SAGEEP 2011 Charleston, South Carolina USA http://www.eegs.org

PANAMA CANAL EXPANSION PROJECT: HOW MARINE ELECTRICAL RESISTIVITY WAS USED IN SUPPORT OF CANAL DREDGING

Gillian Noonan, hydroGeophysics, Inc., Tucson, AZ Dale F. Rucker, hydroGeophysics, Inc., Tucson, AZ

Abstract

The Panama Canal is currently being dredged to deepen and widen the waterway for passage of larger vessels. Dredging efficiency relies heavily on knowledge of the rock type and hardness, as this will determine the dredging method and machinery selected. For example, loose sediment can be removed via suction dredging, whereas more complex drilling and blasting techniques are needed to remove harder rock types. To support this effort, a detailed marine electrical resistivity survey was completed within the Canal to produce a spatial distribution of electrical resistivity values along the canal floor to understand overall rock and sediment distribution. Resistivity data were acquired along a series of parallel, two-dimensional transects, nominally spaced 25 meters apart across the width of the navigable portion of the waterway. Alone, resistivity values could not determine rock type; it was found in this survey that differing geological units can show similar ranges of resistivity values. Although weathered and competent rock generally displays lower resistivity values and loose sediment exhibit higher resistivity, the porosity, water saturation and ionic constituents could potentially affect specific values in localized areas. This survey effort showed however, that when resistivity data are combined with known data from geological maps and Canal boreholes, rock types, strength properties, and geologic boundaries can be identified successfully in the resistivity data.

Introduction

The Panama Canal is the shortest land connection between the Atlantic and Pacific Ocean. The Canal starts near Colón in the Caribbean Sea, and vessels pass through the Gatun locks where they are raised 26 m before entering the Gatun Lake. The transit of the lake accounts for over half of the 51 km of travel before entering the Gaillard Cut. The Gaillard Cut is a 13.7 km long and up to 20 m deep channel that was carved through the rock and shale of the Continental Divide (Moormann and Saul, 2009). At the south end of the Gaillard Cut, the vessels are lowered back to sea level at two sets of locks before leaving the Canal at Panama City.

Presently, work is being conducted to expand the Canal and to build new locks for ‘Post-Panamax’ vessels. Expansion will include dredging the bottom and widening the sides to make way for ships with a maximum draft of 15.2 m in tropical fresh water. Current draft is limited to 12 m for Panamax class vessels. The proposed volume of material to be removed from the Gaillard Cut is approximately 6x106 m3, and the ease by which this material can be excavated will depend on the specific subaqueous rock types and their strength encountered along the waterway.

Geophysical surveys, including electrical resistivity, are ideal for many engineering and environmental studies in that they are able to provide a detailed and continuous map of desired geological, hydrogeological, and geotechnical properties over a large survey area. While land-based electrical resistivity surveys are quite common, water-borne electrical resistivity surveys for the purpose of geological mapping have not been widely performed. More commonly, water-borne resistivity surveys have been used to help understand ocean-aquifer and other hydrological interactions, such as saline intrusion (Abdul Nassir et al., 2000), identifying potential aquifer recharge areas (Snyder et al.,

SAGEEP 2011 Charleston, South Carolina USA http://www.eegs.org 2002), and investigating submarine groundwater discharge (e.g., Belaval et al., 2003; Day-Lewis et al., 2006). A few examples of marine-based resistivity surveys for the purpose of geological mapping do exist, as shown in Rinaldi et al. (2006) and Rucker et al. (2011). Rinaldi et al. (2006), as part of a geotechnical investigation, performed a water-borne survey to support the installation of a wastewater conduit in the estuary of the Rio de la Plata, Argentina. Their work showed that vertical electrical sounding (VES) could be used successfully to map the depth to a sand layer. Rucker et al. (2011) showed that water-borne resistivity data can be a successful tool for mapping specific subaqueous rock types in the Panama Canal.

