Geophysical methods in studying fresh water

salinization problems

Abstract

1. Integration of TDEM and shallow reflection seismics for detailed

study of the Mediterranean Coastal Aquifer

1.1. Introduction

1.2. Hydrogeological background

1.3. TDEM survey

Seismic Survey

Line GL-0082

1.5. Integrated interpretation of TDEM and seismic results

1.6. Discussion

TDEM study of spatial distribution of brines beneath the sea of

Galilee (Lake Kinneret)

2.1. Hydrogeological background

2.2. Marine TDEM system

2.3. Interpretation

2.4. Results

2.5. Resistivity-Salinity calibration

2.6. Discussion

Acknowledgments

References
 Geophysical methods in studying fresh water
salinization problems

Mark Goldman (1)

ABSTRACT

There are two main sources of fresh water in Israel suffering from severe salinization problem. The first one is the Mediterranean
coastal aquifer which is contaminated by sea water intrusion and the second one is the Sea of Galilee contaminated by sub-bot-
tom brines.

It has been shown during last decade that the time domain electromagnetic (TDEM) geophysical technique is mostly suited for
detecting and monitoring saline ground water of different origins. Since early 90th the method is being routinely used for detect-
ing and monitoring sea water intrusion into the Mediterranean coastal aquifer of Israel. In addition to the routine TDEM surveys,
an integrated TDEM/high resolution seismic reflection method has been recently developed for more detailed study of the
aquifer including subdivision of the latter into sub-units separated by impermeable layers and evaluation of the water quality
(salinity) within each sub-unit. The integrated method was successfully tested in a number of areas throughout the
Mediterranean coastal plain of Israel showing its ability to detect additional resources of fresh ground water. The method was
particularly feasible in the case of the so called hydrological reversals, when fresh water bearing sub-aquifers underlain sea
water intrusion.

In order to study the salinization mechanizm in the Sea of Galilee extensive geophysical surveys including seismic reflection and
TDEM methods have been recently carried out both on-shore and off-shore. The off-shore TDEM survey was carried out using
novel surface marine TDEM system which was particularly design for off-shore surveys on fresh water lakes. The system consists
of a 25 m diameter circular transmitter loop with the receiver antenna located in the center (central loop array). The array floats
on the water surface and is towed by a motor boat. Navigation is carried with the differential GPS.

A total of 33 on-shore and 269 off-shore soundings have been carried out within the lake and it’s vicinity. Using the measured
resistivities which were calibrated with the salinity measurements in boreholes and core probes, the detailed study of a spatial
distribution of brines beneath the sea bottom was carried out.

1. INTEGRATION OF TDEM AND SHALLOW REFLEC- such an area, where groundwater is the only source of
TION SEISMICS FOR DETAILED STUDY OF THE fresh water in the whole coastal plain. The coastal
MEDITERRANEAN COASTAL AQUIFER aquifer supplies about one quarter of the country’s
annual water consumption. Like in many other
Two geophysical surveys using shallow reflection regions, the aquifer suffers from severe salinization
seismics and time domain electromagnetics (TDEM), caused by seawater encroachment. The problem is
were carried out in the Mediterranean coast of Israel. further aggravated by the population growth and,
The surveys were a part of an INCO-DC research pro- consequently, the progressively increasing extraction
ject aimed at developing an integrated geophysical of water from the aquifer. For rational management of
approach for rational management of groundwater the aquifer system, a detailed study of the aquifer and
resources. The general objective of the surveys was a its separate sub-units (which can be saturated with
detailed study of the coastal aquifer in the area and, in fresh groundwater) is necessary. This may be
particular, subdivision of the aquifer into subaquifers achieved by drilling observation wells and by using
separated by impermeable units and evaluation of surface geophysical surveys.
water quality within each subaquifer.
Among various geophysical methods, geoelectric and
seismic reflection techniques seem to be the most
1.1 INTRODUCTION suitable for this purpose. However, each of these
methods alone is efficient in solving only a specific
The problem of effective management of groundwa- hydrogeological problem, but is usually unable to
ter resources is of paramount importance in many provide a general solution. For example, TDEM is very
regions throughout the world. It is particularly impor- efficient in detecting sea water intrusion, but is usual-
tant in the areas suffering from the lack of fresh sur- ly much less successful in delineating geological
face water and insufficient rainfalls. The structures. The seismic method, vice versa, is very
Mediterranean coast of Israel is a typical example of efficient in solving the structural problem but is
unable to distinguish between fresh and saline
ground waters. Therefore, the best way seems to
(1)
The Geophysical Institute of Israel (GII), 1 Hamashbir str., apply both methods and then to perform an integrat-
Holon 58122, Israel. ed interpretation of seismic and electromagnetic data.

