Land surface temperature microclimate mapping for open-air heritage conservation

Thermal infrared data from Landsat TIRS and ECOSTRESS reveals heat-stress microzones across open-air heritage sites where temperature cycling drives salt damage and biological colonisation. Site managers gain spatial evidence for shading, drainage and visitor routing decisions.

Sensors

Landsat 8/9 TIRS: Two thermal infrared bands (Band 10 at 10.6–11.2 µm, Band 11 at 11.5–12.5 µm) at 100-metre native resolution, resampled to 30 m in data products. Sixteen-day repeat per satellite; combined Landsat 8 and 9 gives an eight-day revisit. Radiometric precision allows land surface temperature (LST) retrieval to within roughly 1–2 K under clear-sky conditions. Daytime and night-time acquisitions available.
ECOSTRESS (ISS-mounted): Five-band thermal radiometer (8.28–12.13 µm) at approximately 70-metre ground sampling distance. Because the ISS orbit precesses, ECOSTRESS captures sites at varying local times across days and weeks, making it the only freely available sensor that samples diurnal thermal cycling without tasking. LST uncertainty is published at around 1.5 K; evapotranspiration and thermal stress products are standard Level-3 outputs.
ASTER TIR: Five thermal bands (8.125–11.65 µm) at 90-metre resolution, with on-demand day/night acquisition. ASTER's multispectral TIR allows emissivity separation, which matters when comparing thermally distinct materials such as limestone, sandstone and mortar within the same scene. Archive extends to 1999. New acquisitions are now limited following instrument scheduling changes.
Sentinel-3 SLSTR: Dual-view thermal scanner at 1-kilometre resolution in thermal bands. Too coarse for individual monument mapping, but useful for regional context and for validating LST retrieval against finer-resolution sensors. Daily revisit at mid-latitudes makes it a good baseline for anomaly flagging.

Why temperature cycling is the hidden conservation problem

Physical weathering of stone is not simply a function of average temperature. It is driven by the amplitude and frequency of thermal cycles. Repeated expansion and contraction open micro-fractures; salts dissolved in groundwater or deposited by rain migrate into those fractures and crystallise as temperatures rise, wedging grains apart. Biological crusts, lichens and cyanobacteria colonise preferentially on surfaces that stay warm and damp long enough to sustain metabolic activity. None of this is visible in a standard optical image.

A site manager walking the grounds of a stone circle or a terraced garden sees a uniform surface. The thermal camera sees something quite different: south-facing ashlar that peaks at 55–60 °C on a summer afternoon, a shaded north face that never exceeds 25 °C, a drainage hollow that retains moisture and stays 8 °C cooler than surrounding turf at midday. Each of those zones has a different damage trajectory. Mapping them is the first step to doing anything useful about them.

What ECOSTRESS adds that a single daytime pass cannot

Landsat's eight-day combined revisit is adequate for change detection over seasons, but it captures each site at roughly the same local time, typically mid-morning. That single snapshot misses the afternoon thermal peak and the nocturnal cooling curve, both of which matter for salt crystallisation modelling. ECOSTRESS changes this. Because the ISS orbit drifts in local solar time, the same site is imaged at 06:00, 14:00, 20:00 and other times across successive passes, building a diurnal profile over weeks without any additional tasking cost.

The practical output is a site-specific thermal amplitude map: the difference between the daily maximum and minimum LST at each 70-metre pixel. High-amplitude zones are the priority conservation targets, because it is the cycling, not the peak alone, that drives mechanical and chemical damage. Published ECOSTRESS Level-3 products include an Evaporative Stress Index that is a useful proxy for surface moisture retention, another factor in biological colonisation risk.

The honest limit is cloud. ECOSTRESS and Landsat TIRS are both passive thermal sensors; a single overcast day produces no data. In humid climates, building a complete diurnal profile may take months of compositing. Sites in northern Europe or monsoon-affected Asia require longer time windows than sites in the Mediterranean or the Middle East.

Reading the thermal signature of different heritage materials

Different stone types have different thermal inertia, the rate at which they absorb and release heat. Dense granite has high thermal inertia and changes temperature slowly; porous limestone or sandstone heats and cools faster. ASTER's five TIR bands allow emissivity mapping alongside temperature retrieval, which means you can distinguish quartzite from carbonate lithologies spectrally, not just thermally. That distinction matters when a site mixes original fabric with later repair materials: a Roman wall repointed with modern cement will show a different thermal signature from the original ashlar, and that boundary is often where water infiltration concentrates.

Earthworks present a different problem. A grass-covered bank or ditch has low thermal contrast with its surroundings for most of the day. The useful window is early morning, when differential drainage and soil moisture create a brief temperature anomaly before solar heating homogenises the surface. Landsat's night-time thermal band is occasionally used for this, though at 100-metre native resolution it resolves only large earthwork complexes. ECOSTRESS at 70 metres is marginally better but still struggles with features narrower than two pixels.