In this work, we extended the analysis of Rucker et al. (2011) to assess context-specific geotechnical properties of the Panama Canal using a water-borne streaming electrical resistivity survey. The analysis focused on a small section in the Gaillard Cut that comprised either Oligocene-aged tuffaceous agglomerate with intruded andesite, early Miocene-aged marine sedimentary sequences, or competent and less weathered agglomeratic tuff with basalt intrusions of late Miocene age. Strength parameters of rock hardness were mapped over the entire survey area by correlating resistivity to categorical hardness values, extracted from the borehole records. The electrical resistivity method was shown to be a useful tool for delineating the boundaries of differing rock types and strength properties. Consequently, the information could be used for evaluating the ease of dredging and removal of material within the Canal.

Setting

The Panama Canal is a man-made shipping waterway that connects the Caribbean Sea to the Gulf of Panama across the Isthmus of Panama (Figure 1). Using three sets of locks, the canal allows passage of ships through a set of artificial lakes and a long channel cut through the Isthmus. The expansion will increase the width of the entire length of the navigable waterway to allow the passage of larger ships by both increasing the water level in Gatun Lake and creating a deeper and wider channel through Gatun Lake and Gaillard Cut.

The Gaillard Cut portion of the canal is an overland cut channel about 13 km long, connecting the Gatun Lake to the Miraflores Lake. The Gaillard Cut, as originally excavated, was about 92 m wide. Between 1930 and 1970, the cut was widened to more than 150 m. The current expansion project will ensure the minimum width is 280 m along straight portions and 320 m in curves (Rucker et al., 2011). For the analysis presented in this paper, we concentrated on a small region of the Gaillard Cut, called the Cunette-Empire region. The Cunette-Empire region comprises approximately 2.2 km of the central portion of the cut, as shown in Figure 1. Bathymetry data, recorded with an on-board echo sounder, showed the water depth in the study area to be an average of 17 m deep within the navigable channel.

Geology

The Isthmus of Panama is located in the region of four main tectonic plates of South America, Nazca, Cocos, and Caribbean. The isthmus has been referred to as the Panama block (Adamek et al., 1988; Silver et al., 1990) or Panama microplate (Fisher et al., 1994; Coates et al., 2004) and is part of a volcanic arc formed in early Miocene around 17 Ma (Coates et al., 2003; Coates et al., 2004; Molnar, 2008). Within the isthmus, the Panama Canal Basin is structurally complex (Kirby et al., 2008) owing to the presence of a regional shear zone trending northwest–southeast, which is aligned almost nearly parallel to the Panama Canal (Case, 1974). Numerous northeast–southwest trending faults are also present (Pratt et al., 2003). As described by Jones (1950), the faults are typically high angle (N60°) and include normal, reverse, and strikeslip (Lowrie et al., 1982) with a wide range of offsets (cm to km

SAGEEP 2011 Charleston, South Carolina USA http://www.eegs.org scale). Major faults within the Gaillard Cut have been documented in the most recent geological map of the area produced by Stewart et al. (1980). Eocene to late Miocene intrusive, volcanic, and sedimentary rocks underlie portions of the canal (Jones, 1950).

Figure 1: Panama Canal showing the location of the Cunette-Empire study area.

The Gaillard Cut generally transitions from older strata in the north (e.g., the Las Cascadas

Formation) to younger strata in the south (e.g., the Pedro Miguel Formation). The northern part of Cunette-Empire is mainly occupied by the Las Cascadas Formation, consisting primarily of Oligocene tuff and agglomerate, with some andesite and basalt. To the south, Miocene-age agglomerate, basalt and tuff of the Pedro Miguel formation overlay the Cucaracha formation (containing claystone, sandstone, conglomerate, and lignite) which, in turn, overlays the Culebra formation (composed of marine mudstone, sandstone, limestone, and conglomerate) as described by Kirby et al. (2008).

Figure 2 shows a geological map of the study area, which has been modified from Stewart et al. (1980) to accommodate the more recent stratigraphic interpretations by Kirby et al. (2008). Namely, the Culebra and Pedro Miguel Formations were originally recorded by Stewart et al. (1980) as a member of the La Boca Formation. Figure 2 also shows geologic profiles through the Canal, developed from borehole data taken within the present boundary between 1959 and 1972. At times, such as shown in profiles A-A’ and C-C’, the boreholes were drilled on dry land prior to the Canal’s expansion to present-day limits. These data were retained for the figure, but the water column of the canal shows what material has been removed to allow a greater appreciation of the sequences encountered in the Gaillard Cut region. The profiles show mostly agglomeratic sequences in the north, representing the Las Cascadas Formation. About mid-way through the study area, the tuffaceous agglomerate of the Pedro Miguel is incised and bounded by near-vertical faults. The southern portion of the region shows sandstones and siltstones of the Culebra Formation.

