Airborne ultra-wideband radar sounding over the shear margins and along flow lines at the onset region of the Northeast Greenland Ice Stream

. We present a high-resolution airborne radar data set (EGRIP-NOR-2018) for the onset region of the Northeast Greenland Ice Stream (NEGIS). The radar data were acquired in May 2018 with the Alfred Wegener Institute’s multichannel ultra-wideband (UWB) radar mounted on the Polar 6 aircraft. Radar proﬁles cover an area


Introduction
The Northeast Greenland Ice Stream (NEGIS) efficiently drains a large area of the Greenland Ice Sheet and is a crucial component of the ice sheet mass balance (Fahnestock et al., 1993;Rignot and Mouginot, 2012).It extends from the central ice divide over more than 600 km towards the northeastern coast, where it discharges ice through the three marineterminating glaciers (79 • North glacier, Zachariae Isbrae, and Storstrømmen glacier).The currently prevailing hypothesis is that an anomaly of elevated geothermal heat flux (GHF) leads to extensive basal melting (Fahnestock et al., 2001) and induces ice flow.The GHF at the onset of NEGIS is the most important and, at the same time, most uncertain parameter in the representation of basal melt and the subglacial hydrology in ice flow models (Smith-Johnsen et al., 2020).However, the large gap between observed and modeled basal melt rates (Gerber et al., 2021;Zeising and Humbert, 2021;MacGregor et al., 2016;Buchardt and Dahl-Jensen, 2007) inside the ice stream and in the broader surroundings of the NEGIS onset raises questions about the real thermal state at the bed.The exceptionally high GHF proposed by Fahnestock et al. (2001) would be non-compatible with known geological processes (Jóhannesson et al., 2020;Blackwell and Richards, 2004).Therefore, Bons et al. (2021) raise the question of how fast flow at the NEGIS onset is possible without an extraordinarily high basal heat flux and melt rates.
Unlike other ice streams in Greenland, the NEGIS lacks an extensive overdeepened bed and thus lateral topographic constraints (Joughin et al., 2001;Franke et al., 2020).The ice stream more or less symmetrically broadens along flow, as more ice is dragged through the shear margins (Fahnestock et al., 2001;Joughin et al., 2001).Subglacial water routing in combination with subglacial till deformation seems to be a further controlling mechanism of ice flow at the onset of the NEGIS (Keisling et al., 2014;Christianson et al., 2014).The outlet area is characterized by an overdeepened basin covered with unconsolidated sediments (Joughin et al., 2001;Bamber et al., 2013).Ice thinning as a consequence of increasing oceanic water temperatures around Greenland could thus potentially be transmitted far upstream (Christianson et al., 2014), and changes in the hydropotential might have significant effects on the ice stream geometry.The high susceptibility of NEGIS to marine-triggered discharge and the expected increase in ocean water temperatures in the years to come (Yin et al., 2011;Straneo et al., 2012) raises questions about the future ice stream stability and its effect on the Greenland Ice Sheet mass balance.
Large-scale ice flow models are essential tools to predict the future behavior of glaciers and ice sheets and are necessary to estimate future sea-level rise.Until today, these models fail to successfully simulate the NEGIS due to insufficient understanding of the key processes responsible for ice flow dynamics (Mottram et al., 2019;Shepherd et al., 2020), limiting the prediction accuracy.The East Greenland Ice-Core Project (EastGRIP; https://eastgrip.org,last access: 17 March 2021) aims to drill a deep ice core in the upstream area of the NEGIS, providing valuable insights into the climate record, basal properties, and ice flow history at the drill site.Ice cores provide in situ information on physical and chemical properties at high resolution but are limited as being a spatial point measurement.Hence, further geophysical techniques are required to extrapolate this information to obtain a complete picture of the ice dynamic properties.
We present airborne radar data of the onset region of NEGIS recorded in 2018 by a multichannel ultra-wideband radar system.The data set consists of profiles oriented parallel and perpendicular to the ice flow direction.The high along-track and range resolution allows consistent isochrone tracing, providing insights into the three-dimensional structure of the ice stream.In combination with the EastGRIP ice core, this data set contributes to a better understanding of ice flow dynamics of the NEGIS.In our paper, we introduce the study site and survey design.Furthermore, we describe the radar data processing and the respective data products.The data are freely available from the PANGAEA data publisher (https://doi.pangaea.de/10.1594/PANGAEA.928569;Franke et al., 2021a).

