Geothermal heat flux mapping for epithermal mineral system targeting
Nighttime thermal infrared imagery from ASTER and ECOSTRESS reveals anomalous ground heat flux linked to hydrothermal systems that host epithermal gold and silver deposits, separating geological signal from industrial noise.
Sensors
- ASTER TIR (Terra satellite): Five thermal infrared bands from 8.125 to 11.65 µm at 90 m spatial resolution. Nighttime acquisitions suppress solar heating, isolating geothermal flux. Revisit is approximately 16 days at nadir but can be tasked. The sensor has been in orbit since 1999, providing a deep archive for baseline comparison.
- ECOSTRESS (ISS-mounted): Eight TIR bands near 8.6 to 12.5 µm at 70 m resolution. The non-sun-synchronous ISS orbit delivers acquisitions at varying local times, including multiple nighttime passes per week at mid-latitudes, which is operationally useful for separating diurnal thermal inertia effects. Land surface temperature accuracy is quoted at approximately 1.5 K.
- Landsat-9 TIRS: Two thermal bands (Band 10 at 10.6–11.19 µm, Band 11 at 11.5–12.51 µm) at 100 m resolution, resampled to 30 m in standard products. Sixteen-day repeat. Useful for multi-decadal trend analysis and cross-calibration with ASTER given the overlapping archive from Landsat-8 onwards.
- MODIS LST (Terra and Aqua): Daily land surface temperature products at 1 km resolution. Too coarse to resolve individual anomalies but invaluable as a regional background reference and for filtering seasonal and meteorological temperature variation before analysing higher-resolution scenes.
Why epithermal systems bleed heat upward
Epithermal gold and silver deposits form in the upper kilometre or two of the crust where hydrothermal fluids, driven by magmatic or deep-crustal heat, circulate through fault networks and deposit metals as they cool. That same heat source does not switch off once mineralisation is complete. Active or recently active systems continue to conduct and convect heat toward the surface, producing ground temperatures measurably above the regional background.
The anomalies are modest. In well-documented Andean and East African Rift examples, geothermal surface temperature excesses above background typically range from roughly 2 K to 8 K, depending on depth to the heat source, permeability structure and local cover. That is well within the detection capability of ASTER TIR and ECOSTRESS, but it demands careful atmospheric correction and a rigorous separation from anthropogenic heat, which is the methodological challenge the technique has to solve.
Nighttime acquisition is not optional, it is the method
Solar heating during the day swamps subtle geothermal signals. A dark basalt outcrop in afternoon sun can be 15 K warmer than a pale limestone a few metres away purely from differential albedo. Nighttime acquisitions, acquired several hours after sunset when solar-forced temperature gradients have largely equilibrated, reduce that noise dramatically. ASTER's nighttime tasking capability was exploited in published studies across the Bolivian and Chilean Andes precisely for this reason.
Thermal inertia adds a useful second dimension. Materials with high thermal inertia, typically dense, moist or hydrothermally altered rocks, cool more slowly overnight. A geothermally heated zone therefore appears anomalously warm both because it receives subsurface heat and because hydrothermal alteration products such as clay minerals and silica sinter can modify the thermal inertia of the host rock. These two effects are partially separable by comparing pre-dawn and early-evening acquisitions, which ECOSTRESS's variable overpass times make feasible.
Separating geology from generators: the anthropogenic contamination problem
Mining regions are not quiet at night. Diesel generators, processing plants, heap leach pads and accommodation camps all radiate heat. A poorly sited thermal anomaly map can flag a ball mill discharge rather than a buried intrusion. The separation relies on spatial texture, temporal behaviour and contextual masking.
Geothermal anomalies are spatially diffuse, structurally controlled (they follow fault traces or ring fractures), and persistent across years. Industrial heat sources are compact, rectilinear, and disappear or shift when operations change. MODIS daily LST time series provide a multi-year persistence filter: a pixel that is warm every night for a decade is unlikely to be a generator. Combining that filter with a cadastral or OpenStreetMap mask of known infrastructure removes the majority of false positives before any geological interpretation begins.
Fumarolic fields and hot springs are the clearest geological signal. Where surface expression is absent, the anomaly may still be real but shallower in magnitude and more ambiguous. Honest practice requires reporting confidence tiers rather than a single binary map.
Integration with structural geology: heat follows fractures
A thermal anomaly in isolation has limited exploration value. The diagnostic pattern is a warm zone that aligns with a mapped or inferred fault, a caldera margin, or a volcanic vent cluster. Published studies in the East African Rift, where the Olkaria and Menengai geothermal fields have been extensively characterised, show that surface heat flux anomalies correlate spatially with NW-trending fault systems identifiable in Sentinel-1 SAR and optical imagery.
The practical workflow therefore combines the thermal layer with lineament mapping from SAR or multispectral data (covered separately in the structural geology reference page) and with published geological maps. The thermal data ranks candidate fault segments by heat flux intensity; structural analysis assesses whether those segments have the geometry to focus hydrothermal upflow. Neither input alone is sufficient.