From thermal map to management decision

A thermal microclimate map is only useful if it connects to something a site manager can act on. The most direct applications are three. First, shading: identifying which surfaces exceed damaging temperature thresholds most frequently tells conservators where temporary or permanent shade structures will have the greatest protective effect. Second, drainage: persistent cool-wet anomalies in the thermal data often correspond to blocked or inadequate drainage, a problem that is cheap to fix once located. Third, visitor routing: footpaths concentrate visitor heat load and compaction; a thermal map can show where path surfaces are already stressed and where rerouting would reduce the cumulative burden.

Biological colonisation mapping is a related output. Lichens and cyanobacteria absorb in the near-infrared differently from bare stone, and combining a thermal layer with a multispectral vegetation or crust index from Sentinel-2 (10-metre resolution) produces a composite risk layer that is more informative than either dataset alone. This fusion approach is well-documented in the conservation science literature, though it requires careful co-registration when thermal and optical resolutions differ by a factor of seven or more.

Satellize runs this kind of multi-sensor thermal and spectral analysis on open constellations, with the Tonga crop-estimation programme as one example of how agronomic stress indices translate to operational decision support. The analytical pipeline for heritage sites is structurally similar: LST retrieval, anomaly classification, change detection over seasons.

Honest limits and what they mean for procurement

Seventy metres is the finest freely available thermal resolution. For a large site such as Stonehenge, Avebury or the Alhambra gardens, that is workable. For a small enclosed courtyard or a single decorated façade, it is not. Airborne thermal surveys using UAV-mounted FLIR cameras routinely achieve sub-10-centimetre resolution and are the appropriate tool for detailed façade assessment. Satellite thermal data is most valuable at the site-wide and landscape scale, identifying priority zones that then justify targeted ground or airborne survey.

LST retrieval accuracy also depends on atmospheric correction quality. Standard Landsat Collection-2 LST products apply a single-channel algorithm with an estimated uncertainty of 2–3 K under good conditions; split-window algorithms using both TIRS bands reduce this somewhat but require accurate emissivity inputs. Users should treat relative differences across a scene with more confidence than absolute temperature values. For conservation decisions based on threshold temperatures, ground-truth with in-situ loggers at a small number of representative points is strongly advisable.

Archive depth is a genuine advantage. Landsat thermal data is continuous from Landsat 5 TM (1984) through the present, allowing multi-decade trend analysis of whether thermal stress at a site is worsening in line with regional warming. That long baseline is something no airborne or ground survey can replicate.

Typical figures

Finest available thermal resolution (satellite)	~70 m (ECOSTRESS); 100 m native / 30 m resampled (Landsat TIRS); 90 m (ASTER TIR)
Revisit (Landsat 8 + 9 combined)	~8 days at equator under clear sky
Diurnal sampling (ECOSTRESS)	Variable local overpass time due to ISS orbital precession; full diurnal profile built over weeks
LST retrieval uncertainty	~1–2 K (Landsat TIRS, Collection-2); ~1.5 K (ECOSTRESS); better under dry, clear conditions
Thermal spectral bands	Landsat TIRS: 2 bands (10.6–11.2 µm, 11.5–12.5 µm); ECOSTRESS: 5 bands (8.28–12.13 µm); ASTER TIR: 5 bands (8.1–11.7 µm)
Cloud limitation	All passive TIR sensors are blocked by cloud; multi-month compositing required in humid climates
Landsat thermal archive depth	1984–present (Landsat 5 TM onward); free via USGS EarthExplorer
Minimum resolvable feature (thermal)	Approximately 140–200 m for reliable thermal characterisation at Landsat TIRS; sub-site detail requires airborne FLIR
Delivery formats	GeoTIFF LST rasters, classified thermal anomaly shapefiles, time-series CSV per zone, PDF site report

Analytics Satellize can run

Site-wide LST map (single date)	Single-channel or split-window LST retrieval from Landsat TIRS Collection-2, atmospherically corrected	GeoTIFF thermal layer with classified temperature zones; PDF summary for site manager
Diurnal thermal amplitude map	ECOSTRESS multi-overpass compositing to derive daily max/min LST difference per pixel	GeoTIFF amplitude raster; priority zone shapefile ranked by cycling intensity
Seasonal thermal trend analysis	Multi-year Landsat TIRS time series; per-zone mean LST and amplitude trend fitted by season	Time-series chart per defined zone; GIS layer flagging zones with statistically significant warming trend
Biological colonisation risk layer	Fusion of ECOSTRESS thermal stress index with Sentinel-2 crust/lichen index (NDVI, BSI); co-registered at common resolution	Composite risk-score raster; ranked list of façade or surface zones by combined thermal-moisture-biological risk
Drainage anomaly identification	Persistent cool-wet pixel clustering from multi-date LST composites; cross-referenced with site DEM slope and aspect	Annotated map of probable drainage deficiency locations; recommended ground inspection points
Material emissivity differentiation (ASTER)	Temperature-emissivity separation (TES algorithm) applied to ASTER five-band TIR data	Emissivity map distinguishing primary stone types and repair materials; shapefile of anomalous emissivity zones indicating material substitution or moisture infiltration
Visitor-path thermal stress assessment	LST extraction along mapped footpath vectors; comparison with adjacent undisturbed surface temperatures	Table of path segments by thermal excess score; recommended rerouting options with supporting thermal evidence

Sources

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.