Mass concrete curing thermal anomaly monitoring for dams
Hydration of mass concrete in gravity and arch dams generates internal heat that can crack the structure if uncontrolled. Satellite and airborne thermal infrared sensors can track surface temperature gradients as a proxy for curing progress, with honest caveats about pixel size.
Sensors
- Landsat 8/9 TIRS (USGS/NASA): Two thermal bands (Band 10 ~10.6–11.2 µm, Band 11 ~11.5–12.5 µm) at 100 m native resolution, resampled to 30 m in products. Revisit 16 days per satellite, ~8 days combined. Freely available. At 100 m ground sampling, a single pixel covers a 1-hectare area; only the largest dam faces span more than one or two pixels, making monolith-level discrimination impossible.
- ECOSTRESS (NASA/ISS): Multispectral thermal instrument on the International Space Station, covering 8.28–12.13 µm across five bands at approximately 70 m × 38 m pixel footprint. Non-sun-synchronous ISS orbit gives irregular revisit (roughly every few days at mid-latitudes) and captures diurnal variation that Landsat cannot. Land Surface Temperature products are publicly archived. Resolution is still marginal for individual dam monoliths, which are typically 15–20 m wide.
- ASTER TIR (NASA/METI): Five thermal bands (8.125–11.65 µm) at 90 m spatial resolution, 16-day revisit. On-demand acquisition only since 2008; ASTER is in instrument-preservation mode and new tasking is limited. Useful for historical archive comparisons on large structures. The 90 m pixel is comparable to TIRS and carries the same coarseness problem for dam monitoring.
- Airborne thermal infrared (general): Uncooled or cooled microbolometer cameras flown on UAV or manned aircraft can achieve sub-metre to 10 cm ground sampling distance at low altitude, resolving individual monolith joints and lift surfaces. Not a satellite product, but the correct tool when monolith-level thermal mapping is actually required. Satellize can integrate airborne thermal data into the same delivery pipeline as satellite layers.
Why mass concrete generates a thermal problem in the first place
Portland cement hydration is exothermic. In a thin structural member the heat dissipates quickly; in a mass concrete dam pour, where the thermal diffusivity of concrete is roughly 0.003–0.004 m²/hour, heat accumulates in the core faster than it can escape through the surface. Internal temperatures in large gravity dam lifts routinely reach 40–60 °C above ambient during peak hydration, typically within the first three to seven days after placement.
The gradient between the hot core and the cooler surface creates tensile stress. When that stress exceeds the tensile strength of young concrete, which is low, cracking follows. The consequences range from cosmetic surface crazing to through-cracks that compromise water-tightness and structural integrity. Thermal monitoring during curing is therefore not optional on any serious dam project; the question is which sensors are accurate and spatially adequate for the job.
What Stefan-Boltzmann actually tells a satellite sensor
Every surface emits radiation proportional to the fourth power of its absolute temperature and its emissivity. For concrete, emissivity in the 8–14 µm atmospheric window is typically 0.92–0.96, which is well characterised. A satellite thermal sensor measuring at-sensor radiance can, after atmospheric correction and emissivity assignment, retrieve Land Surface Temperature to an accuracy of roughly ±1–2 K under clear-sky conditions for Landsat TIRS and ECOSTRESS. That precision is adequate to detect the elevated surface temperatures of a freshly poured lift relative to an adjacent cured lift or the surrounding rock abutment.
The catch is that what the sensor sees is skin temperature, the outermost millimetres of the surface. The dangerous thermal gradient is internal, between the core and the surface. Surface temperature is a proxy, not a direct measurement. A dam face that looks thermally uniform from orbit may still have a steep internal gradient if insulating formwork or curing blankets are in place. Conversely, a surface hot spot visible from orbit could indicate a lift that has lost its curing blanket prematurely. Interpretation requires knowledge of the construction sequence.
The resolution problem, stated plainly
Landsat TIRS pixels are 100 m on a side before resampling. A typical gravity dam monolith is 15–20 m wide. A single TIRS pixel therefore integrates the thermal signal from five or six monoliths simultaneously, plus any adjacent water, rock or construction equipment within the footprint. Detecting a thermal anomaly in one monolith against the average of its neighbours is not feasible at this resolution unless the anomaly is both large and extreme.
ECOSTRESS does better at roughly 70 m × 38 m, and its irregular revisit occasionally catches a pour within the first 24–48 hours when surface temperatures are highest. ASTER at 90 m sits between the two. None of these sensors can replace embedded thermocouples or distributed fibre-optic temperature sensing for monolith-level quality control. What satellite thermal imagery can do is provide an independent, spatially continuous surface record across the entire dam face and its immediate surroundings, flag gross anomalies, and give a project owner a verifiable archive that does not depend on the contractor's own instrumentation.
For projects where monolith-level resolution is genuinely required, airborne thermal surveys at sub-10 cm ground sampling distance are the appropriate tool. Satellite data then serves as the wide-area context layer.
