Research method
Radar Observations
Radar observations in this library are three different instruments. IceBridge ice-penetrating radar measures ice thickness for BedMachine v3. Ground-penetrating radar (400 MHz) images a buried Antarctic lake and its fractures. Sentinel-1 C-band InSAR (ARIA, 90 m) maps slow-moving California landslides. Sharing the word ‘radar’ does not make a bed map, a hydrofracture survey and an interferogram the same method. None of them is a rain gauge or a tide gauge.
Glaciology and geomorphology reach for radar when the target is buried: ice thickness, englacial water, or millimetre-to-centimetre surface motion through cloud. It answers ‘what does this frequency see at this baseline?’ Its main limitation is that a 150 m BedMachine grid is ~1 km where kriging fills gaps, GPR volume misses the southern shoreline, and 90 m InSAR misses small slides and heading-parallel motion.
Evidence
What the evidence shows
Drawn from 3 studies in this library. Each finding starts with a plain-language takeaway, then the denser detail. Supports means evidence for a finding; Challenges means evidence against a stated position; Qualifies marks scope with a short note on each study’s contribution. Challenged positions are labeled — they are not findings.
BedMachine v3 combines IceBridge radar, mass conservation where ice flows faster than 50 m/yr, and 30 bathymetry sources including OMG multibeam, mapping bed and fjords within 50 km of the Greenland coast at up to 150 m. Ice volume is 2.99±0.02×10⁶ km³; sea-level potential 7.42±0.05 m (7 cm above prior). Of 243 glaciers, 67 connect to ocean deeper than 200 m. Retreat patterns match retrograde beds; v2 beds were often ~100 m too high. True resolution is coarser (~1 km) where kriging/RTopo-2 fill gaps; radar thickness uncertainty is ~50 m.
GPR before and after collapse showed a buried East Antarctic grounding-zone lake ~3.5 m below the surface, average water depth 2.2 m (maximum 4.6 m), and ~1.5 billion m³ of water that drained through preexisting lake-bed fractures. Post-collapse profiles showed ~1 m vertical discontinuities. REMA and Sentinel-1 timed a multi-metre surface drop and a March–April 2016 backscatter jump; volume is likely underestimated because the GPR grid missed the southern shoreline.
Sentinel-1 ARIA interferograms (water years 2015–2020) mapped 247 slow-moving landslides; 230/247 sit in the Coast Ranges and 176 (71%) in Franciscan mélange. Mean areas are 0.5 km²; speeds averaged 0.85–9.7 cm/yr. A 38-slide subset spanning 216–2180 mm/yr rain was faster in wet WY2017 and WY2019 and slower in dry WY2016 and WY2018, despite >10× rainfall and ~4× thickness range. One coastal slide sped up in dry WY2018 via wave erosion. This is InSAR displacement, not ice-penetrating radar.
Open questions
Tensions and limits
Some items are genuine disagreements on the same question. Others mark different assays, populations, or outcomes — limits on how far one study travels — not a forced fight between papers.
Ice-penetrating radar, GPR and C-band InSAR are not interchangeable ‘radar observations.’ BedMachine infers thickness from IceBridge plus mass conservation and ship bathymetry; the Antarctic paper images a single buried lake at 400 MHz; the landslide paper measures surface motion at 90 m without seeing the subsurface. Combining 7.42 m of Greenland sea-level potential with 1.5 billion m³ of buried lake water as two ice-sheet radar volumes mixes a continent-scale product with one grounding-zone survey.
- Greenland’s hidden fjord floors in BedMachine v3
- A buried Antarctic lake drained through hydrofracture
- Wet- and dry-climate landslides both speed up in wet years
Study Role Design N Population Outcome Greenland’s hidden fjord floors in BedMachine v3 Supports OtherMass-conservation bed mapping combining ice-penetrating radar and ocean bathymetry Greenland-wide BedMachine v3 product; volume 2.99×10^6 km³ — not a sample N Greenland Ice Sheet bed topography and adjacent ocean bathymetry Ice volume, sea-level potential (7.42 m), and ocean access to glacier fronts A buried Antarctic lake drained through hydrofracture Supports OtherGPR before/after collapse plus SNOWPACK and Sentinel-1 monitoring of a buried meltwater lake N=1 · Single buried lake (~1.5 billion m³) at an East Antarctic ice-shelf grounding zone Buried meltwater lake on the grounding zone of an East Antarctic ice shelf Drainage through vertical fractures and implications for ice-shelf weakening Wet- and dry-climate landslides both speed up in wet years Supports OtherSentinel-1 InSAR inventory of California slow-moving landslides vs water-year rainfall N=247 · 247 active slow-moving landslides; 38-slide subset tracked across a large rainfall gradient Slow-moving landslides in California (mainly Coast Ranges Franciscan mélange) Water-year velocity response to wetter vs drier precipitation across climates Even within ice radar, grid spacing is not resolution. BedMachine’s 150 m grid is ~1 km where kriging/RTopo-2 fill, and radar thickness uncertainty is ~50 m; thinning since 2003–2008 is ignored except where it exceeds that error. GPR lake volume is a lower bound because the survey missed the southern shore. A student who treats both as ‘150 m truth’ will overstate how well the bed or the lake is known.
