Dam Safety: Review of Geophysical Methods to Detect Seepage and Internal Erosion in Embankment Dams

Several non-intrusive geophysical methods are available to facilitate early detection of seepage, piping, and internal erosion in embankment dams. A review of these methods shows where they can be applied and indicates work needed to further improve the use of each.

Internal erosion is the second largest cause of failure of earthfill dams worldwide. Damages resulting from internal erosion can lead to expensive remediation. Typical dam safety surveillance consists of visual inspections supported by limited instrumentation. However, internal erosion can become quite advanced before the problem is detected via these means. Recently, interest has grown regarding the use of non-intrusive geophysical techniques to facilitate early detection of anomalous seepage, piping, and internal erosion.

To date, the use of geophysical methods to investigate seepage in dams has produced mixed results, partly because the application of these methods is not well-understood and partly because false positives cannot be tolerated. Although geophysical anomalies are easily detected, often what these anomalies represent and their implications are not clear. The application of geophysical methods to dams is in its early stages, and adapting geophysical techniques to geotechnical investigations and dam safety surveillance requires more refinement to answer specific engineering questions.

These needs prompted the launch of a collaborative research project under the auspices of CEATI’s Dam Safety Interest Group (DSIG) to study the current state-of-practice regarding geophysical methods applied to embankment dams. The objective of this project was to evaluate, adapt, and/or develop some of the most promising geophysical techniques as investigation and monitoring tools to detect seepage and internal erosion.

Four techniques were selected for additional research and development:

Temperature measurement

Temperature measurement makes use of natural seasonal temperature variations to locate areas of preferential seepage. Temperature in the saturated part of an embankment dam primarily is governed by the temperature of the water seeping from the reservoir. However, the air temperature from above and geothermal heat flow from below also influence temperature distribution in the dam. Geothermal heat flow is relatively constant, but air and reservoir temperatures vary seasonally and create temperature “waves” that penetrate the dam. Conductive air temperature variations typically penetrate about 10 meters below the dam surface along the crest and downstream slopes. Upstream, reservoir water exhibits seasonal fluctuations that are influenced by stream inflows and mixing. Stratification often exists in large reservoirs, and variations up to about 20 degrees Celsius (C) can occur in the upper tens of meters of the reservoir, with little seasonal fluctuation at depth. Figure 1 shows the effect of seasonal fluctuations on a vertical temperature profile measured in a deep reservoir in northern British Columbia.

Detection of concentrated seepage through an embankment dam generally relies on measuring the attenuating temperature pulse at a given location as the reservoir water seeps through the embankment. Temperature fluctuations can be used to identify higher-permeability zones that could indicate damage. Zones of high temperature variation indicate an area of higher seepage flow than areas with a lower temperature variation. For interpretation, it is critical to monitor and understand the temperature cycles at all depths in the reservoir adjacent to the dam.

The top elevation of these vertical temperature profiles, measured in the reservoir impounded by a dam in British Columbia, corresponds with the reservoir level. Seasonal fluctuations in temperature that are evident in the upper 60 meters of the reservoir facilitate the detection of higher-permeability zones in the embankment damn.

The temperature within an embankment dam can be measured at discrete points by using thermistors (such as those integrated into vibrating wire piezometers) or by profiling the water column inside standpipe piezometers or existing casings. Distributed temperature measurements using optical fibers bring the promise of improved spatial coverage and enable monitoring with an accuracy of 0.01 to 0.1 C at spacings of about 1 meter over a continuous fiber of 10 kilometers or more. Costs of readout units are US$25,000 to US$75,000 (depending on specification requirements).

Data can be interpreted qualitatively or quantitatively. As part of the DSIG research project, a user-friendly time lag software package called DamTemp was enhanced. This software has the capability to use measured temperatures to identify and estimate the seepage flows in a zone of potential damage.

As an extension to the DSIG research work, practical guidance for temperature data measurement and evaluation procedures have been documented in a field manual.1

With significant advances in temperature monitoring and interpretation tools over the past two decades, temperature measurements are rapidly gaining acceptance as a useful method for monitoring seepage in embankment dams. This is particularly true in Sweden and other European countries.

Self-potential method

Self-potential (SP) is a passive technique that measures naturally occurring electrical potentials in the ground. This is the only one of these four geophysical techniques that responds directly to fluid flow. Water flowing through the pore space of soil generates electrical current flow. This electrokinetic phenomenon is called streaming potential and gives rise to SP signals that are of primary interest in dam seepage studies.