Acquisition timing matters more than sensor choice
Peak hydration heat in a typical mass concrete lift occurs within 72 hours of placement. Landsat's 8-day combined revisit means there is roughly a 50 per cent chance of a cloud-free overpass within that window at any given location, and no guarantee of one. ECOSTRESS's non-sun-synchronous orbit improves the odds and adds nighttime acquisitions, which are particularly useful because solar heating of the surface is absent, making thermally elevated concrete more distinguishable from its surroundings.
Cloud cover is the other hard constraint. Thermal infrared does not penetrate cloud. In tropical or monsoon-affected dam construction sites, extended cloud cover periods can eliminate satellite thermal observation entirely during the most critical curing phases. Honest programme design acknowledges this and pairs satellite monitoring with ground instrumentation rather than substituting for it.
Pre-construction baseline imagery, acquired before any concrete is placed, establishes the ambient thermal signature of the rock foundation, river and surrounding terrain. Change detection against that baseline is more reliable than any absolute temperature threshold, because local emissivity variations and topographic shading effects are automatically removed.
What a practical satellite thermal monitoring programme looks like
A workable approach combines automated download of every available Landsat and ECOSTRESS overpass for the dam site, atmospheric correction using standard algorithms (the USGS Collection 2 LST product for Landsat, or the ECOSTRESS Level-2 LST product from NASA Earthdata), and differencing against the pre-construction baseline. Anomalies exceeding a defined threshold, say 5 K above the baseline for the same season and time of day, trigger a flagged report with the acquisition geometry and cloud-mask quality score attached.
The output is not a substitute for the dam engineer's embedded sensor network. It is an independent audit layer. A project financier, an independent engineer or a government dam-safety regulator can use it to verify that the contractor's reported curing schedule is consistent with the observed surface thermal record. That audit function has value even when the pixel resolution is coarse, because gross deviations, such as a lift left uninsulated in cold weather or a section showing no thermal elevation when fresh concrete should be hot, are visible at 100 m resolution.
Satellize runs this kind of change-detection pipeline on open constellations including Landsat and ECOSTRESS, with commercial tasking added where a client licence permits. The methodology is the same published split-window and single-channel retrieval approach used in peer-reviewed remote sensing literature, applied operationally rather than as a research exercise.
Typical figures
| Spatial resolution (Landsat 8/9 TIRS) | 100 m native, resampled to 30 m in USGS products; single pixel spans multiple dam monoliths |
| Spatial resolution (ECOSTRESS) | ~70 m × 38 m per pixel; marginal for individual monoliths (typically 15–20 m wide) |
| Spatial resolution (ASTER TIR) | 90 m; on-demand tasking limited since 2008, instrument-preservation mode |
| Revisit (Landsat 8 + 9 combined) | ~8 days at equator under clear sky; cloud cover may eliminate usable acquisitions for weeks |
| Revisit (ECOSTRESS on ISS) | Irregular, roughly every 1–5 days depending on latitude; includes nighttime passes |
| LST retrieval accuracy (clear sky) | ±1–2 K for Landsat TIRS and ECOSTRESS under standard atmospheric conditions |
| Spectral bands used | 8–14 µm thermal infrared atmospheric window; concrete emissivity typically 0.92–0.96 |
| Archive depth | Landsat: 1972 to present (TIR from 1982 with TM); ECOSTRESS: 2018 to present; ASTER: 2000 to present |
| Cloud penetration | None. Thermal infrared is blocked by cloud; SAR is required for all-weather surface observation |
| Minimum detectable anomaly (practical) | Gross lift-scale anomalies (>5 K surface deviation across >1 pixel) detectable; sub-monolith anomalies are not |
Analytics Satellize can run
| Pre-construction thermal baseline map | USGS Collection 2 LST retrieval or ECOSTRESS Level-2 LST; multi-date median compositing to remove transient effects | GeoTIFF baseline raster with seasonal statistics, delivered before first concrete placement |
| Pour-event thermal anomaly flag | Change detection: post-pour LST minus baseline for equivalent season and solar time; threshold alert at user-defined K deviation | Automated alert with acquisition date, cloud-mask score, and anomaly extent polygon in GIS layer |
| Curing progression time series | Multi-date LST differencing across all available Landsat and ECOSTRESS overpasses during construction period | Time-series chart per dam face zone showing surface temperature evolution against expected hydration curve, in PDF and CSV |
| Nighttime thermal signature extraction | ECOSTRESS nighttime acquisitions processed to remove solar contribution; anomalies interpreted against ambient air temperature from ERA5 reanalysis | Annotated thermal image with temperature-difference overlay, flagging zones of unexpectedly high or low surface emission |
| Independent audit report for project finance or dam-safety review | Archive compilation of all usable overpasses, quality-scored by cloud cover and acquisition geometry, compared against contractor-reported pour schedule | Structured PDF report with satellite evidence, data-gap disclosure, and consistency assessment against reported construction sequence |
| Airborne-satellite data fusion layer | Co-registration of high-resolution airborne thermal raster (client-supplied or commissioned) with Landsat/ECOSTRESS context layer | Merged GeoTIFF at airborne resolution with satellite temporal context, compatible with standard GIS platforms |
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.