61
Although in the past a number of seismic and TDEM about 100 m to approximately 300 m, respectively. In
surveys was carried out in different parts of the order to provide better vertical resolution and accura-
coastal plain of Israel (Goldman et al., 1991; Ben-Gai, cy of interpretation (Goldman et al., 1994), both shal-
1995), they were usually performed independently, low Geonics EM47 and deep EM37 systems were
and no attempt has ever been made to integrate their applied in the western part of the line.
results. However, in many cases such an interpreta-
tion would be necessary, as may be seen from the fol- For interpretation of the TDEM data, the following
lowing. During the previous TDEM surveys, it was approach was applied. First, the interpretational
found at several locations that beneath a very low model consisting of a minimum possible number of
resistivity unit, which was undoubtedly identified with layers was used. After comparing with the seismic
sea water intrusion, there was a highly resistive layer results, additional layers were included in the initial
testifying to the presence of fresh groundwater (the model according to the seismic interpretation. Such
so called hydrological reversals). Unfortunately, in an approach led to a significant improvement of the
most cases the existence of impermeable hydrogeo- interpretational results.
logical units (aquitards or aquicludes), which justifies
the validity of the above mentioned model, could not
be concluded from the TDEM interpretation alone. In 1.2 HYDROGEOLOGICAL BACKGROUND
particular, this problem was encountered in the
Nitzanim area located in the southern part of the The Coastal Plain aquifer of Israel extends along the
coastal plain of Israel (Fig. 1.1). In order to provide a Mediterranean shore line from the Gaza Strip in the
detailed and reliable study of the coastal aquifer in the south to the Mount Carmel in the north, reaching the
area, an integrated geophysical survey, including width of 10 - 15 km eastward. The aquifer consists of
shallow reflection seismics and TDEM, was carried out the Quaternary sequence of marine and continental
in the vicinity of a number of hydrogeological obser- deposits composed predominantly of calcareous
vation wells. sandstone (the Kurkar unit) resting on an erosional
contact. The sequence starts with the Calabrian
The seismic survey included two high-resolution regional regression which terminates the Saqiye
reflection lines shot using the CMP technique. The marine regime. The accumulated succession of sedi-
TDEM survey included nine central loop soundings ments which follows bears evidence of several cycles
carried out along seismic line GI-0082 (Fig. 1.1). The of transgressions and regressions of the sea. With
transmitter loop size varied between 50 by 50 m near each cycle, deposition environment shifted from west
the sea shore to 200 by 200 m in the eastern part of to east and back. The lithologic variability observed in
the profile providing the penetration depth from the Quaternary sequence reflects different types of

Fig. 1.1 Location map showing TDEM sites, seismic lines and observation wells in the Nitzanim area.

62
Fig. 1.2 Lithological logs of the Nitzanim observation wells. The figures to the right of each log indicate the depth (m) from the
surface.

sediments characteristic of a certain environment: cal- 1 meter per 1 km. Discharge of the aquifer is in the
careous sandstone, sandy limestone, sand, clay, silt, form of seepage along the seawater/freshwater inter-
conglomerate and loam. In the west, the aquifer con- face. In certain areas, over-exploitation of the aquifer
sists of marine deposits, while in the east, the has caused inland seawater encroachment to more
deposits are of continental origin. Calcareous sand- than 1 km from the sea shore.
stones are generally porous and hydrologically con-
ductive and considered aquifers, whereas clays are A number of observation wells are located at the
impermeable and act as aquicludes. In the places investigated site in the Nitzanim area (Figs.1.1 and
where the clays are thick and extensive enough, they 1.2). Three of the wells (12/0, 12/A and 12/1) penetrate
divide the aquifer into distinct subunits. The aquifer the entire aquifer down to the Saqiye group which is
rests on impermeable black shales and clays of the about 160 m deep in this area. In the western part of
Saqiye group of Pliocene - Miocene age. the area (wells12/0 and 12/A) marine sediments are
penetrated, whereas in the eastern part (well 12/1)
The coastal aquifer is replenished by rain precipita- continental sediments were encountered. In Fig. 1.2,
tion during the winter months. Groundwater flows various Kurkar subaquifers encountered in the wells
westward, toward the sea. Usually, the groundwater are designated by capital K with corresponding
level rises from west to east with a gradient of about indices.According to hydrological data, the water

63
Fig. 1.3 Schematic W-E hydrogeological cross-section of the Nitzanim test site. The horizontal axis shows the distance from the
sea.

table level in the area is about 2 - 3 m above the mean The interpretation results for four soundings carried
sea level (MSL). The sea water penetrates into the out in the western part of the profile are shown in Fig.
aquifer up to about 1.0 km from the coastal line. 1.4. No a priori information has been used at this
stage. The initial model for each inversion was
A schematic W - E hydrogeological section across the obtained by applying first Occam inversion
investigated area is represented in Fig. 1.3. (Constable et al., 1987) and then reducing to the pos-
sible minimum the number of layers in the recovered
smooth model. According to the above described
1.3 TDEM SURVEY hydrogeological setting of the area, the interpreted
resistivities can be divided into four groups:
Nine TDEM stations were established roughly along
seismic line GI-0082 (Fig. 1.1). All the collected data • Very low resistivities ( less than 3 ohm-m). These
were processed and interpreted using the Interpex resistivities are typical for saline water saturated
TEMIX-XL 1-D interpretation package (Interpex, 1996). lithologies (both aquifers and aquicludes).

64
 Fig. 1.4 Preliminary 1-D interpretation of the TDEM data collected at four stations in the western part of the profile.