SAGEEP 2011 Charleston, South Carolina USA http://www.eegs.org

Figure 2 also reveals some of the hardness characteristics of the rock. The rock hardness was recorded as a qualitative measure during core logging and Table 1 lists the details of the value’s meaning. The scale for rock hardness varied from a RH-1 to RH-5, with RH-5 representing very hard material. These data are shown in Figure 2 alongside the lithology descriptions of the core data. In some instances, the hardness straddled two rock hardness values and was recorded as ‘RH-2 to RH-3’. These are shown as ‘2/3’ in the figure. The rock hardness scale appears to have been developed specifically for the Panama Canal by the Panama Canal Company, mid 20th century, and loosely resembles that of the U.S. Bureau of Reclamation’s scale of H1 through H7.

Figure 2: Geological sequences of the Cunette-Empire region of the Panama Canal, showing a stratigraphic model modified from Stewart et al. (1980) to accommodate the more recent interpretation by Kirby et al. (2008). Also shown are lithologies, recorded from core data along three profiles, as well as their respective rock hardness values (described in Table 1).

SAGEEP 2011 Charleston, South Carolina USA http://www.eegs.org Table 1: Scale of Rock Hardness for use in the Panama Canal

Rock Hardness Value

Qualitative Description

Detailed Description

RH-1 Soft Slightly harder than overburden soil, rock-like structure, crumbles and breaks easily by hand

RH-2 Medium Soft Cannot be crumbled between fingers, but can easily be picked with light blows of the geology hammer

RH-3 Medium hard Can be picked with moderate blows of the geology hammer. Can be cut with knife

RH-4 Hard Cannot be picked with geology hammer, but can be chipped with moderate blows of the hammer

RH-5 Very Hard Chips can be broken off only with heavy blows of the geology hammer

Methodology

Resistivity is a volumetric property that describes the resistance of electrical current flow with in a medium (Telford et al., 1990: Rucker and Fink, 2007). Earth resistivity is a function of soil type, porosity, moisture, and dissolved salts. The concept behind applying the resistivity method is to detect and map changes or distortions in an imposed electrical field due to heterogeneities in the subsurface. For the Panama Canal project, these distortions can be caused by differing geologic media and their degree of weathering. Competent rock with low porosity and low water saturation generally has high resistivity. This makes direct current transmission difficult and the measured voltage gradient high. Weathered rock and loose fill material, on the other hand, should have lower resistivity, depending on the ionic constituents and their concentrations.

For land-based surveys, the electric current may be generated by battery or motor-generator driven equipment and current is introduced into the ground through electrodes (metal rods), with earth-electrode coupling enhanced by pouring water at the electrode locations. In marine acquisition, the electrodes are in direct contact with the water and are towed behind the boat via a cable which is connected to the resistivity meter. In our study, we used the SuperSting R8 (Advanced Geosciences, Inc.), which allowed up to 8 receiving dipoles to be recorded simultaneously with subsequent measurements occurring within seconds. The electrode arrangement is such that the closest set of electrodes behind the boat is the transmitting dipole and the remainders are the receiving dipoles. Ancillary measurements were made such as position using a real-time differential GPS, bathymetry using an echo sounder, and water temperature and conductivity using a hand-held probe. The design criterion for imaging was a minimum of 8 m below the bottom of the canal given a water depth of 15 m. To accommodate the required depth of investigation, the electrode separation was 15 m along the cable; the total cable length was 170 m. In customary notation, the positions of the voltage dipoles are described as “n” distances away from the transmission dipole. Depth of investigation, then, is a function of n.

During this study the boat moved along the canal at 4 km/h, and data were collected approximately every 3.6 s. At this resolution, a sample was collected every 3.75 m, sufficient to distinguish contacts between rock types. The second design criteria of the survey was to image across the width of the canal by acquiring parallel lines spaced 25 m apart. The data were acquired by swathing back and forth between pre-designated turning points and the GPS was used to ensure a sufficient coverage by plotting positional information on a heads-up display. Total resistivity coverage of the canal

SAGEEP 2011 Charleston, South Carolina USA http://www.eegs.org was 663 line-kilometers, with Cunette-Empire comprising approximately 18 line-km of bi-directional coverage. Figure 3 shows the coverage within the Cunette-Empire region.