- Greenland’s hidden fjord floors in BedMachine v3
- A buried Antarctic lake drained through hydrofracture
Study Role Design N Population Outcome Greenland’s hidden fjord floors in BedMachine v3 Supports OtherMass-conservation bed mapping combining ice-penetrating radar and ocean bathymetry Greenland-wide BedMachine v3 product; volume 2.99×10^6 km³ — not a sample N Greenland Ice Sheet bed topography and adjacent ocean bathymetry Ice volume, sea-level potential (7.42 m), and ocean access to glacier fronts A buried Antarctic lake drained through hydrofracture Supports OtherGPR before/after collapse plus SNOWPACK and Sentinel-1 monitoring of a buried meltwater lake N=1 · Single buried lake (~1.5 billion m³) at an East Antarctic ice-shelf grounding zone Buried meltwater lake on the grounding zone of an East Antarctic ice shelf Drainage through vertical fractures and implications for ice-shelf weakening
Common misconceptions
BedMachine v3 is the true Greenland bed at 150 m everywhere.
Where IceBridge and mass conservation are dense, spacing can be 150 m; where kriging or RTopo-2 fill, true resolution is ~1 km. Radar thickness uncertainty is ~50 m, and post-2003–2008 thinning is ignored except where it exceeds that error. v2 beds were often ~100 m too high — a reminder that the product is an inversion, not a survey of every fjord.
Sentinel-1 InSAR is ice-penetrating radar.
ARIA interferograms measure line-of-sight surface displacement of landslides at 90 m. They miss small slides and heading-parallel motion. They do not measure ice thickness or buried water; that is IceBridge or GPR.
One GPR-derived lake volume is a continent-wide meltwater census, and the fractures already reach the ice-shelf front.
The survey is one grounding-zone lake; volume (~1.5 billion m³) is likely underestimated, and the study cannot prove that these fractures currently reach the shelf front or trigger collapse. Fractures can advect onto the shelf — that is a hypothesis supported by ice velocity of 165 m/yr, not a mapped shelf-wide failure.
Exam-style questions
Short-answer questions that ask you to explain or compare, not recall.
BedMachine v3 raises Greenland sea-level potential to 7.42±0.05 m (7 cm above prior) and finds 67 of 243 glaciers connect to ocean deeper than 200 m. Which two data types besides IceBridge radar were required, and where is the 150 m grid a fiction?
Mass conservation where ice flows faster than 50 m/yr, and 30 bathymetry sources including OMG multibeam (fjords within 50 km of the coast). The 150 m grid is coarser (~1 km) where kriging/RTopo-2 fill gaps. Radar thickness uncertainty is ~50 m.
How did GPR show the buried lake drained by hydrofracture rather than by horizontal leakage?
Precollapse 400 MHz profiles showed lake-bed discontinuities (closed vertical fractures) and ~3.5 m burial / ~1.5 billion m³; post-collapse profiles showed ~1 m offset stratigraphy. Sentinel-1 lacked a lateral backscatter change outside the basin, consistent with vertical drainage. Volume is a lower bound because the southern shoreline was unsurveyed.
California landslides moved at 0.85–9.7 cm/yr and sped up in wet water years across a >10× rainfall range. Why can 90 m Sentinel-1 InSAR not be used as an ice-thickness map of those hills?
C-band InSAR measures surface displacement along the look direction, not subsurface structure. It already misses small slides and heading-parallel motion. Ice thickness is an ice-penetrating-radar problem (IceBridge/BedMachine), not an interferogram of soil and mélange.
A student cites all three papers as ‘radar shows ice sheets are thinning.’ What is each paper actually a radar observation of?
BedMachine: Greenland bed/thickness and fjord bathymetry (IceBridge plus mass conservation), not a thinning time series — thinning since 2003–2008 is ignored except where it exceeds ~50 m error. Buried lake: one hydrofracture drainage imaged by GPR. Landslides: California InSAR surface speed, not ice. None of the three is an ice-sheet thinning record.
The studies
3 studies in this library bear on Radar Observations, ordered by citations.
- Greenland’s hidden fjord floors in BedMachine v3
A 150 m Greenland bed and fjord map raises sea-level potential by 7 cm and shows 30–100% more glaciers can meet warm Atlantic Water.
- Wet- and dry-climate landslides both speed up in wet years
Across California’s huge rainfall gradient, slow landslides still accelerate when local water years are wetter than average.
- A buried Antarctic lake drained through hydrofracture
Ground-penetrating radar caught a buried meltwater lake draining through vertical fractures at an East Antarctic ice-shelf grounding zone.
Learn alongside