing the exact date of the survey, the rate of surface-elevation change can be computed. In the map in Fig. 3, the part that was only f lown in 2007 is plotted with black, while the parts only surveyed in 2015 are shown in grey. There are some clearly visible gaps: One leg of the f light line is missing in north-eastern Greenland from Ni- oghalvfjerdsfjorden to Hagen Bræ and similarly and a part of the line is also absent south of Jakobshavn Isbræ. The gap in the north is caused by gaps in the 2015 dataset due to time and weather constraints. The gap south of Jakobshavn Isbræ is due to cloud cover in 2007. Figure 3 shows that the mean annual elevation changes in the period 2007–2015 is clearly dominated by thinning with some main outlet glaciers such as Jakobshavn Isbræ and Kangerlussuaq Gletscher thinning rapidly. Only a few places along the f light line are associated with thickening, e.g. at Storstrømmen. The sections of the f light path in the south- eastern parts that actually show modest thickening might be a result of accumulation since these parts of the 2015 survey were mapped in October after some snowfall in the area. Elevation change data, such as presented here, are scientif- ically very valuable e.g. to validate satellite data and ice-sheet models. Furthermore, the data presented here represens an important supplement to the heights and height differences 2 1 2.0 1.5 1.0 0.5 0.0 –0.5 –1.0 –1.5 –2.0 Surface eleva- tion change (m/yr) 30°W50°W 60°N 65°N 70°N 75°N 80°N 2007–2015 500 km 80°N 1 4 7 6 3 5 Fig. 3. Mean annual surface-elevation changes between 2007 and 2015 along the PROMICE circum-Greenland f light-paths. The part of the f light track for which only 2007 data are available is indicated in black, while 2015-only is indicated in grey. 1: Nioghalvfjerdsfjorden. 2: Stor- strømmen. 3: Constable Pynt. 4: Kangerlussuaq Gletscher. 5: Kangerlus- suaq. 6: Jakobshavn Isbræ. 7: Hagen Bræ. http://promice.dk/DownloadAirborne.html http://promice.dk/DownloadAirborne.html http://promice.dk/DownloadAirborne.html http://promice.dk/DownloadAirborne.html 8282 available from the NASA Operation IceBridge field surveys (Krabill et al. 2009; Krabill 2014) as the f light lines cover different areas, and also our measurements are made at the end of the melt season while Operation IceBridge data are collected mainly in the spring. Comparison to BedMachine v3 bedrock elevations The bedrock elevation dataset described above also represents a valuable legacy dataset that can be used by a wider scientific community. Knowledge of bedrock elevations in Greenland is essential in, e.g. ice-discharge studies and ice-sheet model- ling. One widely used bedrock topography model is the one available in BedMachine v3 (Morlighem et al. 2017) which is based on the conservation of mass and constrained by avail- able measurements. The BedMachine v3 model is provided together with an error map, which shows how the error in- creases with increasing distance to measurement points. To evaluate whether the PROMICE dataset can potentially contribute to an improvement of the BedMachine model in the future, we have extracted the BedMachine error values for all the 2007 and 2011 bedrock elevations in the PRO- MICE datasets. The two corresponding histograms in Fig. 4 show that in c. 50% of the data locations the error in the BedMachine v3 model is greater than 100 m, indicating that the PROMICE dataset with an uncertainty of ± 35 m could indeed contribute positively to a future, improved version of the model. It may also be noted that only 25% of the BedMa- chine data are related with similar or lower errors than the PROMICE dataset. Acknowledgements This is a publication in the framework of The Programme for Monitoring of the Greenland Ice Sheet (PROMICE) – a Danish government initiative funded through the Danish Cooperation for Environment in the Arctic (DANCEA). References Ahlstrøm, A. & the PROMICE team 2008: A new programme for moni- toring the mass loss of the Greenland ice sheet. Geological Survey of Denmark and Greenland Bulletin 15, 61–64. Andersen, M. et al. 2015: Basin-scale partitioning of Greenland ice sheet mass balance components (2007–2011). Earth and Planetary Science Letters 409, 89–95. Christensen, E.L., Reeh, N., Forsberg, R., Jørgensen, J.H., Skou, N. & Woelders, K. 2000: A low-cost glacier-mapping system. Journal of Gla- ciolog y 46, 531–537. Forsberg, R., Keller, K. & Jacobsen, S.M. 2001: Laser monitoring of ice elevations and sea-ice thickness in Greenland. International Archives of Photogrammetry and Remote Sensing 34, 163–168. Krabill, W.B. 2014: IceBridge ATM L2 Icessn Elevation, Slope, and Roughness. Boulder, Colorado, USA. http://nsidc.org/data/ilatm2. html (NASA Distributed Active Archive Center at the National Snow and Ice Data Center). Krabill, W.B. et al. 2009: Operation Ice Bridge =verview and results from aircraft laser altimetry. American Geophysical Union, Fall Meeting 14–18 December 2009. San Francisco: Abstract 3 pp. Morlighem, M. et al. 2017: BedMachine v3: Complete bed topography and ocean bathymetry mapping of Greenland. From multibeam echo sounding combined with mass conservation. Geophysical Research Let- ters 44, 11051–11061, http://dx.doi.org/10.1002/2017GL074954 Rignot, E., Velicogna, I., van den Broeke, M.R., Monaghan, A. & Len- aerts, J.T. 2011: Acceleration of the contribution of the Greenland and Antarctic ice sheets to sea level rise.  Geophysical Research Letters  38, L05503. Shepherd, A. et al. 2012: A reconciled estimate of ice-sheet mass balance. Science 338, 1183–1189. Thomas, R.R., Akins, T., Csatho, B., Fahnestock, M., Gogineni, P., Kim, C. & Sonntag, J. 2000: Mass balance of the Greenland ice sheet at high elevations. Science 289, 426–428. van den Broeke, M.R., Enderlin, E.M., Howat, I.M., Kuipers Munneke, P., Noël, B.P.Y., van de Berg, W.J., van Meijgaard, E. & Wouters, B. 2016: On the recent contribution of the Greenland ice sheet to sea level change. The Cryosphere 10, 1933–1946. http://dx.doi.org/10.5194/tc- 10-1933-2016 C ou nt s 0 100 200 300 400 500 2011 2007 Error (m) 25 000 20 000 15 000 10 000 5000 0 30–40 m 40–100 m >100 m <30 m 24.4 8.8 50.0 16.8 Fig. 4. Histograms showing the errors of the BedMachine bed topography grid in all the points where PROMICE bedrock elevation data are avail- able. The grey area shows the <35 m interval (uncertainty in the bedrock data). The pie chart shows to what extent the BedMachine model error is 0–30 m, 30–40 m, 40–100 m and more than 100 m. Authors’s addresses L.S.S., S.B.S., R.F., L.S. & H.S., Technical University of Denmark, DTU Space, Geodynamics department, DK-2800 Kongens Lyngby, Denmark. E-mail: slss@space.dtu.dk. S.S.K., Technical University of Denmark, DTU Space, Microwave & remote sensing department, DK-2800 Kongens Lyngby, Denmark. W.C., Geological Survey of Denmark and Greenland, Øster Voldgade 10, DK-1350 Copenhagen K, Denmark. http://nsidc.org/data/ilatm2.html http://nsidc.org/data/ilatm2.html http://dx.doi.org/10.1002/2017GL074954 http://dx.doi.org/10.5194/tc-10-1933-2016 http://dx.doi.org/10.5194/tc-10-1933-2016 mailto:slss@space.dtu.dk