Search for: All records

This paper presents an analysis of the bed detecting capabilities of an ice sounding radar integrated onto a small, unmanned aircraft system (UAS). We evaluated the average signal-to-noise ratio (SNR) and signal-to-interference ratio (SINR) of radar measurements collected by the UAS over Greenland’s Helheim Glacier in 2022 and compared those to radar measurements collected over the same region using a radar-equipped Twin Otter around 2008. The statistical analysis presented of the SNR and the SINR shows that both systems have comparable bed detection capabilities. While the average SNR for all points considered is more than 20 dB higher for the Twin Otter system, the average SINR of both has a similar value. The overall average SINR is 9.79 dB for the UAS and 9.19 dB for the MA. As it is discussed in the paper, the lower SNR of the UAS system is attributed to its lower operating frequency, while the comparable SINR depends on various factors. The results of this paper have implications on planning and design of future field deployments. 

Abstract Airborne radar sensors capture the profile of snow layers present on top of an ice sheet. Accurate tracking of these layers is essential to calculate their thicknesses, which are required to investigate the contribution of polar ice cap melt to sea-level rise. However, automatically processing the radar echograms to detect the underlying snow layers is a challenging problem. In our work, we develop wavelet-based multi-scale deep learning architectures for these radar echograms to improve snow layer detection. These architectures estimate the layer depths with a mean absolute error of 3.31 pixels and 94.3% average precision, achieving higher generalizability as compared to state-of-the-art snow layer detection networks. These depth estimates also agree well with physically drilled stake measurements. Such robust architectures can be used on echograms from future missions to efficiently trace snow layers, estimate their individual thicknesses, and thus support sea-level rise projection models. 

<p>NSF COLDEX performed two airborne campaigns from South Pole Station over the Southern Flank of Dome A and 2022-23 and 2023-24, searching for a potential site of a continuous ice core that could sample the mid-Pleistocene transition. Ice thickness data extracted from the MARFA radar system has allow for a new understanding of this region.</p> <p>Here we generate crustal scale maps of ice thickness, bed elevation, specularity content, subglacial RMS deviation and fractional basal ice thickness with 1 km sampling, and 10 km resolution. We include both masked and unmasked grids.</p> <p> The projection is in the SCAR standard ESPG:3031 polar stereographic projection with true scale at 71˚S.</p> <p>These geotiffs were generated using performed using GMT6.5 (<a href="https://doi.org/10.1029/2019GC008515">Wessel et al., 2019</a>) using the pygmt interface, by binning the raw data to 2.5 km cells, and using the <a href="https://github.com/sakov/nn-c"> nnbathy </a> program to apply natural neighbor interpolation to 1 km sampling. A 10 km Gaussian filter - representing typical lines spacings - was applied and then a mask was applied for all locations where the nearest data point was further than 8 km. </p> Ice thickness, bed elevation and RMS deviation @ 400 m length scale (<a href="http://dx.doi.org/10.1029/2000JE001429">roughness</a>) data includes the following datasets: <ul> <li> UTIG/CRESIS <a href="https://doi.org/10.18738/T8/J38CO5">NSF COLDEX Airborne MARFA data</a></li> <li> British Antarctic Survey <a href="https://doi.org/10.5285/0f6f5a45-d8af-4511-a264-b0b35ee34af6">AGAP-North</a></li> <li> LDEO <a href="https://doi.org/10.1594/IEDA/317765"> AGAP-South </a></li> <li> British Antarctic Survey <a href="https://doi.org/10.5270/esa-8ffoo3e">Polargap</a></li> <li> UTIG Support Office for Airborne Research <a href="https://doi.org/10.15784/601588">Pensacola-Pole Transect (PPT) </a></li> <li> NASA/CReSIS <a href="https://doi.org/10.5067/GDQ0CUCVTE2Q"> 2016 and 2018 Operation Ice Bridge </a> </li> <li> ICECAP/PRIC <a href="https://doi.org/10.15784/601437"> SPICECAP Titan Dome Survey </a> </ul> <p>Specularity content (<a href="https://doi.org/10.1109/LGRS.2014.2337878">Schroeder et al. 2014</a>) is compiled from <a href="https://doi.org/10.18738/T8/KHUT1U"> Young et al. 2025a </a> and <a href="https://doi.org/10.18738/T8/6T5JS6"> Young et al. 2025b</a>.</p> <p>Basal ice fractional thickness is complied from manual interpretation by Vega Gonzàlez, Yan and Singh. </p> <p>Code to generated these grids can be found at <a href="https://github.com/smudog/COLDEX_dichotomy_paper_2025"> at github.com </a></p> 

