Research method
Satellite Remote Sensing
10 studies
Satellite remote sensing infers a geophysical field from a sensor — optical reflectance, aerosol optical depth, synthetic-aperture radar backscatter or interferograms, feature-tracked ice velocity, or a stereo DEM — rather than measuring that field in situ. Retrievals in this library include MAIAC 1 km AOD, MODIS/MISR/OMI aerosol columns, Sentinel-1 SAR, ITS_LIVE Landsat velocities, Planet/Sentinel-2 iceberg maps, and REMA height change. They are statistical or physically inverted products, not rain gauges, EPA monitors, or ice-penetrating radar by themselves.
Earth-system papers reach for satellites when they need spatially complete maps of ice speed, iceberg area, aerosol loading, or landslide motion that no gauge network can provide. It answers ‘what does this sensor see after the retrieval?’ Its main limitation is that column AOD is not surface PM2.5, 12-day SAR averages smear daily glacier events, and an optical threshold can miss small icebergs that a trained UNet still counts.
Studies
10
Findings
6
7 supporting · 0 challenging · 0 qualifying citations
Open tensions
2
Currently
What we know
- Yearly two-stage models of MAIAC 1 km AOD, meteorology and land use against EPA monitors mapped 2001–2010 SE US PM2.5 with cross-validated R² about 0.71–0.85. Domain-mean PM2.5 fell about 20% (about 23% in metro Atlanta), steepest between 2007 and 2008, with larger declines along urban/highway corridors than in forests. The maps are AOD-to-PM2.5 predictions, not chemically speciated aerosol.
- Over Tehran, winter satellite AOD is a minimum while Mehrabad visibility is worst (6.3–6.8 km) because Alborz blocking traps urban aerosol near the ground, below what the column retrieval emphasises. Summer AOD rises with dust and dry weather. Friday visibility anomalies and post-rain clearing show local emissions plus wet scavenging — signals a column product alone would invert.
- ITS_LIVE Landsat feature-tracking (120 m annual mosaics) on 102 Greenland lake-terminating outlet glaciers versus land-terminating pairs shows 2017 terminus velocity 231% higher at lake termini. Median speeds were 43 versus 36 m yr⁻¹ overall, and 43 versus 13 m yr⁻¹ at 500 m inland. About 44% of lake-terminating glaciers accelerate toward the terminus, versus 3.9% of land-terminating glaciers; the contrast is significant from about 3.5 km inland to the margin.
- A UNet trained on ~10,000 manual labels in ~3 m Planet imagery mapped 60.4 km² of icebergs in Sermilik — ~50% more area than same-day Sentinel-2 top-of-atmosphere thresholding, with ~5× more small bergs. Size distributions follow a power law only above ~300 m². Along-fjord discharge is ~1310 m³ s⁻¹ (~2020 m³ s⁻¹ in a June–July comparison). Small bergs, which optical thresholding undercounts, melt near the surface.
- A buried East Antarctic grounding-zone lake ~3.5 m down, holding ~1.5 billion m³, drained through preexisting fractures. REMA DEM differencing showed a multi-metre surface drop, and Sentinel-1 backscatter jumped during March–April 2016. GPR supplied the water volume and fracture geometry; the satellite record timed the collapse. This is one lake, not a continent-wide census.
Largest unresolved question
Column AOD and near-surface visibility can move in opposite seasons. Tehran’s winter AOD minimum coincides with the worst Mehrabad visibility (6.3–6.8 km) because aerosol is trapped in a shallow layer; SE US MAIAC maps still track EPA monitors at R² 0.71–0.85 because that retrieval is statistically calibrated to surface PM2.5. Calling both ‘satellite aerosol’ hides whether the product is a column or a ground-trained map.
Procedure
How it works
The main stages of the method. Specific protocols and analysis choices vary between studies.
Record radiance, backscatter, or stereo from orbit
A sensor measures optical reflectance, aerosol optical depth, SAR backscatter or interferograms, feature-tracked ice velocity, or a stereo DEM. The spacecraft does not weigh PM2.5 or clock a glacier with a GNSS stake.
Invert or train a retrieval
A physical inversion or a statistical model turns the raw signal into a geophysical field — MAIAC AOD to PM2.5, Sentinel-1 to ice speed, a UNet to iceberg area. Cloud-missing AOD still leaves gaps; the retrieval is a product, not a monitor.
Map the field in space and time
The output is a spatially complete map of ice speed, iceberg area, aerosol loading, or landslide motion. Twelve-day SAR averages smear daily glacier events; an optical threshold can miss small icebergs that a trained UNet still counts.
Compare with gauges when they exist
MAIAC PM2.5 maps are trained on EPA monitors; column AOD over Tehran can sit at a winter minimum while near-surface visibility is worst because aerosol is trapped in a shallow layer. Column optical depth is not surface PM2.5.