For a homogeneous earth, the resistance (R) can be converted to resistivity (ρ) using a simple geometric factor, K:

ρ = R•K (1) where K = π•n•(n + 1)•(n + 2)•a (2)

where a is the dipole separation. Since the resistance data for dipole spacing from n=1…6 were not measured in a homogenous earth, the applications of equations 1 and 2 to individual measurements created an “apparent resistivity.” Automated inverse methods applied through numerical models were used to convert the apparent resistivity to an estimate of the true resistivity. The inverse method relied on nonlinear optimization, requiring an iterative procedure to march towards a solution. Its objective is to minimize the difference between the modeled and measured apparent resistivity, usually in a weighted least squares sense. The objective function has been updated many times over the years to also include other terms, such as smooth model constraints (i.e., a smooth model based on minimizing the second spatial derivative of the resistivity). Ancillary data, such as bathymetry and water resistivity, are entered into the inversion processing routine during the forward calculation step, allowing a more accurate depiction of the rock below the water body's electrical properties. The water resistivity within the Gaillard Cut (55 to 75 ohm-m) measured lower than other areas in the canal, which could be due to increased dissolved solids from runoff along exposed rock banks.

Figure 3: Streaming resistivity coverage within the Cunette-Empire region of the Panama Canal. The northern and southern 3D inverse model grids are also shown.

SAGEEP 2011 Charleston, South Carolina USA http://www.eegs.org

The code used to inverse model the dataset was RES3DINVx64 (Geotomo Software, Malaysia). RES3DINVx64 is a 64-bit, multi-threaded inversion code (Loke et al., 2010), which was used on a computing platform with 16 processors for modeling the Panama Canal dataset. The inversion code also allows for an arbitrary model grid for the inversion, separate from that of the forward model. The arbitrary grid allows a much coarser analysis of electrical resistivity, commensurate with the expected scale of lithological changes in the subsurface. Specifically, we used a 30m square grid, with 2m layers; the inverse model grids are shown in Figure 3, where the northern and southern portions of the survey were split based on the bend in the Canal. To model the water column, we used the a priori model block option and assigned the cells the values recorded by a hand-held conductivity meter to depths recorded by the echo-sounder (between 15 and 19m); the water resistivity was 67 ohm-m. The final depth of the modeling was limited to 28m.

Results

The inverse modeling results are shown in Figure 4, as a slice through the 3D model domain at a depth of 21m. The resistivity values are presented in a base 10 log scale and span approximately from 0.4 ohm-m to 800 ohm-m. Overlain on the contoured resistivity data are the specific lithology and rock hardness data from the borehole logs, taken at the 21m depth. The lithology data are presented as colored dots with numbers nearby to represent the hardness. Rock hardness recorded with a half value was meant to represent those rock sections that straddled two adjacent hardness categories. The cross section of lithologies appears to be mainly composed of agglomerate in the Las Cascadas and Pedro Miguel Formations. Some regions show andesite, particularly in the high resistivity areas in the north. Some regions of tuff, intrusive basalt, and ash flow are scattered among the boreholes. The Culebra Formation appears to be mainly sandstone at the 21m depth.

Interestingly, there does appear to be a geology-specific qualitative correlation between the rock hardness and resistivity data. To highlight this correlation, Figure 5 shows a scatter of co-located borehole-resistivity data that compares the resistivity to rock hardness, with Figure 5A focusing on the Las Cascadas Formation and Figure 5B focusing on the Pedro Miguel Formation. The data are divided into material type, which allows a further discriminator for the trends observed in the scatter. The scatter shows that there is a positive correlation between resistivity and rock hardness, with high resistivity and high hardness attributed mostly to andesite and basalt. Low resistivity and low hardness values are associated with tuff and the mid-range for both parameters is filled by agglomerate. When comparing the Las Cascadas Formation to the Pedro Miguel Formation, the resistivity for the latter appears to be on average higher than the former for equivalent lithological sequences. This is likely due to the younger Pedro Miguel being less weathered.