Survey region and previous work
In May 2018, we recorded radar data in the vicinity of the drill site of the EastGRIP ice core.An area of ∼ 24 000 km 2 was mapped, with 7494 km of radar profiles along flow lines and perpendicular to ice flow (Fig. 1).The survey region extends ∼ 150 km upstream and downstream of the EastGRIP drill site and ranges from the central part of the ice stream up to 50 km beyond the shear margins.In our survey region, the ice stream accelerates from ∼ 10 to more than 80 m a −1 and widens from ∼ 15 to ∼ 55 km.The radar data also cover the transition in the position of the shear margin as well as strongly folded internal stratigraphy outside of the ice stream (see Fig. 1). Figure 1b and c show the locations of radar profiles in relation to the ice surface velocity.Profiles extending perpendicular to ice flow have a spacing of 5 km in the region close to the drill site and 10 km further up-and downstream.Along-flow profiles either follow flow lines, which in some cases pass through the shear margins, or are constantly located inside the ice stream.Other profiles are oriented parallel to ice flow of the NEGIS but are located outside of the ice stream.
The ice thickness in our survey region ranges from 2059 to 3092 m and shows, on average, a gradual decrease in thickness from the upstream to the downstream part (Franke et al., 2020(Franke et al., , 2019).An analysis of the bed topography, basal roughness, and bed return echoes by Franke et al. (2021c) shows that our survey area can be divided into two different morphological regimes.The upstream part (upstream of East-GRIP) is characterized by a narrow ice stream width with accelerating ice flow velocity, a smooth bed with elongated flow-parallel subglacial landforms, and a soft till layer at the base (Christianson et al., 2014).Downstream of EastGRIP, the ice stream widens, and we note an overall change to a rougher and more variable bed geomorphology.The ice stream widens up to 57 km, and ice flow velocity keeps constant and decreases locally (Franke et al., 2021c).
In the 2012 summer field season, a scientific consortium collected ground-based geophysical data (RES and seismic survey) as well as a shallow ice core (Vallelonga et al., 2014).Christianson et al. (2014) examined the ice-bed interface by means of radar and seismic data analysis.They found highporosity, water-saturated till, which lubricates the ice stream base and most likely facilitates ice streamflow.Keisling et al. (2014) used the same radar data to analyze the internal radar stratigraphy and suggest that the basal hydrology controls the upstream portion of the NEGIS.By contrast, the downstream part is rather confined by the bed topography.Riverman et al. (2019a) and Riverman et al. (2019b) analyzed the shear margins in particular and observed an increased accumulation and enhanced firn densification in the upper ice column as well as wet elongated subglacial landforms at the bed.Furthermore Holschuh et al. (2019) use RES data to evaluate 3D thermomechanical models of the NEGIS.The authors highlight the complexity of the stagnant-to-streaming ice flow transition and provide insights into the englacial heat transport.Finally, a comprehensive chemical analysis of a 67 m deep firn core was conducted by Vallelonga et al. (2014).The results demonstrated that a deep ice core at this location has the potential to retrieve a reliable record of the Holocene and last glacial cycle.