These transect projected radargrams were collected as part of the Center for Oldest Ice Exploration (COLDEX) Science and Technology Center (https://www.coldex.org) in the 2022/23 (CXA1) and 2023/24 (CXA2) airborne field seasons. The raw 3 TB data is deposited at the USAP data center at https://doi.org/10.15784/601768. The set of images in this archive was designed for easy, non expert, access to radargrams, organized according to survey design. <p> The science goal was to characterize the ice sheet between Antarctica's Dome A and Amundsen Scott South Pole Station, to locate sites of interest for the drilling of an ice core with ages spanning the mid-Pleistocene. The radar was deployed on Balser C-FMKB, and flown at ranges of up to 800 km from South Pole Station at velocities of 90 m/s and typical altitude above ground of 600 m. Other instruments included a UHF array system provided by the University of Kansas, a gravity meter, a magnetometer, a laser altimeter, and multiple global navigation satellite systems receivers. The radar data is used for finding ice thickness, bed character, englacial structure and surface assessment. <p> <b>Dataset organization</b> Transects are provided a P/S/T nomenclature, organized by the Project they are flying in, the acquisition System (typically named after the aircraft) and the Transect within the Project. <p> Transects were collected in preplanned systems with the following parameters (examples below): <p> <i>The CLX radials</i> (CLX/MKB##/R###), attempting to emulate flow lines from Dome A and radiating (in the EPSG:3031 polar stereographic projection) from easting 965 km northing 385 km, with a separation of 0.25 degrees. <p> <i>The CLX corridor</i> (CLX/MKB##/X###) rotated from the EPSG:3031 polar stereographic projection at -150 degrees and separated by 10 km in the Y direction and 3.75 km in the X direction <p> <i>The CLX2 corridor</i> (CLX2/MKB##/X###) rotated from the EPSG:3031 polar stereographic projection at -150 degrees and separated by 2.5 km in its Y direction and 2.5 km in its X direction <p> <i>The NPXE radials</i> (NPXE/MKB##/R###) radiating (in the EPSG:3031 polar stereographic projection) from easting 0 km and northing 0 km (ie South Pole), with a separation of 2 degrees. <p> <i>The SAD corridor</i> (SAD/MKB##/X###|Y####) designed to characterize the Saddle region near South Pole approximately perpendicular to the flow lines, rooted from the EPSG:3031 polar stereographic projection at -73.8 degrees and separated by 2.5 km in its Y direction and 2.5 km in the its X direction <p> <i>Untargeted transit lines</i> used the name of the expedition (CXA1) as the project, and used the flight and the increment within the flight to name the Transect (eg (CXA1/MKB2n/F10T02a). <p> <b>Processing</b> These images were processed using the CReSIS Synthetic Aperture Radar Processor (CSARP), as part of the Open Polar Radar Effort. Data were processed using pulse compression and matched filter approach for focusing optimized for producing data with 25 m along track sampling. Radio Frequency Interference was partially removed. See the Open Polar Radar server for more detail. <p> <b>Data format</b> Radar data is provided in three formats: <p> <i>Browse</i> data in PNG format are provided with marked axis depth projected, correcting for the velocity of ice, and projected along track into consistent project coordinates. Turns are trimmed off. Long transects are projected to ~30x vertical exaggeration, shorter transects have constant size. <p> <i>Image</i> data in grayscale JPEG format are provided without ornamentation. but are depth projected, correcting for the velocity of ice, and projected along track into consistent project coordinates. Turns are trimmed off. All images have a constant vertical scale of 1.69 m/pixel and horizontal scale of 25 m per pixel. The minimum black value corresponds to -140 dB, and the maximum white value corresponds to 0 dB, for a resolution of ~0.5 dB. Use of this data for radiometric interpretation has not been validated. <p> <i>Metadata</i> is provided in in comma delimited csv format. Columns included: <p> CSARP record (the number of record or trace in the original flight based processing<br> UNIX time [s] (seconds from midnight January 1, 1970, with no leap seconds) <br> Longitude [degrees] (WGS-84) <br> Latitude [degrees] (WGS-84) <br> Aircraft Elevation [m] (WGS-84) <br> Surface Echo Delay [s] (time delay between surface echo and transmission) <br> Roll [degrees] (right wing down positive) <br> Pitch [degrees] (nose down positive) <br> Heading [degrees] (right of North) <br> EPSG 3031 Easting [m] (projected coordinate) <br> EPSG 3031 Northing [m] (projected coordinate) <br> displayed_distance [km] (x-axis distance) <br> surface_elevation [m] (radar estimate surface elevation, WGS-84)<br> blanking [px] (sampled (blanked above surface return)<br> Elevation of image top [m] (WGS-84 elevation of the top of the projected image) <br> Elevation of image bottom [m] (WGS-84 elevation of the bottom of the projected image) <br> <p> A summary csv file is provided with transect name, start and end points in geographic and projected coordinates, and projection. <p> <b>Acknowledgements</b> This work was supported by the Center for Oldest Ice Exploration, an NSF Science and Technology Center (NSF 2019719). We thank the NSF Office of Polar Programs, the NSF Office of Integrative Activities, and Oregon State University for financial and infrastructure support, and the NSF Antarctic Infrastructure and Logistics Program, and the Antarctic Support Contractor for logistical support. Additional support was provided by the G. Unger Vetlesen Foundation and the NSF-sponsored Open Polar Radar project (NSF 2126503 & 2127606). 

This work describes the design and development of a radar receiver with a large dynamic range by means of carefully designed compression. The receiver is designed for ice sounding applications on the Antarctic and Greenland ice sheets and is designed to be usable over a large frequency range (VHF and UHF) and with multiple analog-to digital converters with only minor modifications. We present the receiver design, in which we have implemented an RF-power limiting feature so that the output power is monotonically increasing with respect to the input power over a large dynamic range. This allows the receiver to operate in the nonlinear region to compress the high-power returns into the dynamic range of the analog to digital converter while still achieving good sensitivity (low noise figure) for low power signals. We discuss design considerations, hardware description, initial lab test results, the architecture of the design and results from recent field deployments. Lastly, we discuss the future work on the decompression mechanism to recover the uncompressed signals.