Field data acquisition

SP is measured by determining the voltage across a pair of non-polarizing electrodes using a high-impedance voltmeter. This inexpensive and deceptively simple data acquisition procedure requires special care and attention in order to reliably interpret and correct for sources of electrical noise that can mask the signal of interest. All noise sources — including time-varying telluric currents associated with solar and atmospheric activity, stray currents, and the corrosion of buried metal — must be recognized and measured. These noise sources can mask the relatively small signals associated with seepage anomalies. For this reason, telluric measurements and magnetic surveys should be carried out to assist in interpreting the SP data. Typically, SP anomalies on the order of tens of millivolts are associated with seepage anomalies of interest, although anomaly amplitudes largely depend on site-specific conditions.

Each of these self-potential profiles from a dam in British Columbia corresponds to potential difference measured between a given electrode and a base electrode positioned at the center of the dam crest. There is a strong correlation between the self-potential data and seasonal reservoir levels. 

The objectives of the survey and the nature of site conditions dictate the choice of SP survey configuration and layout. Distance between electrodes typically ranges from several meters to tens of meters, depending on the resolution required. Unlike other geophysical techniques, pre-assembled sets of SP survey equipment are not commercially available, and widely accepted data quality-control standards and procedures had not been established for the SP method. As a result of this research, guidance on obtaining high-quality SP data in support of dam seepage investigations has been comprehensively documented in an SP field data acquisition manual.2

SP data interpretation

Interpretation of SP measurements to infer seepage patterns and concentrated seepage flows ranges from simple qualitative to more advanced quantitative numerical modeling approaches.

Zones of preferential flow can be inferred qualitatively using patterns in the electrical potential distribution. Interpretation of seepage-related features is aided by taking the difference between two data sets collected at different pool levels. This process reduces the influence of non-seepage sources — such as electrical potential fields associated with buried metal pipes and concrete rebar — and thus facilitates the identification of seepage-related anomalies. Distinct anomalies can be interpreted using simple geometric source modeling to estimate the location and depths of seepage-induced electrical current sources. This information can be used in conjunction with other site information to further delineate the extent and cause of the seepage, or to help guide more detailed investigations.

The current state-of-the-art in SP data interpretation is application of more advanced numerical modeling techniques to interpret characteristics of the hydraulic regime from the SP data. A three-dimensional (3D) forward modeling software package called SP3D was developed as part of the DSIG project. This program enables an interpretation of hydraulic head patterns from the geophysical data using a 3D seepage model of the dam. This level of data interpretation requires estimates of the hydraulic conductivity, electrical resistivity, and cross-coupling coefficient of the embankment materials.

A lack of available data on the electrical resistivity and cross-coupling coefficient of well-graded soils prompted a laboratory study to measure these parameters. A unique apparatus was developed to perform streaming potential and resistivity measurements on the same soil specimen to derive the cross-coupling coefficient. Both unidirectional and cyclic flow methods were used to perform streaming potential measurements. The cyclic method was shown to be a valid test method and the most efficient technique for measuring the streaming potential coupling coefficient in soils.

The influence of soil and fluid properties on the cross-coupling coefficient was investigated for typical embankment soils. The results show that this coefficient does not vary considerably in saturated soils as compared to other properties such as electrical resistivity.3 This suggests that practitioners may not need to characterize the cross-coupling coefficient to the same degree as electrical resistivity for practical SP field data interpretation.

Practical guidelines for interpreting SP data resulting from dam seepage investigations have been developed.4

Figure 2 illustrates the temporal var-iations evident in an SP data set collected using an array of electrodes installed along the crest of a dam in British Columbia. This monitoring array was deployed to obtain information about the seasonal SP time variation within the dam and to assess the long-term performance of the prototype system. The data shown in Figure 2 are all referenced to a common base station at the center of the dam crest. The SP signals vary with changes in the seepage flow through the dam as the reservoir level cycles.

Electrical resistivity

The direct current resistivity method has well-established data acquisition and interpretation techniques for standard survey configurations. The method uses pairs of electrodes to inject current into the ground and measure the resulting electrical potential distribution. Its application to dam seepage investigations is two-fold. The method may be used to monitor spatial and/or temporal variations in electrical resistivity in response to changing soil conditions caused by internal erosion and anomalous seepage. The method also may be used to characterize the electrical resistivity of the subsurface for the purposes of interpreting SP data.