Do not treat the retrieval as an in situ sample
Satellite remote sensing answers what this sensor sees after the retrieval. It is not a rain gauge, not chemically speciated PM, and not ice-penetrating radar by itself.
Common misconceptions
A satellite retrieval is a rain gauge or an EPA monitor.
MAIAC PM2.5 maps are two-stage statistical predictions from AOD, meteorology and land use; cloud-missing AOD still leaves gaps. They are not chemically speciated samples, and they do not measure health outcomes.
Satellite AOD is surface PM2.5.
Tehran winter AOD is a minimum while visibility at the ground is 6.3–6.8 km because the aerosol sits in a blocked near-surface layer. Column optical depth can miss the layer a visibilimeter sees.
Sentinel-1 ice velocity is a daily speedometer, and optical iceberg area is complete.
Twelve-day SAR averages smear daily speed-ups; ITS_LIVE 2017 speeds are annual mosaics of Landsat pairs. Sentinel-2 TOA thresholding missed ~50% of Sermilik iceberg area and ~5× the small bergs relative to a Planet UNet.
Related
Claim ledger
What the evidence shows
Drawn from 10 studies in this library. Mix labels say which citation roles are present; they are not a strength score. Supports means evidence for a finding; Challenges means evidence against a stated position; Qualifies marks scope.
Yearly two-stage models of MAIAC 1 km AOD, meteorology and land use against EPA monitors mapped 2001–2010 SE US PM2.5 with cross-validated R² about 0.71–0.85. Domain-mean PM2.5 fell about 20% (about 23% in metro Atlanta), steepest between 2007 and 2008, with larger declines along urban/highway corridors than in forests. The maps are AOD-to-PM2.5 predictions, not chemically speciated aerosol.
Over Tehran, winter satellite AOD is a minimum while Mehrabad visibility is worst (6.3–6.8 km) because Alborz blocking traps urban aerosol near the ground, below what the column retrieval emphasises. Summer AOD rises with dust and dry weather. Friday visibility anomalies and post-rain clearing show local emissions plus wet scavenging — signals a column product alone would invert.
ITS_LIVE Landsat feature-tracking (120 m annual mosaics) on 102 Greenland lake-terminating outlet glaciers versus land-terminating pairs shows 2017 terminus velocity 231% higher at lake termini. Median speeds were 43 versus 36 m yr⁻¹ overall, and 43 versus 13 m yr⁻¹ at 500 m inland. About 44% of lake-terminating glaciers accelerate toward the terminus, versus 3.9% of land-terminating glaciers; the contrast is significant from about 3.5 km inland to the margin.
A UNet trained on ~10,000 manual labels in ~3 m Planet imagery mapped 60.4 km² of icebergs in Sermilik — ~50% more area than same-day Sentinel-2 top-of-atmosphere thresholding, with ~5× more small bergs. Size distributions follow a power law only above ~300 m². Along-fjord discharge is ~1310 m³ s⁻¹ (~2020 m³ s⁻¹ in a June–July comparison). Small bergs, which optical thresholding undercounts, melt near the surface.
A buried East Antarctic grounding-zone lake ~3.5 m down, holding ~1.5 billion m³, drained through preexisting fractures. REMA DEM differencing showed a multi-metre surface drop, and Sentinel-1 backscatter jumped during March–April 2016. GPR supplied the water volume and fracture geometry; the satellite record timed the collapse. This is one lake, not a continent-wide census.
Sentinel-1 is not only ice: 12-day velocities on four Icelandic soft-bed outlets show spring speed-ups ~17 days after melt onset at Skálafellsjökull, while 90 m ARIA interferograms of 247 California landslides (176, 71%, in Franciscan mélange) moved at 0.85–9.7 cm/yr — faster in wet WY2017 and WY2019, slower in dry WY2016 and WY2018, despite a >10× rainfall range. Twelve-day SAR averages smear daily glacier events; InSAR misses heading-parallel motion.
- Soft-bed glaciers show a hydrology continuum
- Wet- and dry-climate landslides both speed up in wet years
Study Role Design N Population Outcome Soft-bed glaciers show a hydrology continuum Supports OtherGNSS, till probes, GPR, and Sentinel-1 seasonal velocities on soft-bed outlet glaciers N=4 · Four glaciers (Skálafellsjökull, Fjallsjökull, Breiðamerkurjökull, Briksdalsbreen) Soft-bedded Icelandic and Norwegian outlet glaciers Seasonal speed-up patterns indicating mixed soft-bed subglacial drainage 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
Debates
Tensions and limits
Some items are genuine disagreements on the same question. Others mark different assays, populations, or outcomes.