• Low resistivities (between 3 to 8 ohm-m). These Fig. 1.5 shows the pseudo-2D resistivity cross-section
resistivities are characterizing either brackish water of the west part of the profile compiled from appro-
saturated aquifers or aquicludes (mainly clays). priate 1-D resistivity vs. depth models (Fig. 1.4).
Beneath station N3 one can see two moderately resis-
• Moderate resistivities (between 8 and 15 ohm-m). tive layers which can be identified with fresh water
Typical for fresh water saturated aquifers and some saturated subaquifers. However, the lateral extension
aquicludes (loams). of the lower subaquifer is unclear since it was not
detected at stations N2 and N4. It seams it is termi-
• High resistivities (greater than 15 ohm-m). These nated somewhere between stations N3 and N4 by an
resistivities are typical for either dry or fresh water aquiclude layer having resistivity of approximately 5
saturated aquifers. ohm-m, while between N3 and N2 the subaquifer
probably becomes saturated with saline water. Thus,
The lateral extent of sea water intrusion into the according to the preliminary TDEM interpretation, the
aquifer can be estimated from the TDEM measure- lower subaquifer saturated with fresh groundwater is
ments alone just taking into account the depth to the confined to an area around station N3.
aquifer base (top Saqiye), which is very consistent
along the whole profile (Figs. 2 and 3). According to
the borehole data the depth to the aquifer base is SEISMIC SURVEY
approximately 160 m in the whole test area. This
means that if the very low resistivity unit is located at The high-resolution seismic reflection survey carried
depths shallower than 160 m, it represents sea water out in the Nitzanim area included two seismic lines
intrusion, otherwise it is identified with the Saqiye shot using conventional P-wave technique (lines GI-
clays underlying the aquifer. Fig. 1.4 clearly indicates 0082 and GI-0083, Fig. 1.1). The length and locations of
that sea water intrusion terminates somewhere the lines were chosen in accordance with the survey’s
between stations N3 and N4 (i.e. at a distance of target and local surface conditions. The acquired seis-
approximately 800 m from the sea). Thus, according mic data were processed at the Geophysical Institute
to the preliminary TDEM interpretation the aquifer is of Israel (GII) processing center using the industry-
fully saturated with fresh water eastward of station standard PROMAX package. The typical seismic time
N4. Unfortunately, due to an insufficient resistivity section is shown in Fig. 1.6. The horizontal axis on the
contrast, the depth to the water table can not be section shows station numbers while the vertical axis
obtained from the TDEM measurements. Therefore the is two-way traveltime in milliseconds. The section is
most interesting part of the profile for evaluating the related to a horizontal datum placed at the MSL. The
quality (salinity) of groundwater is located between part of the section located above the datum appears
stations N1 and N4. at negative times. Since no velocity information from

65
 Fig. 1.5 Pseudo-2D resistivity cross-section constructed from the preliminary interpreted 1-D models (fig 1.4).

boreholes was available in the investigated area, no mate the elevation of the reflected events appearing
attempt at depth conversion of the sections was below the datum. Based on this velocity, positive
made. However, rough estimates of the elevations of reflection times in milliseconds roughly correspond to
the reflected events appearing on the sections were negative elevations in meters.
made on the basis of the velocities obtained from
seismic refraction surveys carried out in different The detailed description of seismic results along both
parts of the Coastal Plain. According to the refraction lines is given by Shtivelman and Goldman (1999).
data, the velocity of P waves in the Kurkar unit (cal- Here, we shall restrict ourselvs by the description of
careous sandstone) is about 2000 m/s. Since below only the western part of line GI-0082, which provides
the MSL the section is represented mainly by the the most significant results for the integrated inter-
Kurkar unit (Fig. 1.2), this velocity can be used to esti- pretation.

Fig. 1.6 Seismic section of the western part of line GI-0082. Every trace is displayed; trace spacing is 1.25 m.

66
Line GI-0082 The reflector appearing at time of about 50 ms, may
be related to a thin clay unit penetrated at elevation of
The line runs in a NW-SE direction and is about 2.6 km about -48 m; this reflector apparently separates two
long (Fig. 1.1). The borehole data shown in Fig. 1.2 upper Kurkar layers K01 and K02. The layers may be
were used for correlation of the seismic data. traced (albeit somewhat problematically) up to bore-
hole 12/B (station 155) where they seem to corre-
The western part of the line: In this region, the seismic spond to layers KB1 and KB2. Other events appearing
section (Fig. 1.6) looks rather complicated. A sequence above and below the reflector, seem to be uncorrelat-
of reflected events with various local dips can be ed to the borehole data; probably, these events corre-
detected below the datum down to times of about 250 spond to local clay lenses which do not reach the
ms. The continuity of the events is clearly interrupted borehole.
at a number of locations which were interpreted as
shallow faults and marked on the section by thin
lines. The faults have a form of flower structures and 1.5 INTEGRATED INTERPRETATION OF TDEM AND
are apparently related to strike-slip motions with SEISMIC RESULTS
minor vertical displacements.
The individual TDEM interpretation described in
Consider a sequence of reflections in the vicinity of Section 1.3 led to a hydrogeological model in which
borehole 12/A (Fig. 1.2) located near station 365. the lower subaquifer was saturated with fresh water
Correlation with the borehole data shows that the within a narrow strip located below station N3 (Fig.
reflector appearing at times of about 130 ms, may be 1.5). However, according to the seismic interpretation,
related to the top Saqiye interface penetrated at ele- the lower Kurkar units designated as KA6 and KA7
vation of -143 m. This reflector can be traced up to sta- extend from a vicinity of TDEM station N2 continu-
tion 515 to the east (Fig. 1.6) and up to station 200 to ously through stations N3 and N4 further eastward
the west. Its extension further to the west is unclear, (Fig. 1.6). The moderately resistive layer detected
apparently due to decreasing thickness of the lower- beneath station N3 at approximately the same eleva-
most Kurkar layers, as can be seen in borehole 12/0. tion (-120 m) is most likely identified with these units
saturated with fresh water. If this assumption is cor-
The reflector appearing at time of about 115 ms, can rect, the data collected at station N4 where the lower
be correlated to the top of the Kurkar layer KA6 pene- subaquifer was not detected, should be reinterpreted.
trated in the borehole at elevation of -122 m. In order to make the TDEM interpretation consistent
with the seismic results, an additional moderately
The reflector appearing at time of about 90 ms might resistive layer has been included in the initial model
be correlated to the top of the relatively thick clay used for the TDEM inversion. This led to the solution
layer penetrated at elevation of -86 m. It is difficult to shown in Fig. 1.7. For the sake of convenience, the fig-
detect and trace reflections from two thin Kurkar lay- ure also shows the results of the preliminary interpre-
ers KA4 and KA5 below the clay layer, although some tation. The uncertainty of the model parameters can
indications on their presence as a single unit can be be roughly estimated by using the so called linear
found in the section. equivalence analysis (Goldman et al., 1994). The best
fit models are shown by solid lines in the resistivity
Two upper reflectors appearing at times of about 40 vs. depth sections in Fig. 1.7. Dashed lines in the figure
ms and 60 ms, may be correlated to two thin clay lay- represent alternative models for which the misfit error
ers penetrated at elevations of -40 m and - 53 m. These is only slightly greater than for the best fit model. As
reflectors separate three upper Kurkar layers KA1, KA2 could be expected, both the resistivity and thickness
and KA3. In the seismic section, the event associated of the lower moderately resistive layer are poorly
with KA3 layer clearly pinches out in the vicinity of sta- resolved and its inclusion in the interpreted model is
tion 280, while KA2 layer can be traced further to the only justified by the existence of the independent
west. seismic interpretation. It should be emphasized that
the reinterpretation was based on seismic results
The sequence of events at the beginning of the sec- only, without being biased by any borehole informa-
tion (in the vicinity of station 20) can be correlated to tion. The misfit error of the final interpretation
borehole 12/0 (Fig. 1.2) located about 135 m south- decreased from approximately preliminary interpreta-
ward of the beginning of the line. Here the identifica- tion to slightly more than 2 %. But the most important
tion of the Top Saqiye interface is problematic, proba- result is that the final interpretation of the TDEM data
bly due to a relatively small thickness of the lower now becomes consistent with the independent seis-
Kurkar layers K05 and K06. The strong reflector appear- mic interpretation and ultimately with the borehole
ing at time of about 100 ms, may possibly be related data available. The final resistivity cross-section
to the top of the thick clay layer penetrated at eleva- accompanied by the borehole data is shown in Fig.
tion of -110 m. The reflector appearing at time of about 1.8. One can see that the location of the lower moder-
80 ms, is apparently related to the clay-loam layer ately resistive unit roughly coincides with the lower
penetrated at elevation of -79 m. The time interval subaquifer in well 12/A. Note that the coincidence is
between the above two reflectors may be associated rather poor due to the above mentioned uncertainty in
with the Kurkar layers K03 and K04. Tracing these layers the layer parameters. The resistivity of the aquiclude
to the east indicates that they may be apparently relat- layer separating two subaquifers varies laterally from
ed to layer K A2 mapped in the vicinity of borehole 12/A. approximately 5 ohm-m in the eastern part of the sec-

