Sensors
- SMOS (ESA): L-band passive microwave radiometer at 1.4 GHz. Native spatial resolution approximately 35–50 km, which limits use to large site complexes or regional anomaly mapping. Global revisit every 2–3 days. Brightness temperature sensitivity to soil moisture and dielectric contrast at depth makes it the primary emissivity anomaly detector in sabkha contexts.
- AMSR2 (JAXA, GCOM-W1): Passive microwave radiometer covering 6.9 GHz to 89 GHz. Resolution ranges from roughly 3 km at 6.9 GHz to 0.3 km at 89 GHz. Daily global revisit. The lower-frequency channels penetrate salt crust and shallow overburden; brightness temperature contrasts at 6.9 and 10.65 GHz have been used alongside SMOS to cross-validate subsurface dielectric anomalies.
- ALOS-2 PALSAR-2 (JAXA): L-band (1.27 GHz) synthetic aperture radar with spatial resolution down to 3 m in spotlight mode. L-band penetrates dry sandy or saline overburden to depths of 1–2 m depending on moisture content. Used alongside passive microwave data to characterise surface roughness and confirm that brightness temperature anomalies are not artefacts of surface texture alone.
- Sentinel-1 (ESA): C-band (5.4 GHz) SAR with 5 × 20 m resolution in Interferometric Wide Swath mode and 6-day repeat at mid-latitudes. C-band penetration in dry salt environments is shallower than L-band, typically less than 0.5 m, but the dense archive back to 2014 supports multitemporal backscatter analysis to isolate persistent subsurface anomalies from seasonal surface moisture variation.
Why salt is a problem and a clue
Sabkha surfaces are among the most reflective terrain on Earth in the visible and near-infrared. A salt crust with albedo above 0.6 saturates optical sensors, washing out any spectral contrast from buried features. Thermal infrared fares somewhat better but is itself sensitive to surface emissivity variations caused by the salt mineralogy rather than what lies beneath. Active C-band SAR sees the surface roughness of the crust and little else in wet conditions, when the brine layer acts as a near-perfect reflector.
Passive microwave emission is different. The brightness temperature recorded by a radiometer depends on the physical temperature of the material multiplied by its emissivity, and emissivity at L-band is controlled primarily by the dielectric constant of the medium. Compacted mudbrick, fired brick rubble, or dense occupation debris has a dielectric signature distinct from the loose saline sediment surrounding it. Where that contrast exists at depth, it perturbs the upwelling brightness temperature by amounts that SMOS and AMSR2 can detect, even through a salt crust that would defeat every optical approach. The salt is simultaneously the obstacle and the reason the signal survives: it keeps the surface dry enough that L-band penetration reaches the archaeology.
What the physics actually permits
L-band microwave penetration depth in dry, low-moisture saline sediment is governed by the skin depth equation. In practice, published field studies suggest effective penetration of 1 to 2 metres in hyperarid sabkha, dropping sharply if interstitial brine is present. This means the method is sensitive to shallow buried features, not deep stratigraphy. A tell with two metres of overburden is detectable in principle; a Bronze Age deposit beneath five metres of alluvium is not.
The resolution limit is the harder constraint. SMOS footprints of 35–50 km mean that a single anomalous pixel could correspond to a site complex the size of a small town, or to a geological dielectric inhomogeneity with no archaeological significance at all. AMSR2 at 6.9 GHz narrows this to roughly 35 km; at 36.5 GHz the footprint is around 8 km but penetration is far shallower. No currently operational passive microwave sensor resolves individual structures. The method identifies candidate zones for follow-on investigation, not site plans.
Combining passive microwave with L-band SAR to reduce false positives
A brightness temperature anomaly in SMOS data is a hypothesis, not a discovery. The standard published workflow pairs that anomaly with ALOS-2 PALSAR-2 backscatter to check whether the surface at the candidate location shows roughness or moisture patterns consistent with a buried feature rather than, say, a patch of different sediment facies. Where PALSAR-2 backscatter is spatially correlated with the microwave anomaly and the correlation persists across multiple acquisition dates, the probability of a genuine subsurface dielectric contrast rises substantially.
Sentinel-1 multitemporal coherence adds a third layer. Buried compacted structures tend to stabilise the surface above them slightly, producing marginally higher interferometric coherence than the surrounding loose sabkha. The effect is subtle and can be masked by wind-blown salt redistribution, but in a time series of fifty or more Sentinel-1 acquisitions the persistent coherence signal becomes statistically distinguishable from noise. The three-sensor combination, passive microwave for the initial anomaly, L-band SAR for penetrating confirmation, C-band coherence for surface stability, is the current published best practice for this environment.
Honest limits: what this approach cannot do
Resolution is the ceiling on ambition. No amount of processing converts a 35 km SMOS pixel into a site plan. Researchers working in the Arabian Peninsula and the Saharan margins have used this method to flag kilometre-scale zones warranting ground survey, not to map individual buildings. Any claim of finer discrimination should be treated with scepticism unless it rests on a sensor not yet in orbit.
Geological mimics are a persistent problem. Subsurface changes in sediment grain size, buried evaporite lenses, or variations in palaeo-water-table depth all produce dielectric contrasts that look archaeologically interesting from orbit. Ground-truth remains essential. The method is best understood as a triage tool that directs scarce field resources toward statistically anomalous zones rather than as a substitute for excavation or even for ground-penetrating radar survey.