Inverse modeling methods are preferred for interpreting an electrical resistivity distribution from the geophysical data. The interpretation of electrical resistivity data acquired using a single line of electrodes along the crest of an embankment poses a challenge due to the sloping geometry of the dam. Two-dimensional interpretations may misrepresent the true resistivity at depth. However, monitoring applications are not adversely affected as the focus of these investigations is to detect changes in resistivity with time, which may be linked to the development of internal erosion in the core of the embankment.

A report is available that provides detailed, practical guidance on resistivity survey design and equipment, data acquisition, and data interpretation for embankment dams.5

Embankment and reservoir conditions are dynamic. Fluctuations in pool levels, seasonal temperatures, and total dissolved solids all affect the electrical properties of the embankment, particularly its electrical resistivity. Long-term monitoring affords an increased sensitivity to temporal changes and enables more effective identification of local changes that may be linked to the development of internal erosion. In long-term monitoring applications, a large amount of data is collected and processed, such that efficient data handling becomes a special requirement. Case histories of long-term monitoring measurements in Sweden, using electrical resistivity, temperature, and SP methods, are available. These case histories illustrate the significant effect of seasonal variations on the measured data and provide insight for the design and installation of permanent monitoring arrays.6


Figure 3 shows temporal and spatial variations in resistivity interpreted from data collected using a long-term monitoring array of electrodes installed along the crest of a dam in Sweden. Values of resistivity at a depth of 20 meters below the crest have been interpreted from the raw data measured from the array at four stations along the dam crest. The lower resistivity and higher variation evident in the profile at location (chainage) 450 meters indicates the presence of an eroded zone.

Seismic methods

Common seismic techniques include refraction, reflection, downhole, and cross-hole methods. With all these techniques, the time required for seismic energy to propagate from its source to a receiver is measured. If the length of the travel path is known, the velocity of the seismic energy can be derived. The seismic velocity can be used to garner information about soil stiffness and density. In dam seepage applications, internal erosion can cause low-stress conditions, which can manifest as zones of low seismic velocity.

Cross-hole seismic tomography has been used to better define the configuration of sinkholes at WAC Bennett Dam in northern British Columbia. Results suggested that a through-dam seismic configuration not requiring drill holes also might be capable of detecting sinkholes and/or zones of internal erosion. This procedure makes use of the geometry of the dam to image conditions beneath the crest by propagating seismic waves from the upstream to the downstream slope, or vice versa.

Two types of body waves can propagate through a medium. Compressional or P-waves relate to changes in the volume of a medium. Shear or S-waves relate to the distortional changes of a medium. (Surface waves, such as Rayleigh and Love waves, exist in an elastic half-space but are less commonly exploited for geotechnical purposes.)

Generally, shorter wavelength sources provide better resolution, thus S-waves are preferred for geotechnical applications. However, S-waves tend to attenuate more rapidly than P-waves, and it is more difficult to generate high-energy S-waves. Seismic vibrator sources (e.g. Vibroseis) have been shown to generate and propagate S-wave energy across distances of more than 120 meters in a zoned earthfill dam.

Interpretation of through-dam data can range from simple to complex. In the common station gather approach, the travel path is assumed to be a straight line between source and receiver. The simplicity of this interpretation is at the cost of resolution, and only the average velocity between the source and receiver is obtained. For repeat testing or ongoing monitoring, this may be sufficient to detect a change in condition. If a change is detected, more sophisticated data interpretation and more comprehensive field testing could be initiated.

The seismic velocities measured from the field testing can be used to infer density, stress, and saturation conditions.7 It is interesting to note that P-waves should not have been capable of detecting the sinkholes at WAC Bennett Dam due to their longer wavelength. However, P-wave testing clearly detected an anomaly, which was interpreted and confirmed as a zone of lower stress surrounding the sinkholes.


Geophysical methods are useful as non-destructive remote sensing tools that can provide information over large volumes as compared to point measurements. However, the anomalies of interest that are associated with internal erosion in embankment dams often are very small. The effectiveness of geophysical techniques to detect changes in seepage conditions is improved through repeating surveys or adopting a long-term monitoring approach. In addition, application of more than one geophysical technique will provide added confidence in the interpretation and detection of anomalous features.

The CEATI study showed that complex inter-relationships exist between various parameters such as water content, porosity, total dissolved solids, mineralogy, temperature, electrical resistivity, coupling coefficient, and SP.8 Not recognizing some of the fundamental relationships and carrying out a one-time survey without supporting information could lead to misleading and often disappointing results.