67
 Fig. 1.7 Preliminary (above) and final (below) 1-D interpretations of the TDEM data at station N4.
Dashed lines represent equivalent models.

tion to slightly more than 2 ohm-m in its western part. aquiclude. Finally, the boundary between the moder-
This lateral variation can be explained by different ately and highly resistive layers and very low resistiv-
salinities of water within the aquiclude: the closer to ity unit exactly coincides with the freshwater/seawater
the sea, the higher salinity (i. e. the lower resistivity) interface detected in wells 12/0 and 12/B. The interface
is expected. It should be noted that this variation may was not detected in well 12/A obviously because of
explain the above mentioned difference in the inter- the impermeable clays appearing somewhere in the
pretation of the TDEM data at points N3 and N4. The vicinity of station N3.
much smaller resistivity contrast between the
aquiclude and underlying subaquifer at point N4 as Comparison of the seismic and TDEM sections with
compared to N3 may be the reason why this sub- the geological cross-section compiled from the bore-
aquifer was not detected at N4 during the preliminary hole data shows some discrepancy in the layer geom-
interpretation. etry in the area between boreholes 12/A and 12/B.
Specifically, while the geological cross-section
The boundary drawn within the very low resistivity obtained by linear interpolation between the bore-
unit between approximately 1.5 ohm-m and 2.5 ohm- holes shows monotonous inclination of all layers
m can be most likely identified with the boundary westward (Fig. 1.3), both geophysical sections show a
between the sea water saturated aquifer and structural high in the corresponding region (Figs. 1.6

68
 Fig. 1.8 Pseudo-2D resistivity cross-section based on the final interpretation of the data at station N4. Lithological logs are
shown for comparison.

and 1.8). Such a situation can be encountered over a subaquifer was not always revealed by the routine
region with sparse borehole control. In this case, seis- conventional TDEM inversion.
mic sections can provide a continuous and, therefore,
more diagnostic image of the subsurface. The reinterpretation of the TDEM data based on the
combined use of seismic and electromagnetic results,
enabled us to successfully solve the above inversion
1.6 DISCUSSION problem thus considerably improving the hydrogeo-
logical significance of the geophysical results.
The geophysical surveys carried out in the Nitzanim
area provided an important information necessary for The integrated approach to the interpretation of the
a detailed study of the aquifer in the area. geophysical data can be applied for solution of simi-
lar problems in other areas.
The seismic sections obtained along the reflection
lines display a sequence of reflected events down to
times of about 250 ms (about 270 - 300 m depth). TDEM STUDY OF SPATIAL DISTRIBUTION OF
Correlation of the seismic data with borehole infor- BRINES BENEATH THE SEA OF GALILEE
mation shows that the reflections can be related to (LAKE KINNERET)
various impermeable units located within and below
the aquifer. By tracing the reflectors along the sec- An extensive TDEM survey covering the lake with a
tions, the geometry and lateral extension of the units fairly dense grid of points has been carried out during
can be estimated. Based on this interpretation, the 1998. A total of 269 offshore and 33 supplementary
aquifer can be subdivided into a number of sub- onshore TDEM soundings were performed along six
aquifers separated by the impermeable units, as it has N-S and thirteen W-E profiles and at selected points
been done using the borehole data alone (Fig. 1.3). At both offshore and onshore along the whole coastal
several locations along the sections, disturbance line (Fig. 2.1). The interpreted resistivities were cali-
zones apparently related to shallow faults were brated with the direct salinity measurements in a few
mapped; this mapping may have implications for the boreholes available in the vicinity of the lake and deep
hydraulic continuity of the various aquifers in the core probes taken from the lake bottom.
region..

