Dam Inspection Drones: Concrete, Spillway, and Embankment Data

On This Page

• Define the engineering question first
• Match the capture to concrete, spillway, and embankment
• Separate image detail from measurement accuracy
• Build the commercial workflow around site inputs
• Specify deliverables the owner can check
• Questions to settle before awarding the work

Define the engineering question first

Start with the dam's existing inspection and monitoring program. Is the task to locate new surface deterioration, revisit a known joint, map erosion after an event, or measure possible movement? Give the provider those questions and the relevant previous records before asking for an aircraft or sensor proposal.

FERC's Chapter 14 engineering guidance organizes monitoring around potential failure modes and combines observations with instrumentation and engineering evaluation. It concerns FERC-regulated hydropower projects; it is not a blanket specification for every dam. The procurement implication is useful more broadly: connect each requested observation to the person and decision it will support.

Assign interpretation separately from capture. The pilot can supply a photograph of staining; the responsible engineer decides how that observation relates to drainage, previous inspections, instrument readings, and follow-up. Agree on immediate notification for concerning findings instead of allowing them to wait in the final report queue.

Match the capture to concrete, spillway, and embankment

A visible-light camera records surface appearance. Photogrammetry matches features across overlapping photographs to reconstruct geometry and produce a point cloud, textured model, or corrected image mosaic. These outputs serve different purposes: a navigable model locates a finding, while the original close photograph may preserve the detail needed to review it.

In a 2018 field study at Tibble Fork Dam, researchers evaluated targeted image collection with different levels of detail for different inspection features. The practical lesson is to plan broad coverage and selected close views together. Results from that experimental method at one dam do not establish performance at another.

Concrete faces and exposed spillways

Concrete observations can include visible cracking, spalling (loss of surface concrete), deteriorated joints, and staining. Reclamation's photogrammetry research project describes engineers tracing and measuring cracks from dam-face orthophotos, images corrected to support mapping of the surface. Request both the mapped finding and its source image so the reviewer can examine the surface rather than depend entirely on a processed model.

For a spillway, organize coverage by the owner's slab, bay, wall, or joint identifiers. Record the water and operating conditions, including which areas were exposed during capture. A return visit at another water level is a separate coverage task to agree with the owner. Label obscured portions explicitly, and keep an unseen surface distinct from a reviewed surface with no finding. An image of exposed concrete cannot establish the condition beneath a slab or behind a lining.

Similar access and visibility issues appear in bridge inspection drone data. Where close examination or another inspection method is required, include that handoff in the scope.

Embankments, downstream toes, and abutments

For embankments, separate the surface record from any claim about internal condition. Ask for mapped observations of exposed erosion, depressions, visible wet areas, and vegetation that prevents inspection. Specify which surfaces require repeat profiles or elevation comparison, rather than ordering a model of the whole site with no defined use.

Thermal imagery can help locate temperature patterns associated with wet ground, but a temperature pattern alone does not identify the water's source or quantify leakage. Reclamation's seepage-methods report describes how solar heating, time of day, and vegetation affect interpretation. Its canal examples explain relevant physical limits, not a validated seepage-detection rate for every dam.

Treat a suspected wet area as a location for follow-up. Pair the thermal record with visible images and site conditions; have the dam team decide which ground observations or instruments are needed. The same report shows why vegetation can leave gaps or misleading surfaces in photogrammetric ground models. Do not describe vegetation height changes as embankment movement.

Separate image detail from measurement accuracy

Three questions should appear separately in a proposal: what detail can be seen, how accurately a feature can be positioned or measured, and how reliably two surveys can be compared.

Ground sampling distance describes the surface distance represented by an image pixel. It does not, by itself, establish the smallest crack that can be measured reliably. Focus, motion blur, contrast, viewing angle, and the measurement method also matter. Request a representative sample of the proposed detail before accepting a crack-width measurement service.

For mapped geometry, require the coordinate system, vertical reference, control method, and independent checks. Control points help position the model; checkpoints test the result without being used to fit it. The USGS calibration guidelines address calibration, acquisition, quality control, and documentation needed for quantitative use of drone imagery.

