Case Study: as Rs-based Damage Detection in Heritage Buildings

Heritage buildings are irreplaceable cultural assets, and their preservation demands constant vigilance. Traditional inspection methods often involve scaffolding, manual surveys, and even invasive techniques that can themselves cause damage. Remote sensing (RS) technology offers a non-contact, scalable alternative that is transforming how we monitor and maintain these historic structures. By capturing data from drones, satellites, and aerial platforms equipped with advanced sensors, conservation teams can now detect cracks, material degradation, and moisture infiltration long before they become critical. This case study examines a recent application of RS-based damage detection on a historic monument, demonstrating how multispectral and thermal imaging, combined with analytical algorithms, can provide early warnings and guide targeted interventions.

The Need for Non-Invasive Damage Detection in Heritage Conservation

Heritage buildings face a unique set of challenges: aging materials, exposure to environmental stresses, urban pollution, and the inevitable wear of time. Structural failures, water damage, and biological growth often develop slowly and remain hidden until they cause visible harm. Traditional inspection relies on visual assessments and point measurements, which are time-consuming and can miss subsurface issues. Moreover, many heritage sites have fragile surfaces that cannot tolerate direct contact or chemical testing. Remote sensing fills this gap by capturing detailed, repeatable data from a safe distance. It enables conservationists to create baseline records, monitor changes over years, and prioritize repairs based on objective evidence. For a deep dive into the principles of heritage monitoring, see ICCROM’s comprehensive guide on heritage monitoring.

Case Study Background: A Historic Urban Monument

The subject of this study is a 19th-century stone building located in a densely populated city center. It has served various civic and cultural roles over its lifetime and is listed as a protected heritage structure. Over the past decade, the building exhibited signs of surface cracking and spalling, especially around window arches and load-bearing walls. Conservation authorities desired a thorough assessment without erecting scaffolding, which would disrupt public access and risk damaging ornate stonework. They turned to remote sensing as a low-risk, high-resolution solution.

Methodology: From Flight Planning to Data Interpretation

The research team designed a multi-phase approach that integrated drone-based data acquisition with advanced image processing. The methodology was carefully tailored to the building’s geometry, material composition, and environmental exposure.

Sensor and Platform Selection

A quadcopter drone equipped with a gimbal-stabilized payload was used. The payload consisted of:

• Multispectral camera – capturing four spectral bands (blue, green, red, near-infrared) at 2 cm ground resolution.
• Thermal infrared camera – operating in the 8–14 µm range, with a thermal sensitivity of 0.05°C, providing surface temperature maps.
• High-resolution RGB camera – for orthophoto generation and visual reference (0.5 cm resolution).

These sensors were chosen because they are known to detect damage indicators: multispectral imagery reveals vegetation stress and moisture patterns, while thermal imagery highlights thermal anomalies caused by moisture, delamination, or differing material densities.

Flight Mission Design

Two flight missions were conducted: one in the early morning (low ambient temperature, high thermal contrast) and one at midday (maximum solar loading). The drone followed a pre-programmed grid pattern at an altitude of 25 meters, maintaining a forward overlap of 80% and side overlap of 70% for robust photogrammetric reconstruction. All flights were performed under low wind conditions and in compliance with local aviation regulations. Ground control points were placed around the building to ensure georeferencing accuracy within 1 cm.

Data Preprocessing

After each flight, raw images were radiometrically calibrated and orthorectified. Multispectral bands were co-registered using tie points and SIFT-based alignment. Thermal images required manual correction for lens distortion and non-uniformity. The processed data were assembled into dense point clouds and digital surface models using structure-from-motion software. For a technical overview of drone photogrammetry workflows, refer to ASPRS publications on unmanned aerial systems.

Damage Detection Algorithms

Three complementary analysis techniques were applied to the derived datasets:

Vegetation and Moisture Indices (NDVI and NDWI)

Normalized Difference Vegetation Index (NDVI) was calculated from the red and near-infrared bands to detect biological growth (moss, algae) on stone surfaces, which can retain moisture and accelerate decay. The Normalized Difference Water Index (NDWI) used green and NIR bands to highlight areas of elevated moisture content. Both indices were thresholded to produce binary maps of potential risk zones.

Thermal Anomaly Detection

Surface temperature distributions were normalized to account for diurnal variation. Pixels whose temperature deviated beyond two standard deviations from the local mean were flagged as thermal anomalies. These often correspond to voids behind the facades, water intrusion, or areas of differing thermal mass (e.g., repaired sections).

Geometric Deformation Analysis

The digital surface model was compared to an ideal, conservational reference model (derived from archival drawings). Deviation maps were generated, revealing subtle bulges and depressions indicative of structural movement. Local changes in curvature were used to detect crack initiation points.