The TDEM survey resulted in a pseudo-2D resistivity 2.1 HYDROGEOLOGICAL BACKGROUND

cross-section located along the western part of seis-
mic line GI-0082. The section clearly shows sea water The Sea of Galilee (Fig. 2.1) is the lowest fresh water
intrusion appearing as a highly conductive geoelectric lake on earth, and is located within the Dead-Sea Rift
unit. A number of fresh water saturated subaquifers Valley. The average water level is 210 m below mean
can be detected in the section. However, the lower sea level (MSL), the total surface area of the lake is

69
 Fig. 2.1 A location map showing the locations of the offshore and onshore TDEM measurements.

approximately 170 km 2 with the maximum water aquitards of predominantly chalk and marl (Arad and
depth being 46 m. The average salinity of the lake is Bein, 1986). Although faulted, these aquifers are later-
220 mgCl/l (Chlorine is the major ion in the saline ally continuous (Fig. 2.2). The salinity in deep aquifers
water, hence, salinity is usually expressed in Cl con- increases from less than 100 mgCl/l in the highlands
centration), an order of magnitude higher than the to more than 10,000 mgCl/l in the downfaulted blocks
concentration in surface streams which enter the lake. (Mazor and Mero, 1969).

Mesozoic to tertiary sediments and Neogene to qua- The subsiding basin in which the Sea of Galilee is
ternary basalt crop out in the highlands on both sides seated (Fig. 2.2) is capped by a Miocene-Holocene
of the rift valley and constitute the recharge area of sequence thicker than 6 km in the deepest part (Ben-
the regional aquifers. Most of the groundwater is Avraham et al., 1996), consisting of alluvial and lacus-
drained to the rift valley through the sandstone, lime- trine deposits, evaporates and intrusive and extrusive
stone or dolomite aquifers of the lower Cretaceous to igneous rocks (Marcus and Slager, 1985). These rocks
Eocene, which are separated from each other by are mainly aquitards which confine flow through the

70
Fig. 2.2 Schematic hydrogeological cross-section showing the groundwater sources of the saline springs at the western shore
of the Sea of Galilee. The basin fill is an aquitard.

regional aquifers (Fig. 2.2), forming two independent (Stiller, 1994). The water in all cores showed a roughly
systems on the eastern and western sides of the basin linear increase in chlorine concentration from about
(Arad and Bein, 1986). On the western margin, some 220 mg/l at the top of the core (water-sediment inter-
faulted blocks expose the upper Cretaceous aquifer at face) to 350-600 mg/l at 50 cm, and 2000-3500 mg/l at
three locations: Tabha, Fuliya and Hamei Teverya (Fig. 500 cm in the long cores (Fig. 2.3). A rough estimate
2.1). A few saline springs are located adjacent to these based on the observed gradients in the cores has
exposures as well as offshore at Barbutim. These revealed a diffusion rate of 5.3 mgCl cm-2 yr-1 from the
springs discharge water with salinity ranging from a lake bottom, resulting in an annual contribution of
few hundred to about 20,000 mgCl/l, and are a mix- 9,000 tons of Cl from the base sediments of the lake
ture of fresh and saline groundwater (Mazor and by diffusion and advection (Stiller, 1994). These results
Mero, 1969; Issar, 1983). The fresh component drains clearly indicate that saline units are located at shallow
latterly from the regional shallow aquifers, and the subbottom depths over large parts of the lake.
saline component from deeper aquifers (Fig. 2.2). On However their spatial distribution and actual subbot-
the other hand, on the eastern margin, low perme- tom depth are unknown.
ability sediments block discharge from the regional
aquifers to outlets near the lake shore. A few bore- From a practical point of view, as a source for irriga-
holes (<200 m) along the eastern shore have revealed tion water the lake salinity is too great for some crops.
that water salinity ranges from a few hundred mgCl/l Moreover, when lake water is pumped and recharged
to about 15,000 mgCl/l (Shaharabani et al., 1980). into the Coastal Plain aquifer for seasonal storage, it
further increases the aquifer salinity. Therefore, great
The salinity of the Sea of Galilee poses a major prob- efforts have been and are presently devoted to
lem to Israel’s water system since the lake supplies decreasing the lake salinity. In order to further reduce
about 500 million m3 of water annually, some 30% of the lake salinity, it is vitally important to evaluate
the country’s water consumption. The average quantitatively the spatial distribution of salts in the
amount of Cl entering the lake is 150,000 tons per sediments beneath the lake floor. This objective is ful-
year (Simon and Mero, 1992), out of which about filled, at least in part, by using the proposed surface
56,000 tons are diverted from some onshore springs marine modification of the well known time domain
through an aqueduct before entering the lake electromagnetic (TDEM) method.
(Assuline et al., 1994). Chlorine contribution from sur-
face streams is about 13,000 tons per year and there-
fore ungauged onshore and offshore springs and dif- 2.2 MARINE TDEM SYSTEM
fusion from the lake sediments contribute about
80,000 tons of Cl annually into the lake. The discharge The conventional onshore TDEM system is well
location of the majority of these ungauged salts is known and widely elucidated in the geophysical liter-
unknown as only a few offshore springs have been ature (e.g. Fitterman and Stewart, 1986; Goldman et
found. al., 1991). The system consists of a transmitter and
receiver. The transmitter is usually represented by a
Nineteen short cores (50 cm) and three long cores square loop of insulated wire laid on the surface. A
(500 cm) were dredged at various locations on the multi-turn air coil (about 1 m in diameter) is placed
lake floor in order to examine their chlorine content either inside or outside the loop and serves as the