Column AOD and near-surface visibility can move in opposite seasons. Tehran’s winter AOD minimum coincides with the worst Mehrabad visibility (6.3–6.8 km) because aerosol is trapped in a shallow layer; SE US MAIAC maps still track EPA monitors at R² 0.71–0.85 because that retrieval is statistically calibrated to surface PM2.5. Calling both ‘satellite aerosol’ hides whether the product is a column or a ground-trained map.
Column AOD and near-surface visibility can move in opposite seasons. Tehran’s winter AOD minimum coincides with the worst Mehrabad visibility (6.3–6.8 km) because aerosol is trapped in a shallow layer; SE US MAIAC maps still track EPA monitors at R² 0.71–0.85 because that retrieval is statistically calibrated to surface PM2.5. Calling both ‘satellite aerosol’ hides whether the product is a column or a ground-trained map.
Study Role Design N Population Outcome Why Tehran AOD and visibility move opposite seasons Supports OtherMulti-sensor satellite, surface visibility, and trajectory/speciation synthesis for Tehran aerosols 2000–2009 multi-year characterization; not a discrete sample-size study Greater Tehran Area aerosol column and surface visibility (Mehrabad) Seasonal AOD vs visibility contrast under mountain trapping and summer dust How 1 km satellite maps show SE US PM2.5 fell Supports OtherTwo-stage MAIAC AOD models mapping daily 1 km PM2.5 across the southeastern US, 2001–2010 Domain-wide satellite/monitor fusion over 10 years; not a person-level sample N Southeastern United States ambient PM2.5 field (EPA monitors + MAIAC AOD) Spatial–temporal PM2.5 trends (~20% domain-mean decline 2001–2010)
PaperFren reads this as a limit on how far one study travels — different assays, populations, or outcomes — not a forced fight between papers.
Optical and radar ice products do not count the same objects at the same timescale. Planet UNet finds ~50% more iceberg area and ~5× more small bergs than Sentinel-2 thresholding; 12-day Sentinel-1 glacier speeds smear events that GNSS resolves daily; ITS_LIVE annual mosaics average many Landsat pairs. A student who adds Sermilik meltwater, Greenland lake-terminus speed, and Iceland spring speed-up as one ‘SAR velocity’ is mixing sensors, resolutions and averaging windows.
Optical and radar ice products do not count the same objects at the same timescale. Planet UNet finds ~50% more iceberg area and ~5× more small bergs than Sentinel-2 thresholding; 12-day Sentinel-1 glacier speeds smear events that GNSS resolves daily; ITS_LIVE annual mosaics average many Landsat pairs. A student who adds Sermilik meltwater, Greenland lake-terminus speed, and Iceland spring speed-up as one ‘SAR velocity’ is mixing sensors, resolutions and averaging windows.
- Small icebergs add a lot of Greenland fjord freshwater
- Proglacial lakes speed Greenland outlet glaciers
- Soft-bed glaciers show a hydrology continuum
Study Role Design N Population Outcome Small icebergs add a lot of Greenland fjord freshwater Supports OtherUNet mapping of icebergs in Planet imagery vs Sentinel-2 thresholding in Greenland fjords N=10000 · ~10,000 manual iceberg labels used to train the UNet Icebergs in Sermilik and Kangerlussuaq fjords, East Greenland Higher small-iceberg area and meltwater discharge estimates (~1300–2000 m³ s−1 Sermilik) Proglacial lakes speed Greenland outlet glaciers Supports OtherPaired comparison of lake- vs land-terminating Greenland outlet glaciers with ITS_LIVE velocities N=102 · 102 lake-terminating outlets paired with land-terminating counterparts Greenland outlet glaciers terminating in ice-marginal lakes vs on land Near-terminus ice-surface velocity enhancement associated with proglacial lakes Soft-bed glaciers show a hydrology continuum Supports OtherGNSS, till probes, GPR, and Sentinel-1 seasonal velocities on soft-bed outlet glaciers N=4 · Four glaciers (Skálafellsjökull, Fjallsjökull, Breiðamerkurjökull, Briksdalsbreen) Soft-bedded Icelandic and Norwegian outlet glaciers Seasonal speed-up patterns indicating mixed soft-bed subglacial drainage
PaperFren reads this as a limit on how far one study travels — different assays, populations, or outcomes — not a forced fight between papers.
Timeline
How understanding moved
Study years are when the paper was published. Evidence edits are dated changes to this page's claims. Explanations are when PaperFren added a Discovery — not a claim that the science happened that day.
2026
- Prairie wetlands set the pace of annual runoff
- Proglacial lakes speed Greenland outlet glaciers
- Why two giant icebergs fertilize the ocean differently
Concept page published
Satellite Remote Sensing
Change log
What changed
Dated edits to this page's evidence: studies added or removed from a claim, claims added or withdrawn, and new explanations tagged here. Coverage edits are distinct from reviewed research changes and corrections. Dates record PaperFren updates; study publication dates appear in the timeline.