Honest limits of the method
Cloud cover is the most immediate operational constraint. ASTER and ECOSTRESS are optical-thermal sensors; cloud opaque to thermal infrared renders a scene unusable. In the humid tropics and during the wet season in the Andes, cloud-free nighttime acquisitions may be available only a few times per year. MODIS's daily revisit partially compensates by building seasonal composites, but at 1 km resolution it cannot resolve anomalies smaller than a few square kilometres.
Depth sensitivity is a harder limit. The technique detects heat that has reached the surface. A deep, poorly permeable system may produce no detectable surface expression regardless of its thermal budget. Conversely, a shallow magmatic intrusion with no associated mineralisation will produce a strong anomaly. Thermal remote sensing is a targeting tool that reduces the search space; it is not a direct ore indicator. Ground-based follow-up, including soil gas surveys and gravity or magnetotelluric profiling, remains necessary to assess whether a thermal anomaly is associated with a mineralised hydrothermal system.
ASTER's TIR sensor has experienced intermittent operational issues in recent years, and acquisition scheduling requires advance tasking requests. ECOSTRESS provides a partial operational substitute but has its own duty-cycle constraints tied to ISS power and communications. Landsat-9 TIRS, while lower in thermal band count, offers the most reliable repeat coverage for long-term monitoring.
From pixels to a prospect ranking
The analytic output a geoscience team actually needs is not a raw temperature map. It is a ranked list of anomaly clusters with spatial coordinates, estimated temperature excess above background, persistence score, structural alignment score, and a confidence tier. That product can be delivered as a GIS layer with an accompanying technical note, ready to inform a field campaign prioritisation decision.
Satellize runs this workflow on open-archive ASTER, ECOSTRESS and Landsat data, applying atmospheric correction via the MODTRAN-based routines documented in the ASTER science team literature, followed by the persistence and anthropogenic-masking filters described above. The Tonga crop-estimation programme demonstrated the organisation's approach to multi-sensor fusion on constrained archive data; the same compositing logic applies here in a very different physical context. Exploration teams wanting to scope a target region can request a desktop thermal anomaly assessment before committing to field access.
Typical figures
| Best thermal spatial resolution | 70 m (ECOSTRESS) / 90 m (ASTER TIR) |
| Typical revisit at nadir | 16 days (ASTER, Landsat-9); variable, up to several times per week nighttime (ECOSTRESS via ISS) |
| Land surface temperature accuracy | Approximately 1–2 K after atmospheric correction (ASTER, ECOSTRESS published specs) |
| Thermal bands used | ASTER: 5 bands, 8.125–11.65 µm; ECOSTRESS: 8 bands, 8.6–12.5 µm; Landsat-9 TIRS: 2 bands, 10.6–12.51 µm |
| Minimum detectable temperature excess | Approximately 2 K above background under good atmospheric conditions; smaller anomalies require multi-scene compositing |
| Background reference resolution | 1 km (MODIS daily LST products) |
| Archive depth | ASTER: 1999 to present; Landsat thermal: 1982 to present (varying band configurations); ECOSTRESS: 2018 to present |
| Cloud limitation | Optical-thermal sensors; cloud cover renders acquisitions unusable. Wet-season coverage may be limited to a few clear nights per month in tropical regions |
| Delivery formats | GeoTIFF temperature anomaly rasters, GeoPackage or Shapefile anomaly polygon layer, PDF technical note with ranked prospect table |
Analytics Satellize can run
| Nighttime land surface temperature anomaly map | Atmospheric correction of ASTER/ECOSTRESS TIR radiance using MODTRAN-based split-window or temperature-emissivity separation; subtraction of regional background derived from MODIS LST climatology | GeoTIFF raster of temperature excess above background, clipped to area of interest |
| Multi-year thermal persistence score | Pixel-wise percentile analysis of MODIS daily LST time series (up to 20+ years) to identify consistently anomalous locations regardless of seasonal variation | GeoTIFF persistence layer; anomaly polygons ranked by persistence score |
| Anthropogenic heat source mask | Spatial and temporal filtering using cadastral infrastructure data, OpenStreetMap building footprints, and temporal intermittency flags (industrial sources vary; geological sources do not) | Binary exclusion mask integrated into the anomaly layer; flagged uncertain pixels reported separately |
| Structurally controlled anomaly cluster delineation | Spatial intersection of thermal anomaly polygons with lineament traces extracted from SAR or optical data; alignment scoring against published fault databases | GeoPackage layer of anomaly clusters attributed with structural alignment score and nearest mapped fault |
| Ranked prospect shortlist | Multi-criteria scoring combining temperature excess magnitude, persistence, structural alignment and absence of known anthropogenic explanation | PDF technical note with coordinates, ranked table, and per-anomaly confidence tier (high / moderate / low) |
| Emissivity-based alteration proxy | ASTER TIR emissivity spectra (5-band) compared against spectral libraries for silica sinter, kaolinite and alunite, which are common hydrothermal alteration products with distinctive emissivity signatures | Alteration probability raster overlaid on thermal anomaly map; caveats on spectral mixing noted in accompanying note |
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.