71
 Fig. 2.3 Chloride concentration in the pore water of the 5 m long cores versus depth (after Stiller, 1994).

Fig. 2.4 Schematic configuration of the marine TDEM system.

receiver antenna. The array, in which the receiver coil The emf is measured by a receiver coil during the
is located in the centre (central loop array), provides transmitter “time-off” period only, thus providing the
the most accurate data and best lateral resolution and measurements of a purely secondary response, this
is, therefore, the most widely used configuration in being caused by the currents induced in the ground.
TDEM soundings. The size of the transmitter loop usu- As a result, the transmitter/receiver separation can be
ally varies between 50 by 50 m for exploration depths lesser than the required exploration depth and this
of up to approximately 100 m, and 500 by 500 m for unique feature of the TDEM method leads to signifi-
exploration depths of up to about 1000 m. It should be cant improvement in lateral resolution and accuracy
noted that the specified figures are very approximate of measurements as compared with all other electric
and strongly depend on the geoelectric parameters and EM techniques presently available.
and the noise level in the area.
As far as marine TDEM surveys are concerned, to the
The current waveform driven through the transmitter best of our knowledge all of them have been carried
loop consists of equal periods of time-on and time-off. out using sea floor located transmitter and/or receiver

72
arrays (Cairns, et al., 1994). Taking advantage of the basement are almost uniquely determined (see
fact that the water in the Sea of Galilee has low salin- Goldman et al., 1994 for some more details).
ity (and therefore low conductivity), a new highly Moreover, comparison of Figs. 6b and 6c shows that
effective surface marine TDEM system is proposed the standard interpretation, which results in a very
(Figs. 3 and 4). The array consists of a 25 m diameter simple three layer solution, recovers almost the same
circular transmitter loop which floats on the water sur- parameters of the geoelectric basement as the far
face and is towed by a motor boat. The loop is con- more complicated “gradient” layer interpretation.
structed from a hermetically sealed PVC pipe, to Thus, it seems that, in the absence of a sufficient
which the transmitter current cable is tied using stan- amount of a priori information about the “gradient”
dard plastic clips. A 1 m diameter receiver coil is rigid- layer, the standard interpretation is preferable, espe-
ly connected to the center of the transmitter loop and cially if parameters of the geoelectric basement
to another rubber non-motorised boat which is carry- (saline water) are particularly important. If, however,
ing an operator and a receiver console. Beside the coil the salinity-depth profile is known from independent
and receiver console, there are no other metallic measurements, then the ambiguity in the interpreta-
objects inside the transmitter loop. The navigation of tion of the “gradient” layer parameters can be
the system is carried out by a differential GPS instru- reduced by calibrating geoelectric measurements
ment. The latter, as well as the motor generator and with the appropriate salinity measurements.
transmitter, can be located on the tow-boat outside
the loop (Fig. 2.4). It is interesting to note that, unlike the upper part of
the section, the geoelectric basement resistivity does
Since the measurements with the suggested marine not allow any significant gradual changes with depth.
TDEM do not require assembling and disassembling The standard interpretation is thus unique as far as all
of the system at each measuring point, the productiv- geoelectric parameters of the basement are con-
ity of the survey is much greater than that of the con- cerned.
ventional onshore TDEM. This, in turn, makes it possi-
ble to obtain more detailed information and to
increase the reliability of the interpretation, by carry- 2.4 RESULTS
ing out the marine measurements along profiles
rather than at isolated points as is typical of conven- Most of the offshore soundings were acquired along
tional surveys (though each sounding is being carrid roughly straight line profiles crossing the lake in both
out in a stationary mode of operation). N-S and W-E directions. Some of the profiles have
also onshore continuations to distances of up to 1 km
from the shore line (Fig. 2.1). Additional thirty eight
2.3 INTERPRETATION offshore measurements were carried along the
coastal line due to a greater variability of the resistiv-
The TDEM data collected at both onshore and offshore ities near the coast. Results are presented both as
locations were processed and interpreted using the maps representing the resistivity at various depths
standard 1-D algorithm for stratified earth (Interpex below the lake’s bottom (Fig. 2.6) and as pseudo-2D
Ltd, 1989). However, the previously mentioned direct resistivity cross-sections at selected profiles (Figs. 2.7
salinity measurements carried out by Stiller (1994) and 2.8). Only the results obtained with the gradient
throughout the lake, show a continuous increase of inversion method are presented here since, in most
salinity with depth ( Fig. 2.3) and such a salinity distri- cases, the maps and cross-sections for both layered
bution suggests a geoelectric model different from and gradient inversion methods are very similar. The
that used in the standard interpretation software (hor- depth maps were prepared with the krigging method
izontally-layered earth) should be adopted. It is there- whereas the cross-sections were plotted using linear
fore recommended that a model consisting of a grad- triangulation with a vertical/horizontal anisotropy
ual distribution of resistivity with depth (gradient lay- ratio of 0.1. The latter procedure was selected to mini-
ers) should thus be examined as an alternative inter- mize horizontal effects caused by the much larger hor-
pretation model. In order to do so, the data collected izontal distance between stations compared to the
at one of the offshore stations were taken as an exam- vertical distance between different resistivity values
ple data set (Fig. 2.5). The standard interpretation within the same station. In both processing proce-
leads to a geoelectric model which consists of three dures, a resistivity values cutoff of 5 ohm⋅m was intro-
layers: the upper layer approximately 25 m thick hav- duced. The different interpolation methods have
ing a resistivity of about 10 ohm-m, an intermediate incorporated some slight discrepancies between the
layer, about 17 m thick and more conductive (having plotted resistivity at a certain depth between the map
a resistivity of 3.8 ohm-m), and finally, an extremely and the cross-section.
conductive geoelectric basement having a resistivity
of about 0.8 ohm-m (Fig. 2.5b). It should be noted that The resistivity values may be divided into two major
similar models were recovered at the overwhelming units, each having a distinct spatial distribution. The
majority of the off-shore TDEM measurements. Fig. first, High Resistivity Unit (HRU) has values greater
2.5c shows the alternative interpretation using a than 4 ohm⋅m. The unit includes the lake water (Figs.
quasi-gradual decrease in resistivity with depth. 2.6, 2.7 and 2.8) with values of 7 to 12 ohm⋅m, usually
Analysis shows that, despite fairly high ambiguity in closer to 10 ohm⋅m, in agreement with an indepen-
the detection of the parameters of the “gradient” dent measurement of the lake water using DC current
layer, the depth and the resistivity of the geoelectric (A. Katz, personal communication). The unit covers