- Coverage updated · Concept page published
Papers
10 studies in this library bear on Satellite Remote Sensing, ordered by citations.
- How 1 km satellite maps show SE US PM2.5 fell
A two-stage MAIAC AOD model maps daily 1 km PM2.5 and shows about a 20% drop across the southeastern US from 2001 to 2010.
- Why Tehran AOD and visibility move opposite seasons
Winter mountain trapping cuts visibility while column AOD is lowest; summer dust raises AOD even as surface visibility recovers.
- 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.
- Small icebergs add a lot of Greenland fjord freshwater
High-resolution UNet mapping finds far more small icebergs than Sentinel-2 thresholding, raising Sermilik meltwater estimates toward ~2000 m³ s−1.
- Arabian Sea cyclones briefly refill desert aquifers
A few intense tropical cyclones now supply most of interior Oman’s rain and can add cubic kilometres of groundwater, interrupting chronic depletion.
- Soft-bed glaciers show a hydrology continuum
Seasonal ice-speed patterns on Icelandic and Norwegian glaciers reveal mixed, not purely channelized or distributed, soft-bed drainage.
- Why two giant icebergs fertilize the ocean differently
A recently calved iceberg dumped meltwater and triggered a bloom, while a decades-grounded berg barely changed nearby waters.
Show 2 more studiesShow fewer studies
- Proglacial lakes speed Greenland outlet glaciers
Lake-terminating Greenland outlet glaciers flow much faster near their fronts than otherwise similar land-terminating glaciers.
- Prairie wetlands set the pace of annual runoff
In 69% of 109 Prairie Pothole catchments, maximum wetland inundation outpredicts climate for annual runoff, often with a storage threshold.
Compare studies
Select 2–10 studies. Design and N are labels, not a ranking.
Nothing selected yet.
Questions
What is still open
Column AOD and near-surface visibility can move in opposite seasons. Tehran’s winter AOD minimum coincides with the worst Mehrabad visibility (6.3–6.8 km) because aerosol is trapped in a shallow layer; SE US MAIAC maps still track EPA monitors at R² 0.71–0.85 because that retrieval is statistically calibrated to surface PM2.5. Calling both ‘satellite aerosol’ hides whether the product is a column or a ground-trained map.
Optical and radar ice products do not count the same objects at the same timescale. Planet UNet finds ~50% more iceberg area and ~5× more small bergs than Sentinel-2 thresholding; 12-day Sentinel-1 glacier speeds smear events that GNSS resolves daily; ITS_LIVE annual mosaics average many Landsat pairs. A student who adds Sermilik meltwater, Greenland lake-terminus speed, and Iceland spring speed-up as one ‘SAR velocity’ is mixing sensors, resolutions and averaging windows.
Ask PaperFren about Satellite Remote Sensing
Study this conceptflashcards and short-answer questions
Why can Tehran winter AOD be a minimum in the same weeks Mehrabad visibility falls to 6.3–6.8 km, while SE US MAIAC AOD still yields PM2.5 maps with R² 0.71–0.85?
AOD is a column. Alborz blocking traps winter urban aerosol near the ground, so the column can look clean while surface visibility is poor. The SE US maps are statistically trained on EPA monitors (day-specific AOD slopes plus geographically weighted regression), so they are forced to track surface PM2.5 — they are not a raw column converted 1:1.
Sermilik P-UNet mapped 60.4 km² of icebergs, ~50% more than Sentinel-2 TOA thresholding. What size class drives that gap, and why does it matter for fjord freshwater?
Small bergs: ~5× more than S-TOA, and the size distribution is a power law only above ~300 m². Those small bergs melt near the surface, so undercounting them undercounts near-surface freshwater (~1310 m³ s⁻¹ along-fjord; ~2020 m³ s⁻¹ in a June–July comparison).
Lake-terminating Greenland outlets were 231% faster at the terminus in 2017 (43 versus 13 m yr⁻¹ at 500 m inland). What did ITS_LIVE actually measure, and what mechanism remains inferred?
Annual Landsat feature-tracking mosaics of ice-surface speed, sampled along flowlines. Calving rates and submarine melt were not measured, so faster flow is a velocity contrast with land-terminating pairs, not a direct observation of the lake’s mechanical forcing.
How did Sentinel-1 and REMA contribute to the buried-lake paper without replacing GPR?
REMA DEM differencing showed a multi-metre surface drop; Sentinel-1 backscatter jumped in March–April 2016 and timed drainage. GPR measured burial depth (~3.5 m), volume (~1.5 billion m³) and lake-bed fractures. Satellites did not image the water body; they recorded the surface and microwave consequences of collapse.