Quantitatively, a simple cluster analysis was conducted that grouped the data into lithologically-based clusters for tuff (moderately soft material), agglomerate (moderately hard material) and andesite (hard material). The centroid of each cluster was then used to fit a regression model that could be used to convert between hardness and resistivity. For Las Cascadas, the cluster centroids were located at a rock hardness, log resistivity of: (2.5, 0.928); (3.5, 1.251); and (4.75, 2.059) for tuff, agglomerate, and andesite respectively. For the Pedro Miguel Formation, the centroids were located at: (2.75, 1.215) and (4.25 and 1.400) for tuff and agglomerate.

SAGEEP 2011 Charleston, South Carolina USA http://www.eegs.org

Figure 4: Electrical resistivity modeling results for the Cunette-Empire region of the Panama Canal. The plan view of resistivity was taken at a depth of 21 m below the water surface, which is about 3-4 m below the bottom of the canal. Overlain on the resistivity are lithological sequences and rock hardness data from boreholes in the canal.

Figure 5: Cluster and regression analysis for resistivity versus rock hardness for co-located borehole and geophysical data. A) Data from the Las Cascadas Formation with three clusters associated with tuff, agglomerate, and andesite/basalt. B) Data from the Pedro Miguel Formation with two clusters associated with either tuff or agglomerate.

SAGEEP 2011 Charleston, South Carolina USA http://www.eegs.org

The last step of the analysis was to convert the resistivity to the different hardness attributes identified by the cluster analysis. The break between rock hardness attributes was at the mid-point between centroids, and all resistivity data were treated with respect to the specific stratigraphic sequence in which they were located. For example, the resistivity within the Las Cascadas Formation was given a base attribute of moderately hard, with the exception for resistivity values less than 12 ohm-m (attribute of moderately soft) and an attribute of hard for values greater than 45 ohm-m. Similarly, the Pedro Miguel formation was given a base attribute of moderately hard except for resistivity values greater than 20 ohm-m being assigned an attribute of hard. All data from the Culebra Formation was assigned an attribute of moderately soft. The results of the conversion are shown in Figure 6. The figure shows that most of the Pedro Miguel is hard rock except near faults and the Las Cascadas appears evenly divided among the three different rock hardness attributes.

Figure 6: Resistivity converted to rock hardness attributes of moderately soft rock, moderately hard rock, and hard rock.

Conclusions

A water-borne streaming resistivity survey was performed in the Panama Canal to help with geological and geotechnical mapping in support of the expansion project. The survey was conducted by towing a 170m cable behind a boat and swathing back-and-forth to cover the width of the navigable channel. For this study, we focused on the small Cunette-Empire region of the Canal, which is approximately 2.2 km in length. The resistivity data were reviewed and interpreted relative to the context-specific, stratigraphic sequence in which they were acquired. For example, it was observed that the late Oligocene tuff and agglomeratic sequences of the Las Cascadas Formation was lower in strength than the late Miocene tuff and agglomerate of the Pedro Miguel Formation. This had a direct influence on the measured resistivity likely due to the degree of weathering of the older material. Additionally, it was observed that resistivity and rock hardness of different lithologies within a single formation clustered into logical groups. These context-specific clusters were then used to formulate a model to

SAGEEP 2011 Charleston, South Carolina USA http://www.eegs.org directly convert the resistivity to a rock hardness attribute (moderately soft, moderately hard, or hard). The results of the conversion showed a logical and coherent hardness map that could be used to help formulate the type of dredging and time required for material removal.

References

Adbul Nassir, S.S., Loke, M.H., Lee, C.Y., and Nawari, M.N.M., 2000. Salt-water intrusion mapping by geoelectrical imaging surveys. Geophysical Prospecting, 48:4, 647-661.

Adamek, S., Frohlich, C. and Pennington, W.D., 1988. Seismicity of the Caribbean-Nazca boundary: constraints on microplate tectonics of the Panama region. Journal of Geophysical Research, 93, 2053-2075.

Belaval, M., Lane, J.W., Fr., Lesmes, D.P., and Kineke, G.C., 2003. Continuous-resistivity profiling for coastal groundwater investigations: three case studies. Symposium on the Application of Geophysics to Engineering and Environmental Problems (SAGEEP), April 6-10, 2003, San Antonio, Texas, Proceedings: Denver, Colorado, Environmental and Engineering Geophysics Society, CD-ROM, 14p.