Radar data acquisition
The ultra-wideband (UWB) airborne radar is a Multichannel Coherent Radar Depth Sounder (MCoRDS, version 5), which was developed at the Center for Remote Sensing of Ice Sheets (CReSIS) at the University of Kansas (Hale et al., 2016).It has an improved hardware design compared to predecessor radar depth sounders by CReSIS (Gogineni et al., 1998;Wang et al., 2015).The radar configuration deployed in 2018 consists of an eight-element radar array mounted under the Polar 6 Basler BT-67 aircraft's fuselage.The eight antenna elements function as transmit and receive channels using a transmit-receive switch.The total transmit power is 6 kW, and the radar can be operated within the frequency band of 150-600 MHz.The pulse repetition frequency (PRF) is 10 kHz, and the sampling frequency is 1.6 GHz.The characteristics of the transmitted waveform as well as the recording settings can be manually adjusted.We refer to the combined transmission-reception settings as waveforms in the following.
All profiles were recorded using linear frequencymodulated chirps in the frequency band of 180-210 MHz, antenna elements oriented with the E plane aligned with the along-track (HH polarization), and the transmit antenna beam pointed toward nadir.We used three alternating waveforms to increase the dynamic range of the system (see Table 1).Short pulses (1 µs) and low receiver gain of 11 dB are used to image the glacier surface, and longer pulses (3-10 µs) with higher receiver gain (48 dB) to image internal features and the ice base.Recorded traces were coherently pre-summed with zero-pi modulation in the hardware (Allen et al., 2005) to reduce the data rate and to increase signal-to-noise ratio (SNR), leading to a reduced effective PRF.The pre-sum factors were selected with regard to the pulse length of the respective waveform.To reduce range side lobes without losing much signal power, the transmitted and the pulse compression filter were amplitude-tapered using a Tukey window with a taper ratio of 0.08 (Li et al., 2013).
Before the data acquisition, the amplitude, phase, and time delay of the antenna elements were equalized during a test flight over open water during the transit to Greenland.Durhttps://doi.org/10.5194/essd-14-763-2022 Earth Syst.Sci.Data, 14, 763-779, 2022 ing data acquisition, the position of the aircraft was determined by four NovAtel DL-V3 GPS receivers, which sample at 20 Hz.The GPS system operates with dual-frequency tracking so that the position accuracy can be enhanced during post-processing.

Radar data processing
The acquired data comprised a total of 24 total radargrams, each consisting of a pairwise combination of 1 of 8 receivers and 1 of 3 waveforms.The post-flight processing goal was to create single radargrams of the profiles covering the ice sheet from surface to base with high SNR, fine resolution, and high dynamic range.The main processing included pulse compression in the range dimension, synthetic aperture radar focusing in the along-track dimension, and array processing in the cross-track and elevation-angle dimension.Lastly, we vertically concatenated the radargrams of the three waveforms.The post-processing tools are implemented in the CReSIS Toolbox (CReSIS, 2020b).At first, the recorded traces were pulse-compressed using a Tukey time domain weighting on the pulse-and frequencydomain-matched filtering with a Hanning window to reduce side lobes.For this purpose, the matched filter duplicates the transmitted waveforms based on the radar transmit settings.
Synthetic aperture radar (SAR) processing was carried out to focus the SAR radargrams in the alongtrack direction.The SAR processing is based on the fk (frequency-wavenumber) migration technique for layered media (Gazdag, 1978), which was adapted for radioglaciology (Leuschen et al., 2000).We used a two-layered velocity model with constant permittivity values for air ( air = 1) and ice ( ice = 3.15).We consider the constant value of ice = 3.15 to be justified since the uncertainties in the propagation velocity in the firn layer have a negligible effect on the fk migration.Furthermore, we use ice = 3.15 to be consistent with the value used for processing of Operation Ice Bridge (OIB) data.The air-ice interface was tracked using quicklook imagery, which is generated using 20 coherent averages followed by 5 incoherent averages by an automated threshold tracker.Platform motion compensation is applied during averaging.The SAR aperture length at each pixel was chosen to create a fixed along-track resolution of 2.5 m.A requirement for the fk migration is a uniformly sampled linear trajectory of the receivers along the SAR aperture extent.
To estimate the flight trajectory we used Novatel OEM6 receivers at 20 Hz data rate.The precise point positioning (PPP) post-processed accuracy (commercial software package Waypoint 8.4) is estimated to be better than 3 cm for lat-itude and longitude and better than 10 cm for altitude.INS data were acquired by the onboard laser gyro inertial navigation system (Honeywell LASERREF V).Its accuracy is given to be better than 0.1 • for pitch and roll and better than 0.4 • for true heading (Honeywell product description).Changes in aircraft elevation, roll, pitch, and heading lead to phase errors in the migrated data and thus to decreased SNR and blurring.Processed GPS and INS data in high precision from the aircraft were used to correct these effects.
The motion compensation consisted of (1) uniformly resampling the data in along-track using a windowed sinc interpolation, (2) fitting lines to the resampled trajectory with the length of the SAR aperture, and (3) correcting any flight path deviations from the straight lines with phase shifts in the frequency domain.
After the along-track focusing, the channels were combined to increase the SNR and reduce the impact of surface clutter and off-nadir reflections.The delay-and-sum method allows for steering of the antenna array beam.The antenna array beam is steered toward nadir by coherently summing the data from each channel while accounting for the actual position of each measurement phase center.A total of 11 along-track averages (multilooking) are then performed to reduce speckle in the imagery.
Finally, the different waveform images were vertically combined to increase the dynamic range of the result.The two-way travel time (TWT) at which the radargrams are combined were chosen with regard to the pulse durations of the transmitted waveforms and the surface return in order to avoid saturation of the high-gain channels due to the strong surface return.For three-waveform collection with 1, 3, and 10 µs nadir waveforms, image 2 is combined with image 1 after 3 µs after the surface reflection and image 3 with image 2 10 µs after the surface return (see Fig. 5).