73
 Fig. 2.5 Layered against gradient interpretation:

(a) apparent resistivity versus time

(b) layered interpretation
(c) gradient interpretation.

almost the entire area of the map at 5 m depth below two small areas, adjacent to the Tiberias springs and
the lake bottom (Fig. 2.6a) and is dominant in the 10 Haon site (Fig. 2.6a). In the 10 m map (Fig. 2.6b), it is
m map (Fig. 2.6b). In the 15 m and 20 m maps, the unit exposed at much larger areas and in the 15 m and 20
appears along most of the lakes perimeter to a dis- m maps the unit covers most of the internal part of the
tance of approximately 2 km from the shore, except lake (Figs. 2.6c and 2.6d). The unit appears also
for areas in the southern and southeastern parts and onshore as isolated pockets in the Beteha valley and
a small area near the Tiberias springs (Fig. 2.6c and in the southern shore of the lake.
2.6d).
The Medium Resistivity Unit (MRU) is characterized
The Low Resistivity Unit (LRU) represents resistivity by resistivities varying between 2 and 4 ohm-m. In
values below 2 ohm⋅m. In the 5 m map, it appears in most cases it can be identified with the above men-

74
 Fig. 2.6 True resistivity maps at different depths beneath the lake bottom.

tioned “gradient” layer and it appears throughout the 2.8).

lake at depths of 5 m and 10 m below the bottom and 2.5 RESISTIVITY-SALINITY CALIBRATION
in the margins of the LRU at greater depths (Figs. 2.6
a,b,c and d). The above results represent the sub-surface spatial
resistivity distribution. Previous calibrations of resis-
A prominent feature of the recovered resistivity cross- tivity values to salinity concentrations carried out at
sections is that the geometry of the geoelectric layers the Mediterranean Coastal Plain of Israel (Goldman et
is roughly parallel to the lake bottom (Figs. 2.7 and al., 1991) and the Dead Sea shore (Kafri et al., 1997),

75
 Fig. 2.7 Pseudo-2D resistivity cross-sections in the W-E direction.

have revealed that resistivity values around 1 ohm⋅m tivity data recovered around the well.
correspond to concentrations of about 12,000 to 2) Linear extrapolation of the chlorine concentration
60,000 mgCl/l, respectively. Actually, the resistivity gradient from the bottom part of the three 5 m
values represent the concentration of the total dis- cores (Stiller,1994) to the top of the conductive
solved solids (TDS). However, since , the proportion layer at the location of the core.
of chlorine in the TDS (in equivalents) varies in a nar-
row range between 45 to 50 %, both calibrations dif- In these calibrations, a few assumptions have been
fer by roughly constant factor of two. used:

Two types of calibration were performed, in both 1) The proportion of chlorine in the TDS is constant
cases the gradient inversion method was used: throughout the lake’s sub-bottom.
2) The concentration gradient in the bottom part of
1) Comparison of the chlorine concentration mea- the cores continues linearly till the conductive
sured in the Haon-2 well (Fig. 2.1), with the resis- layer.

76
 Fig. 2.8 Pseudo-2D resistivity cross-sections in the S-N direction.

3) Only the brine has affected the measured resistiv- distribution entrapped within the lake’s sediments are
ity signal; parameters such as host rock lithology highly heterogeneous. Most changes, both in brine
and porosity are negligible. concentration and depth are within the 1-2 km near
the shore. A significant change also occurs between
The results of the calibration are presented in Table 1. the southern and northern parts of the lake. The maps
and cross sections define two zones with different
The above data implies that the resistivity is related to transport mechanisms. Under most parts of the lake,
the chlorine concentration within the range of 8,000 to the diffusion is the major transport process, whereas,
27,000 by the expression: Resistivity (ohm⋅m) = 4507⋅ in the margins, the advective flux dominates and dif-
C (mgCl/l) -0.9083. Based on the above relationship, a fusion is negligible.
value of 1 ohm⋅m corresponds to approximately
10,500 mgCl/l, 1.5 ohm⋅m corresponds to approxi- In the central part of the lake, the very similar shape
mately 6700 mgCl/l and 0.5 ohm⋅m corresponds to a of the top LRU and the lake bottom suggests that in
concentration of 22,600 mgCl/l. the past, a saline lake covered the area and its water
was trapped in the sediments. Since the lake was cov-
ered with freshwater, diffusion of salts from the sedi-
2.6 DISCUSSION ments into the lake takes place.