Case, J.E., 1974. Oceanic crust forms basement of Eastern Panama. Geological Society of America Bulletin, 85, 645-652.

Coates, A.G., Aubry, M.P., Berggren, W.A., Collins, L.S., and Kunk, M., 2003. Early neogene history of the central American arc from Bocas del Toro, western Panama. Bulletin of the Geological Society of America, 271-287.

Coates, A.G., Collins, L.S., Aubury, M.P., Berggren, W.A., 2004. The geology of the Darien, Panama, and the late Miocene–Pliocene collision of the Panama arc with northwestern South America. Bulletin of the Geological Society of America, 116, 1327-1344.

Day-Lewis, F.D., White, E.A., Johnson, C.D., and Lane, J.W., Jr., 2006. Continuous resistivity profiling to delineate submarine groundwater discharge-examples and limitations. The Leading Edge, 25:6, 724-728.

Fisher, D.M., Gardner, T.W., Marshall, J.S., and Montero, P., 1994. Kinematics associated with late Cenozoic deformation in central Costa Rica: western boundary of the Panama microplate. Geology 22, 263-266.

Jones, S.M., 1950. Geology of Gatun Lake and Vicinity, Panama. Geological Society of America Bulletin, 61, 893-922.

Kirby, M.X., Jones, D.S., MacFadden, B.J., 2008. Lower Miocene stratigraphy along the Panama Canal and its bearing on the Central American Peninsula. PLoS ONE, 3, e2791.

Loke, M.H., Wilkinson, P. and Chambers, J., 2010. Parallel computation of optimized arrays for 2-D electrical imaging. Geophysical Journal International, 183, 1202-1315.

Lowrie, A., Stewart, J., Stewart, R.H., Van Andel, T.J., McRaney, L., 1982. Location of the eastern boundary of the Cocos Plate during the Miocene. Marine Geology, 45, 261-279.

Molnar, P., 2008. Closing of the Central American Seaway and the ice age: a critical review. Paleoceanography, 23.

Moorman, Chr., Saul, R., 2009. High capacity bored piles in soft volcanic rock-experiences based on single- and multi-level tests. Deep Foundations on Bored and Auger Piles – Van Impe & Van Impe (eds), Taylor & Francis Group, London, ISBN 978-0-415-47556-3.

Pratt, T.L., Holmes, M., Schweig, E.S., Gomberg, J., Cowan, H.A., 2003. High resolution seismic imaging of faults beneath Limon Bay, northern Panama Canal, Republic of Panama. Tectonophysics, 368, 211-227.

SAGEEP 2011 Charleston, South Carolina USA http://www.eegs.org Rinaldi, V., Guichon, M., Ferrero, V., Serrano, C, and Ponti, N., 2006. Resistivity survey of the

subsurface conditions in the estuary of the Rio de la Plata. Journal of Geotechnical and Geoenvironmental Engineering, 132:1, 72-79.

Rucker, D.F., Fink, J.B., 2007. Inorganic plume delineation using surface high-resolution electrical resistivity at the BC cribs and trenches site, Hanford. Vadose Zone Journal, 6, 946-958.

Rucker, D. F., Noonan, G., Greenwood, W.J., 2011. Electrical resistivity in support of geological mapping along the Panama Canal. Engineering Geology, 117, 121-133.

Silver, E.A., Reed, D.L., Tagudin, J.E., Heil, D.J., 1990. Implications of the North and South Panama thrust belts for the origin of the Panama Orocline. Tectonics, 9, 261-281.

Snyder, D.D., Wightman, W.E., 2002. Application of Continuous Resistivity Profiling to Aquifer Characterization. Symposium on the Application of Geophysics to Engineering and Environmental Problems (SAGEEP).

Stewart, R.H., Stewart, J.L., Woodring, W.P., 1980. Geologic map of the Panama Canal and vicinity, Republic of Panama. United States Geological Survey Miscellaneous Investigations Series Map I-1232, scale 1:100,000, 1 sheet.

Telford, W., Geldart, L., Sheriff, R., 1990. Applied Geophysics. Cambridge Univ Press, Cambridge, UK. 790 pp.