Why it matters
Without AI, misreadings risk missing buried features, delaying trench decisions and causing unreported contexts that complicate post-excavation assessment and archiving.
From design to safety
Design
AI aids desk survey and site forecasting
AI analyzes LiDAR DTM, hillshade, and aerial images to predict feature locations, inform written scheme of investigation, and guide trench design priorities with redundancy checks and multitemporal comparisons.
- LiDAR DTM
- hillshade
- aerial photographs
- feature mapping
Build
Guided trench planning and feature targeting
AI outputs targeted coordinates, feature likelihood, and context lists from terrain data to shape trench plans, strip maps, sampling strategies, and contingency access routes.
- cropmarks
- earthworks
- aerial photographs
Operation
Real-time feature validation and mapping
During excavation, AI cross-checks observed features against LiDAR-derived models and historic records, updating context sheets and single context plans with versioned logs.
- feature mapping
- cropmarks
- earthworks
Maintenance
Archive-ready data curation and quality checks
AI flags data gaps, ensures calibration records, and prepares ALARP-compliant archive deposition metadata for post-excavation assessment while validating file integrity and metadata schemas.
- LiDAR DTM
- aerial photographs
- hillshade
Safety
Monitoring for buried services and risk zones
AI integrates LiDAR terrain models with flagged zones to warn on buried services, slope concerns, PPE zoning, and potential UXO during site operations.
- earthworks
- cropmarks
- aerial photographs
Part by part
- 1
LiDAR DTM
- Watch for
- Inconsistent elevation readings indicating terrain shift or data gaps in the model.
- AI action
- AI monitors point-cloud density and vertical accuracy, triggers re-scan orders or alternative data sources.
- 2
hillshade
- Watch for
- Unexplained shading anomalies suggesting terrain occlusions or data gaps.
- AI action
- AI flags anomalies and proposes targeted imagery acquisition.
- 3
cropmarks
- Watch for
- Fading or conflicting cropmark signals across seasonal images.
- AI action
- AI thresholds cropmark confidence and suggests field validation lines.
- 4
earthworks
- Watch for
- Newly exposed earthworks deviating from predicted patterns.
- AI action
- AI updates feature maps and advises additional trenching where needed.
- 5
aerial photographs
- Watch for
- Geometric misalignment or low-resolution imagery.
- AI action
- AI performs co-registration checks and requests higher-res data if required.
- 6
feature mapping
- Watch for
- Misalignment between map features and context sheets.
- AI action
- AI reconciles features with Harris matrix and updates context records.
Worked example
Site X: AI-guided feature detection on a 12-hectare development, Iron Age to Roman layers
Illustrative scenario with sample data.
A development-led site requires rapid assessment of potential ditches and enclosures before trenching to inform the written scheme of investigation and subsequent monitoring.
Signal in
Area: 12 ha; periods: Iron Age to Roman; data: LiDAR DTM 0.5 m grid, hillshade, cropmarks, aerial photos 2025-07-14; initial feature map with provisional contexts and an outline Harris matrix.
System out
AI-generated feature map with 1.2 m tolerance, priority trench list, and context sheets aligned to Harris matrix; archive groundwork planned with metadata schema checks.
Human sign-off: Project manager approves trench priorities, cross-checks with HER and planning conditions, and signs off on the archive deposition plan.
How to roll it out
Phase 1
Weeks 1 to 4: Baseline
Establish data feeds from LiDAR, hillshade, and aerial imagery; configure feature-detection thresholds, multi-temporal templates, and initial map schemas.
Phase 2
Weeks 5 to 12: Integration
Ingest field data, calibrate AI outputs with observed contexts, integrate SME inputs, and refine the written scheme of investigation within permits and conditions.
Phase 3
Weeks 13 onwards: Operational
Maintain feedback loops between field teams, finds processing, and post-excavation planning; ensure records are archive-ready with transfer protocols to HER systems.
Where people stay in charge
- Decisions about interpretation of detected features must be signed off by a qualified person with HER and planning consent oversight.
- AI-generated recommendations on trench locations must be reviewed against written scheme of investigation and documentary evidence by the project manager.
- Data deletion and archive deposition must follow professional standards and regulatory retention schedules, approved by the curator or planning archaeologist.
Questions owners and crews ask
What data sources feed the AI for LiDAR and aerial imagery in archaeology?
The AI combines LiDAR DTM, hillshade models, cropmarks captured from aerial photographs, and up-to-date aerial imagery to inform feature detection and context mapping, with provenance logs for audit.
How does AI help with trench design and targeting?
AI analyzes terrain cues, cropmark patterns, and historical maps to propose high-value trench locations, prioritizing sections with strongest feature likelihood while ensuring compliance with the written scheme of investigation.
Who validates AI outputs in the field?
The project manager and site supervisor review AI-driven maps, cross-check against context sheets, and ensure compliance with the written scheme of investigation, with curator sign-off when required.
What records are produced for post-excavation analysis?
AI-generated feature maps, context sheets, plan and section records, Harris matrix updates, and archiving metadata are prepared for post-excavation assessment and HER deposition.