The results presented above indicate that the brine In the SE part of the lake, the distance between the top

77
 Table 1
Ω c (z)*** z (m) TDEM core /
(ohm⋅m) (mgCl) site (mgCl l-1 m-1)** well*

1.07 9542 24.5 KIN-D 390 D

0.85 11922 30.6 KIN-D 390 D
0.75 14615 37.5 KIN-D 390 D
0.70 17662 45.3 KIN-D 390 D
1.20 10017 16.2 KIN-F 620 F
0.91 13287 21.4 KIN-F 620 F
1.09 8549 15.5 KIN-U 550 U
0.85 12273 22.3 KIN-U 550 U
0.66 16486 30.0 KIN-U 550 U
0.46 21252 38.6 KIN-U 550 U
0.45 26644 48.4 KIN-U 550 U
0.63 16670 10 KH-9 ——— Haon-2

* Location of cores and Haon-2 well are shown in Fig. 2.1

** Chlorine concentration gradient on the bottom part of the core.
*** Calculated chlorine concentration at top of the conductive layer based on the concentration gradient (except
for that of Haon-2 well).

of the LRU and the lake bottom is small, sometimes M., 1996. Gravity field over the Sea of Galilee: Evidence for a
negligible. This implies that this part of the lake was composite basin along a transform fault. Journal of
covered by the freshwater lake much later than the Geophysical Research, 101, 533-544.
northern part.
BEN-GAI, Y. 1995. Sequence stratigraphy of the Plio-
Pleistocene in the continental margin of the Southern
The Beteha, seems as an isolated pocket, this is sur- Levant. Ph.D. Thesis, Tel Aviv Univ.
prising since it is a delta of fresh water rivers entering
the lake. Delta sediments are usually coarse grained CAIRNS, G., EVANS, R. and EDWARDS, R.N., 1994. A tran-
and have a high permeability. This suggests that sient electromagnetic investigation of the TAG hydrothermal
either the Beteha was also covered by the same lake mound. Paper p.o2-19, 12th workshop on electromagnetic
as the one below the Sea of Galilee or it was an iso- induction in the Earth, Brest, France, 52.
lated lake separated by a narrow strip.
CONSTABLE, S. C., PARKER, R. L. and CONSTABLE, C. G.,
1987. Occam’s inversion: A practical algorithm for generating
smooth models from electromagnetic sounding data.
ACKNOWLEDGMENTS Geophysics, 52, 289-300.

The integrated TDEM/seismic reflection survey in the FITTERMAN, D.V. and STEWART, M.T., 1986. Transient elec-
Nitzanim area was carried out in the framework of the INCO- tromagnetic sounding for groundwater. Geophysics, 51, 995-
DC project financed by the European Commission (contract 1005.
# IC18CT96-0122). The seismic part of the research was con-
ducted by V. Shtivelman from GII. GOLDMAN, M., GILAD, D., RONEN, A., and MELLOUL, A.,
1991. Mapping of seawater intrusion into the coastal aquifer
The TDEM survey in the Sea of Galilee was supported by the of Israel by the time domain electromagnetic method.
Israeli Water Authority and by the Ministry of National Geoexploration, 28, 153-174.
Infrastructures. S. Hurwitz and H. Gvirtzman from the
Hebrew University of Jerusalem provided hydrogeological GOLDMAN, M., DU PLOOY, A. and ECKARD, M., 1994. On
support to this research. reducing ambiguity in the interpretation of transient electro-
magnetic sounding data. Geophysical Prospecting, 42, 3-25.
M. Ezersky from GII was involved in all stages of the TDEM
surveys. INTERPEX LIMITED, 1989, TEMIX user’s manual, pp. 126.

ISSAR, A., 1983. On the source of water from the ther-

REFERENCES momineral springs of Lake Kinneret (Israel). Journal of
Hydrology, 60, 175-183.
ARAD, A. and BEIN, A., 1986. Saline versus freshwater con-
tribution to the thermal waters of the northern Jordan Rift KAFRI, U., GOLDMAN, M. and LANG, B., 1997. Detection of
Valley, Israel. Journal of Hydrology, 83, 49-66. subsurface brines, freshwater bodies and the interface con-
figuration in between by the Time Domain Electromagnetic
ASSULINE, S, SHAW, M. and ROM, M.,1994. The effect of method (TDEM) in the Dead Sea rift, Israel, J. Env. Geology,
diversion and desalination of the saline springs on the salt 31:42-49.
concentration in Lake Kinneret and the water addition to the
national system. Mekorot water company, pp.44 (in Hebrew). MARCUS, E. and SLAGER, J., 1985. The sedimentary-mag-
matic sequence of the Zemach-1 well (Jordan-Dead Sea rift,
BEN-AVRAHAM, A., TEN-BRINK, U., BELL, R. and REZNIKOV, Israel) and its emplacement in time and space. Israel Journal

78
of Earth Sciences, 34, 1-10. SHTIVELMAN, V. and GOLDMAN, M., 1999. Integration of
MAZOR, E. and MERO, F., 1969. The origin of the Tiberias- shallow reflection seismics and TDEM for detailed study of
No’it mineral water association in the Tiberias-Dead Sea Rift the coastal aquifer in the Nitzanim area of Israel. Journal of
Valley, Israel. Journal of Hydrology, 7, 318-333. Applied Geophysics, 41 (in press).

SHAHARABANI, M., MICHELSON, H. and SIMON, E., 1980. SIMON, E. and MERO, F., 1992. The salinization mechanism of
Location of saline flows and seepeges along the eastern
Lake Kinneret. Journal of Hydrology, 138, 327-343.
shore of Lake Kinneret. TAHAL report 01/80/30 Tel Aviv, pp. 25
(in Hebrew).
STILLER, M., 1994, The chloride content in pore water of Lake

79