Repeat surveys add another requirement: a common, stable reference. Reclamation's photogrammetry project warns that fine detail within a model can coexist with larger georeferencing errors between models. A convincing colored change map therefore needs an explanation of alignment and uncertainty before its differences are treated as movement.

Ask the survey lead to state the minimum change the delivered method can distinguish and the basis for it. Request separate reporting for uncertain or poorly covered areas. If an apparent shift is comparable to the reported uncertainty, describe the result as unresolved rather than labeling it deformation. Neither a dense point cloud nor an aircraft's positioning specification is a substitute for checking the final survey.

Build the commercial workflow around site inputs

Give bidders an input pack containing the asset layout and identifiers, inspection objectives, known findings, earlier imagery or surveys, available control, and intended output formats. Include the operating window, access limits, water-level information, and who will review the results. The inspection scope-of-work guide explains how to turn those inputs into comparable proposals and rework terms.

Use the following sequence as a recommended project structure:

1. Scope with the dam team. Name the components, required observations, measurements, and excluded work. Assign the engineer who will accept the findings.
2. Plan capture and access. Identify the viewpoints, resolution checks, control work, and fallback methods needed for obscured surfaces. Confirm coordination with site operations.
3. Check a representative sample. Review close images, mapping quality, and the proposed finding register before repeating the approach across the site.
4. Review coverage before demobilizing. Reconcile required surfaces with usable captures and record missing or inconclusive areas while another visit may still be avoidable.
5. Process, interpret, and hand over. Preserve originals, document processing, have the designated reviewer assess findings, and record the next action for unresolved observations.

For U.S. work, include location-specific flight checks before committing to the schedule. The FAA identifies restrictions at designated sensitive facilities, including Hoover Dam. Verify the rules for the actual location and operation; the dam owner's access arrangements alone do not establish flight authorization.

Also settle where imagery will be processed and stored, who can access it, and what the owner receives when a portal subscription ends. Use the commercial fleet cybersecurity guide to extend that discussion across devices, accounts, and data transfers.

Specify deliverables the owner can check

The following is a proposed purchasing schedule, not a universal dam-inspection standard. Adapt it with the receiving engineer. Its technical basis is the observation, survey-quality, and interpretation limits described above.

Define a finding ID that stays attached to the same location across inspections, and retain the previous observation when adding a new one. Include delivery of the raw files and an exportable finding register in the contract. A portal can make review convenient, but the owner should specify how it will retrieve and compare the records on the next inspection cycle.

Questions to settle before awarding the work

Ask providers to answer with sample outputs and a site-specific method:

• Which concrete, spillway, and embankment surfaces will be covered, and which require another access method?
• Is each offered measurement supported by a demonstrated method, or is it a visual estimate?
• How will independent checks and repeat-survey alignment be reported?
• What conditions could make a thermal or visual result inconclusive, and who orders a return visit?
• Who interprets findings, and how are urgent observations communicated?
• Which files remain available to the owner after processing and hosting services end?

Choose the proposal that connects the dam team's questions to inspectable records and clearly assigned follow-up. Start with a representative area if the method has not yet demonstrated the required detail or repeatability. Expand once the owner can use the results to make the intended maintenance or investigation decision.

Source notes

• FERC: Engineering Guidelines, Chapter 14: hydropower dam performance monitoring, observations, instrumentation, and engineering review; revision 3, May 2017.
• Bureau of Reclamation: Photogrammetric data accuracy and analysis, project 20105: concrete crack mapping and limitations in comparing georeferenced models.
• Barrett: Field Validation of an Advanced Autonomous Method of Exterior Dam Inspection: 2018 master's thesis reporting a targeted image-acquisition study at Tibble Fork Dam.
• Bureau of Reclamation: Select Techniques for Detecting and Quantifying Seepage from Unlined Canals: report linked from its official project record; thermal, vegetation, and topographic interpretation limits in a canal context.
• USGS: Guidelines for Calibration of Uncrewed Aircraft Systems Imagery: 2023 technical guidance on acquisition, calibration, quality control, and documentation.
• FAA: Critical Infrastructure and Public Venues: flight restrictions at designated sensitive locations.

Last checked: September 10, 2026.