Range resolution
The theoretical range resolution after pulse compression is where c denotes the speed of light in a vacuum, r the real part of the ordinary relative permittivity, B the bandwidth of the transmitted chirp, and k t the windowing factor due to the frequency and time domain windows.For the bandwidth of 30 MHz, the theoretical range resolution in ice with r = 3.15 and k t = 1.53 is 4.31 m.
In addition, to estimate the accuracy of a specific target (internal layer or bed return), we have to consider the RMS error in the dielectric constant (CReSIS, 2020a).Here we depend on the exact detection of the ice surface reflection, which is well constrained for our data.

S. Franke et al.: UWB radar sounding at the onset region of the NEGIS
To determine range resolution variability for the bed reflection for the radar data, we performed a crossover analysis of bed pick intersections (see Franke et al., 2020) and calculated the mean deviation h c .We consider an error on the order of 1 % for the dielectric constant for typical dry ice (Bohleber et al., 2012), with the ice thickness T and a mean value for crossover deviation, h c .The full analysis of the range resolution of the bed reflection is documented in Franke et al. ( 2020) and shows a variability from 13 to 17 m.

Bed return resolution
A further parameter is the size of the area illuminated by the radar wave in the bed reflection signal.Here we consider a cross-track resolution for a typical rough surface σ y .It is constrained by the pulse-limited footprint and depends on the Tukey and Hanning window parameters, where H is the elevation of the aircraft over the ice surface.
All EGRIP-NOR-2018 flights were performed at an aircraft elevation of ∼ 365 m above ground.When off-nadir clutter is visible in the radargrams, the cross-track resolution depends on the full beam width, β y , of the antenna array, where λ c is the wavelength at the center frequency, N is the number of array elements, and d y is the element spacing of the antennas.The cross-track resolution is now defined as where k y is the approximate cross-track windowing factor for a Hanning window applied to a small cross-track antenna array.
For areas without signal layover, the cross-track footprint ranges between 300-350 m for an ice column of 2000-3000 m in our survey region (Franke et al., 2020).Where layover occurs, we have to consider Eq. ( 5) with a beam angle of ∼ 21 • .Here the cross-track resolution ranges between 800 and 1100 m.

Results
The design of the EGRIP-NOR-2018 radar survey enables a detailed analysis of the bed and englacial stratigraphy:

Parameter Value
Covered area 25 000 km 2 Total profile distance 7494 km Along-track resolution ∼ 27-30 m (qlook) /∼ 15 m (SAR 1 ) /∼ 6 m (SAR 2 ) Range resolution 4.31 m Data amount 4.5 GB (qlook) /9.5 GB (SAR 1 ) /25.6 GB (SAR 2 )   (i) along radar profiles which are parallel to the ice flow (Fig. 1c) and (ii) along radar profiles perpendicular to the ice flow, crossing the shear margins (Fig. 1b).In addition, several radar lines cover the location where the northern shear margin shifts its location (Fig. 1a) and an area southwest of EastGRIP where we observe patterns of strong internal ice deformation (location indicated in Fig. 1a).
In flow-parallel profiles, we observe three regimes showing different characteristics in the radar stratigraphy.Figure 2 shows a ∼ 200 km long flow-parallel radargram extending from the far upstream end outside of the NEGIS up to 30 km downstream of EastGRIP.The area outside of the ice stream at a distance of 0-60 km shows slightly folded internal layers, which have no connection to the local basal topography.Steeply dipping internal layers are characterized by a decrease in reflectivity with depth.The part of the profile which is crossing the shear zone (at 60-90 km distance) is characterized by tight folds in the internal layers, which extend almost up to the ice surface.The folds' apparent wavelength in profile direction depends on the intersection angle with respect to the shear margin orientation.Folds are tightest in profiles oriented ∼ 90 • to the shear margin fold axis.We observe the best resolution of internal layers inside the ice stream, particularly in the lower part of the ice column.All distortions in the internal stratigraphy inside the ice stream seem to be related to the underlying bed.
Radar profiles oriented perpendicular to ice flow show a strong imprint of NEGIS' dynamics.The most striking features are tight folds in the area of the shear margins.Bright stripes characterize them in the radargrams (e.g., Fig. 3), which represent a loss in return power due to steeply inclined internal reflectors (Holschuh et al., 2014;Keisling et al., 2014).The onset of this kind of folding starts at the shear zone's outer boundary (marked by yellow triangles in Fig. 3c  and d).Depending on the ice stream's location and width, these folds can be traced towards the ice stream center for up to tens of kilometers, also towards locations where no shearing at the ice surface is observed.This becomes evident by comparing the upstream and farther-downstream radargram in Fig. 3c and d.
At several locations in flow-perpendicular radargrams we observe a drawdown of the radar stratigraphy towards the shear zone's outer margin (see Fig. 3c).The drawdown could be explained by temperate ice at the base (Franke et al., 2021c).Melting at the base at this location would be in agree-  2019a), who suggest that melt out of sediments within the ice column potentially creates the subglacial bedforms they identified.In general, the stratigraphy of internal layers northwest of the NEGIS differs from the stratigraphy southeast of the NEGIS.The northwest is much more undisturbed than the southeast.A distinct example is shown in Fig. 3a and b.The stratigraphy is marked by long-wavelength anticlines and synclines with elevation differences of almost half of the ice column.In the anticlines' cores, we find strong englacial reflections, which have been misinterpreted before as bedrock (Franke et al., 2020).We note that some of the englacial reflections appear to be attached to the basal reflection (Fig. 3a).This could be an indication that the strong englacial reflection could be due to basal material (e.g., soft sediments) that was transported upwards by the folding process.Figure 3b shows that the deformation patterns in the anticline cores are very complex.Some of the reflection patterns (e.g., in Fig. 3b) result in connected structures similar to the isochrones in the upper part of the ice column.In addition, we find isolated patches with high reflectivity, which, however, do not show a coherent pattern (highlighted in Fig. 3b).

Data products
We offer three different data products of the EGRIP-NOR-2018 radar survey: (i) quick-looks (qlook), (ii) SARfocused (SAR 1 ), and (iii) SAR-focused with a large aperture (SAR 2 ).The general data record properties are shown in Table 2, and detailed documentation can be found in the CReSIS MCoRDS documentation (https://data.cresis.ku.edu/, last access: 17 March 2021) and on the CRe-SIS Wiki website (https://ops.cresis.ku.edu/wiki/index.php/Main_Page, last access: 17 March 2021).In Fig. 4 we pro-vide an overview of the differences between these three data products of the radargrams.

qlook
This product uses unfocused synthetic aperture radar processing for each channel and assumes that all reflections arrive at the receiver from nadir.The data are coherently stacked in slow time, and no correction for propagation delay changes is applied.Here, no motion compensation is applied.Finally, the signals from all eight channels are averaged incoherently.The range resolution is the same as for all other products.The trace spacing is ∼ 27-30 m.

SAR with default settings (SAR 1 )
This data product uses focused synthetic aperture radar processing (fk migration) on each channel individually.The SAR processing requires a uniformly sampled linear trajectory along the extent of the SAR aperture.Motion compensation is applied using high-precision processed GPS and INS data from the aircraft.The direction of arrival is estimated by delay-and-sum beam forming to combine the channels.A Hanning window is applied in the frequency domain to suppress side lobes.This product is comparable to the CReSIS standard data product.The trace spacing is ∼ 15 m.