Geophysical techniques applied to the detection of seepage and internal erosion in embankment dams are at various stages of development. Temperature appears to be one of the most-developed and best-understood techniques. With the recent advances in improved accuracy and resolution in measuring temperatures along fiber optic cables, there are exciting possibilities.

For dam safety applications, SP and resistivity methods generally appear to hold more promise than seismic methods as non-intrusive techniques applied at the surface of a dam. However, in specific settings, cross-hole seismic techniques could prove indispensable.

Although the understanding of the SP and resistivity methods as applied to embankment dams has come a long way in recent years, more research is required before these techniques can enter into standard practice and be applied with confidence on a routine basis. It is imperative that the dam owner and practicing engineer recognize the limitations and the care required in planning, executing, and interpreting the results. Geophysical data interpretation is non-unique and should be constrained by incorporating all available site information and integrating the interpretation of complementary data sets. Thus, strong cooperation between the geophysicist and engineer is essential to improve the interpretation and usefulness of the results.


  1. Johansson, S., and P. Sjodahl, “A Guide to Temperature Measurements for Seepage Investigation and Monitoring of Embankment Dams,” T062700-0214, CEATI, Montreal, Quebec, Canada, 2009.
  2. Corwin, R.F., “Self-Potential Field Data Acquisition Manual,” T992700-0205B, CEATI, Montreal, Quebec, Canada, 2005.
  3. Sheffer, M.R., “Laboratory Testing of the Streaming Potential Phenomenon in Soils for Application to Embankment Dam Seepage Investigations,” T992700-0205B/2, CEATI, Montreal, Quebec, Canada, 2005.
  4. Corwin, R.F., “Interpretation of Self-Potential Data for Dam Seepage Investigations,” T992700-0205B/3, CEATI, Montreal, Quebec, Canada, 2007.
  5. Dahlin, T., P. Sjodahl, and S. Johansson, “A Guide to Resistivity Investigation and Monitoring of Embankment Dams,” T992700-0205B/4, CEATI, Montreal, Quebec, Canada, 2008.
  6. Johansson, S., J. Friborg, T. Dahlin and P. Sjodahl, “Long-term Resistivity and Self-Potential Monitoring of Embankment Dams — Experiences from Hà£llby and Sà£dva Dams, Sweden,” T992700-0205C, CEATI, Montreal, Quebec, Canada, 2005.
  7. Gaffran, P., and M. Jeffries, “A Study of Through-Dam Seismic Testing at WAC Bennett Dam,” T992700-0205E, CEATI, Montreal, Quebec, Canada, 2005.
  8. Johansson, S., J. Friborg, J. Claesson, T. Dahlin, G. Hellstrom, and B. Zhou, “A Parameter Study for Internal Erosion Monitoring,” T992700-0205A, CEATI, Montreal, Quebec, Canada, 2005. 

CEATI’s geophysical methods research 

In 1999, a group of dam owners, engineers, and geophysicist from Canada, the U.S., and Europe met to evaluate the state of practice and identify research needs in the use of geophysical methods. Participants in the “Internal Diagnostics for Embankment Dams” workshop identified temperature, self-potential (SP), resistivity, and seismic techniques as having the greatest potential for identifying anomalous seepage and deteriorating conditions within embankment dams. This lead to initiation of the CEATI Dam Safety Interest Group (DSIG) research project, “Investigation of Geophysical Methods for Assessing Seepage and Internal Erosion in Embankment Dams.” The project was sponsored by BC Hydro, Elforsk AB, Great Lakes Power Ltd., Hydro-Quebec, Manitoba Hydro, New Brunswick Power Generation Corp., New York Power Authority, Ontario Power Generation, and the U.S. Department of the Interior’s Bureau of Reclamation.

Results of the research project are documented in nine reports:

Two computer programs also were developed for interpreting temperature and SP data:

Ken Lum, a principal engineer at BC Hydro, is project manager for CEATI’s research project. Megan Sheffer, a senior engineer at BC Hydro, is a principal investigator in the study. 

This article has been evaluated and edited in accordance with reviews conducted by two or more professionals who have relevant expertise. These peer reviewers judge manuscripts for technical accuracy, usefulness, and overall importance within the hydroelectric industry. 

More Hydro Review Current Issue Articles
More Hydro Review Archives Issue Articles

Previous articleNational Register of Historic Places Application for Grand Rapids
Next articleGeorgia auctions six hydro plants, three utilities

No posts to display