Seasonal flooding of sabkha environments temporarily destroys the signal entirely. When brine saturates the upper sediment column, L-band skin depth collapses and passive microwave emission is dominated by the water surface. Analysis windows must be chosen from the dry season, which in the Arabian Gulf region typically runs from May through September. Archive depth for SMOS extends to 2010 and for AMSR2 to 2012, giving over a decade of dry-season composites for trend analysis.
Workflow from anomaly to candidate site list
A practical detection workflow begins with a multi-year dry-season composite of SMOS brightness temperatures at horizontal and vertical polarisation. Anomalies are identified by their departure from a background model fitted to the regional geology. Candidate pixels are then cross-referenced against AMSR2 at 6.9 and 10.65 GHz to check spectral consistency, and against the Sentinel-1 archive for persistent coherence. Surviving candidates are ranked by the number of independent sensor lines of evidence supporting them.
The output is a georeferenced candidate list with confidence tiers, not a map of confirmed sites. Tier-one candidates, supported by all three sensor types across multiple seasons, are recommended for airborne survey or targeted field inspection. Tier-two candidates, flagged by passive microwave alone, are logged for opportunistic ground-truth when fieldwork is planned nearby. Satellize applies this workflow as part of its remote-sensing analytics service; the Tonga crop-estimation programme uses a structurally similar multi-sensor anomaly triage approach, adapted for a very different physical environment. Delivery is as a GIS layer with attached confidence metadata, ready for import into standard heritage management databases.
Who uses this and what they do with it
National heritage authorities in the Arabian Peninsula and North Africa face the practical problem of surveying millions of square kilometres of sabkha and playa terrain with limited field budgets. Satellite-based triage that narrows the search space from a continental scale to a list of a few hundred candidate zones is operationally valuable even if the resolution never improves beyond what SMOS currently provides.
Academic programmes at several universities have published results using SMOS and AMSR2 data for sites in Oman, the UAE, and the Libyan Fezzan. The method has also been proposed for application in the Atacama and the Lut Desert, where similar surface conditions exist. The published literature is still relatively sparse compared with optical or active SAR archaeology, which means the false-positive rate in new environments is not yet well characterised. That is an honest statement of where the science stands, and it is the reason ground-truth investment remains non-negotiable.
Typical figures
| Passive microwave spatial resolution (SMOS) | 35–50 km per pixel; site-complex scale only |
| Passive microwave spatial resolution (AMSR2, 6.9 GHz) | ~35 km; ~8 km at 36.5 GHz with reduced penetration |
| L-band SAR resolution (ALOS-2 PALSAR-2) | 3 m (spotlight) to 10 m (ScanSAR); subsurface penetration 1–2 m in dry saline sediment |
| C-band SAR resolution (Sentinel-1 IW) | 5 × 20 m; penetration <0.5 m in dry salt crust |
| Revisit period | SMOS: 2–3 days global; AMSR2: ~1 day; Sentinel-1: 6 days at mid-latitudes; ALOS-2: 14 days |
| Frequency bands used | 1.4 GHz (SMOS L-band passive); 6.9–36.5 GHz (AMSR2 passive); 1.27 GHz (PALSAR-2 active); 5.4 GHz (Sentinel-1 active) |
| Effective detection depth | 1–2 m in hyperarid dry saline sediment; signal lost when brine-saturated |
| Minimum detectable target size | Site complex of several km²; individual structures not resolvable with current sensors |
| Archive depth | SMOS from 2010; AMSR2 from 2012; Sentinel-1 from 2014; ALOS-2 from 2014 |
| Optimal acquisition window | Dry season only; Arabian Gulf region May–September; Saharan margins year-round with caveats |
Analytics Satellize can run
| Dry-season brightness temperature anomaly map | Multi-year SMOS composite with regional background modelling; polarisation ratio analysis at H and V | GeoTIFF anomaly layer with z-score values, tiered by statistical significance |
| Multi-sensor candidate site list | Intersection of SMOS anomalies, AMSR2 spectral consistency check, and Sentinel-1 multitemporal coherence persistence | GIS point/polygon layer with per-candidate confidence tier and supporting sensor evidence flags |
| L-band backscatter subsurface confirmation layer | ALOS-2 PALSAR-2 multitemporal backscatter analysis; spatial correlation with passive microwave anomaly footprints | Raster overlay and tabular correlation report, exportable to heritage GIS platforms |
| Seasonal signal stability assessment | Time-series decomposition of Sentinel-1 coherence across wet and dry seasons to separate surface moisture effects from persistent structural signals | Annual stability index raster and PDF summary report per candidate zone |
| Geological mimic screening report | Cross-reference of candidate zones against published geological maps and evaporite deposit databases to flag likely non-archaeological dielectric contrasts | Annotated candidate list with mimic-risk rating and recommended ground-truth priority |
| Field survey prioritisation brief | Ranked candidate synthesis combining all sensor lines of evidence with logistical accessibility scoring | Short-form PDF briefing note with coordinate list and recommended survey sequence for heritage authority field teams |
Who does the work
We can get this done for you. Satellize runs its own analyst desk and a strong science team. You do not buy a data feed and work out what it means; our people source the imagery, run the analysis described on this page, and hand you the answer with its confidence limits stated. Discuss this requirement.