SAR with wider angular range (SAR 2 )
By processing at a finer SAR resolution, the SAR processor uses scattered energy from a wider angular range around nadir to form the image.Since the angle of scattered returns from a specular internal layer is proportional to the internal layer slope, the SAR processor's increased sensitivity to larger angle returns translates to an increased sensitivity to layers with larger slopes.We achieve a better resolution of https://doi.org/10.5194/essd-14-763-2022 Earth Syst.Sci.Data, 14, 763-779, 2022 steeply inclined internal reflectors by changing the alongtrack resolution before SAR processing to 1 m (≈ σ x = 1, whereas the default setting is σ x = 2.5); 1 m is not the smallest possible value for processing but is within the limit to achieve a sufficiently high SNR.The SNR is smaller for larger angles because the range to the target increases for greater angles, which leads to additional signal loss (spherical spreading loss and additional signal attenuation in ice).
The differences between radargrams processed with σ x = 1 and 2.5 are shown in Fig. 4. The final trace spacing is ∼ 6 m.

Individual waveforms
The combination of the three waveform images will increase the dynamic range of the whole radargram.However, specific analyses may require only a single waveform.Therefore, we provide the respective echogram data for each waveform separately (see Fig. 5).The files are labeled with img_01, img_02, and img_03 for image 1, 2, and 3, respectively.

Image mode flights
In addition to the data recorded in the so-called sounding mode, the data set also contains two segments (20180510_02 and 20180515_01) recorded in the image mode (see Table A1).The acquisition settings of these segments are slightly different since the transmission signals are composed of four instead of three pulses: 1 and 3 µs waveforms with nadir-directed transmit beam followed by two 10 µs waveforms, one with the transmit beam directed to the left and one with it directed to the right, to increase the imaged swath width at the ice bottom.We processed the data in a way that the 10 µs left and 10 µs right signal are steered towards nadir during data processing.This is possible because both waveforms contain nadir information since the beams overlap at nadir.However, the reflection power from nadir is reduced because the two side-looking transmit beams have reduced gain by about 3-4 dB relative to when the beam is pointed directly at nadir.The third waveform (see Fig. 5c and d) in the combined image for the two particular segments has been computed from the coherent combination of both the 10 µs left and 10 µs right return signals.

Data formats
We provide the main radar data and auxiliary data in the following formats: 1.The radar data containing a matrix of the echogram and the corresponding GPS information, such as coordinates, aircraft elevation, and timing of every trace, are stored as matfiles (HDF-5 based format).The echograms are provided for the combined waveform product as well as for the individual waveforms.Furthermore, these files contain cell arrays with information about all processing parameters used.3. A set of shapefiles (lines) containing the location of every frame are provided.
4. An Excel is provided which contains all parameter settings applied during radar data processing.
In the Appendix we describe the data architecture and how the data are stored in the PANGAEA repository.

Relevance of the data set
This radar data set provides essential observations of internal and bed reflections to determine spatial distribution of ice thickness, internal layering, and reflectivity.These observables constitute boundary conditions and elucidate properties and processes of the NEGIS.The tightly spaced survey lines allow information on the past and present ice flow dynamics to be derived, revealing the paleo-processes of the NEGIS, in particular when combined with ice core data.Hereby it might be possible to address the question of how long the NEGIS has been active in its present form.The radar data furthermore allow a systematic analysis of the return power https://doi.org/10.5194/essd-14-763-2022Earth Syst.Sci.Data, 14, 763-779, 2022 of internal layers and the bed and could potentially reveal information about the englacial temperature regimes and properties of the bed and basal hydrology (e.g., Franke et al., 2020Franke et al., , 2021c) ) as well as provide crucial boundary conditions for ice flow models (Gerber et al., 2021).In addition our data hold the potential to map the crystal orientation fabric (COF) and especially the horizontal anisotropy from the birefringence effects (Eisen et al., 2021;Young et al., 2021), e.g., at the shear margins or at intersecting radar profiles of different polarization directions.Because ice stream regimes have a strong effect upon the COF (Lilien et al., 2021), it is another important, but yet poorly constrained parameter in ice flow modeling.
The data comprise ∼ 7500 km high-resolution radar data for the scientific community to use.In contrast to previous surveys in this area, the data presented here were specifically recorded over a broad spatial extent to understand the history of the NEGIS and our general understanding about ice streams.The tightly spaced radar profiles perpendicular to ice flow allow a 3D interpretation of the ice-internal stratigraphic architecture.Because most radar lines are directly or indirectly spatially connected to the location of the drill site of the EastGRIP ice core, these data are significant for various objectives regarding the ice dynamic understanding of NEGIS as well as interpretation of the climate proxy record retrievable from the ice core.With this data set the scientific community will be able to upscale the findings of the East-GRIP project from the location of the ice core to the immediate surroundings of the upstream part of the NEGIS.The prospect that parts of the ice core can be rotated back into their correct geographic direction (Westhoff et al., 2020) also allows a systematic analysis of ice crystal orientation fabric together with the radar data.Furthermore, the data presented here can be combined with the radar data acquired during Operation Ice Bridge (OIB) to extend the large-scale understanding of glaciological properties in the Greenland Ice Sheet.A1 of the EGRIP-NOR-2018 data set.The respective segments are highlighted with a white line and the radar profiles of the complete survey with a finer black line.The background map represents ice surface velocity from Joughin et al. (2017).The shear margin is indicated with a dashed white outline and the location of EastGRIP with a red dot.