Results and Validation: What the Data Revealed

The RS-based analysis identified 14 distinct areas of concern. The most significant findings are outlined below:

Finding	Detection Method	Confirmation (Ground Truth)	Urgency
Active crack in SE parapet	Geometric deviation > 3 cm	Visual inspection (confirmed width ~4 mm)	High – risk of water ingress
Moisture infiltration around NW window	NDWI high, thermal cold spot	Moisture meter (18% MC)	Medium – requires sealing
Biological growth on north facade	NDVI > 0.25	Surface swab (algae present)	Low – cosmetic but can worsen
Delamination on decorative cornice	Thermal hot spot (air gap)	Tap test (hollow sound)	High – risk of falling debris

Ground-truth surveys by a structural engineer confirmed 12 of the 14 flagged areas (86% accuracy). The two false positives were later explained by temporary surface conditions (leaves and shadows) that the algorithms had misinterpreted. Overall, the RS approach provided a reliable, spatially comprehensive damage map that would have taken weeks of manual inspection to produce.

Why Remote Sensing Works for Heritage Buildings

This case study reinforces several advantages of RS over conventional methods. First, the non-contact nature eliminates the risk of abrasion or chemical damage to fragile surfaces. Second, the ability to capture data across multiple spectral bands reveals invisible indicators of deterioration, such as early-stage moisture or biological colonization. Third, repeated flights allow for consistent monitoring – the same dataset type can be acquired annually to track progression and evaluate intervention effectiveness. Fourth, RS data can be integrated into a digital twin, enabling simulation of environmental loads and structural behavior over time. A recent paper in the Journal of Cultural Heritage discusses similar integrated approaches: see their special issue on remote sensing for heritage.

Cost and Time Efficiency

While the upfront cost of drone equipment and software can be substantial, the per‑inspection cost is far lower than erecting scaffolding. In this case study, the total fieldwork duration was less than six hours, including two flights and ground control setup. Image processing and analysis required another two days. In contrast, a full manual scaffold inspection would have taken over two weeks and cost nearly three times as much. For large or inaccessible structures, the savings can be even greater.

Challenges and Limitations

Despite its promise, RS-based damage detection is not a panacea. Several practical constraints must be acknowledged:

• Weather dependence: Flights require calm winds and no precipitation. Cloud cover can affect thermal images, and low light limits multispectral accuracy.
• Regulatory hurdles: Flying drones in urban heritage zones often requires special permits. Privacy concerns may also arise in populated areas.
• Data volume and processing: High-resolution multi-band datasets can exceed 50 GB per flight. Processing them demands powerful computers and skilled operators.
• Interpretation expertise: Algorithms can flag anomalies, but validating them as true damage still requires experienced conservators. RS is a complement, not a replacement, for expert judgment.
• Subsurface limitations: Thermal imaging only penetrates a few millimeters. Internal structural issues, such as dry rot in timber supports, may remain hidden unless other evidence points to them.

Best Practices for Implementing RS in Heritage Monitoring

Based on the lessons of this case study, conservation teams should consider the following guidelines:

1. Start with a pilot study on a small, accessible portion of the building to calibrate sensor settings and algorithms against known conditions.
2. Combine multi-sensor data – no single band or index is sufficient. Fusing multispectral, thermal, and geometric data provides a more complete picture.
3. Establish a baseline before any repairs are made. Repeat surveys at consistent times of year to ensure comparability.
4. Integrate with traditional inspections – use RS to identify high-priority zones, then direct manual inspection there for detailed assessment.
5. Document metadata thoroughly (sun angle, humidity, temperature during flight) to help interpret variations over time.

Future Directions: AI and Autonomous Monitoring

The next frontier in RS-based damage detection is the application of deep learning to automatically classify and measure damage features. Convolutional neural networks (CNNs) are already being trained on building crack databases to achieve detection accuracies above 95%. Combined with automated drone flight path planning and on‑board processing, the technology could soon enable near‑real‑time structural health monitoring. Researchers are also exploring the use of Synthetic Aperture Radar (SAR) satellites to detect millimeter‑scale deformations in entire historic districts over months and years. For a forward‑looking perspective, the latest work in this field is compiled by ICOMOS.

Conclusion: A Tool for Proactive Preservation

Remote sensing technology has matured to the point where it can be a standard element of heritage building management. This case study demonstrates that drone‑based multispectral and thermal imaging, combined with geometric analysis, can reliably detect a wide range of incipient damage mechanisms. The non‑invasive nature, scalability, and repeatability of RS make it an ideal complement to traditional conservation practices. As sensor costs decline and AI interpretation becomes more robust, even small historical societies can adopt these methods. The ultimate goal is to shift heritage maintenance from reactive repair to proactive, data‑driven stewardship. When we can see damage before it becomes visible to the naked eye, we buy precious time for the world’s most treasured structures.