Figure 1 .
Figure 1.Location of the EGRPI-NOR-2018 survey area in northeastern Greenland with the MEaSUREs ice surface velocity from Joughin et al. (2017) in the background.(a) The radar survey lines are centered at the EastGRIP drill site (white triangle) and extend up to 150 km upstream and downstream of the NEGIS.The locations of one along-flow radar section (Fig. 2; blue) and three cross-flow radar sections (Fig. 3; red) are shown in panel (a).The ice flow direction of the ice stream is indicated with a white arrow.Three-dimensional images of the cross-flow profiles and along-flow profiles are shown in (b) and (c), respectively.The radar sections are shown with a vertical exaggeration factor of z = 10.The ice surface velocity (Joughin et al., 2017) is shown on a logarithmic scale (log10) for all of Greenland (upper left image) and on a linear scale for panels (a), (b), and (c).For panels (b) and (c) the ice surface velocity is projected on top of the bed topography model from Franke et al. (2020).The projection for all maps is WGS 84/NSIDC Sea Ice Polar Stereographic North (EPSG:3413).

Figure 2 .
Figure 2. Flow-parallel radar profile A composed of four frames: 20180512_01_001-004.The location is indicated in Fig. 1a and covers the area outside of the ice stream, the shear zone, and the ice stream's trunk.The position of the EastGRIP drill site is labeled and indicated with a vertical dashed white line.

Figure 3 .
Figure 3. Radar profiles C 1 −C 1 , C 2 −C 2 , and C 3 −C 3 (frames 20180508_06_003, 20180511_01_007, and 20180514_01_011-012, respectively).The corresponding locations are shown in Fig. 1a.Panel (a) shows a profile located outside of the NEGIS, southeast of EastGRIP.A close-up of an anticline and other patterns of deformation are shown in panel (b), resembling the skeleton of a dead penguin.A slice through the ice stream in the upstream region is shown in panel (c) and a radar section further downstream in (d).The position of the shear margins (the maximum in the surface velocity gradient) is indicated with a yellow triangle and the folds in the shear margin areas with purple lines.The ice flow direction is into the page.

Figure 4 .
Figure 4. Radargrams of a flow-perpendicular radar profile showing the three different data products: (a) quick-look (qlook) processed, (b) SAR-focused (SAR 1 ), and (c) SAR-focused with a larger aperture (SAR 2 ).The main difference between qlook and SAR 1 is the focusing of signals via fk migration (d, e).In contrast to the default SAR focusing (f) a larger aperture enables a better resolution of steeply inclined internal layers (g).
2. A set of figures for each profile, showing the radargram and its respective location in the EGRIP-NOR-2018 survey, are provided as png images.

Figure 5 .
Figure 5. Radargrams of the frame 20180517_01_008 subdivided into (a) waveform 1, (b) waveform 2, and (c) waveform 3 and (d) the radargram composed of all three waveforms in combination.The locations where the different waveforms are concatenated are indicated with red lines in (d).Waveforms 1 and 2 are plotted at full range.Note the different resolution of the near-surface stratigraphy in waveforms 1 to 3.

Figure A1 .
Figure A1.Radar profile locations of the 12 segments shown in Table A1 of the EGRIP-NOR-2018 data set.The respective segments are highlighted with a white line and the radar profiles of the complete survey with a finer black line.The background map represents ice surface velocity from Joughin et al. (2017).The shear margin is indicated with a dashed white outline and the location of EastGRIP with a red dot.
* Pre-sums are set for each waveform individually.

Table 2 .
Data record properties.