Recharge and Flow Mechanisms in the Vadose Zone of Carbonate Aquifers

Groundwater recharge and contaminant transport in fractured karstic aquifers are
difficult to quantify due to the heterogeneity and complexity of the rock mass,
including preferential flow paths along karst conduits. The present study aimed to
assess how changes in the lithology and structural differences in fractured karst
systems influence flow and transport in the unsaturated zone. One particular goal of
this research was to develop a mathematical model for quantifying water flow and
contaminant transport processes in the karst/fractured-porous unsaturated zone and
groundwater.
The research was conducted within the fractured carbonate Western Mountain aquifer
(Yarkon-Taninim) of Israel, one of the country's major water resources which
partially flows through a karst system. The aquifer extends from south of the Carmel
Mountains, in the north, to the Sinai Desert in the south and from the Judea and
Samaria Mountains in the east to the Mediterranean coastline in the west. The aquifer
is composed of the Judea group, which mainly contains carbonate sections aged Late
Albian to Turonian (Arkin and Braun, 1965; Arkin and Hamaoui, 1967; Ben Gai et
al., 2007). The group represents a stable carbonate platform depositional sequence
(Sass and Bein, 1982). Syndepositional and postdepositional environmental
conditions created spatial variations in carbonate contents which created different
basic settings for the aquifer’s development. The sedimentary sequence, when
saturated, is divided, traditionally, into two separate limestone/dolomite sub-aquifers
by the Moza formation’s chalk/marl aquitard (Bida, 1986; Guttman, 1986; Mercado,
1980). However, other sub-aquifers exist locally above and between relatively
impermeable layers. This division is mostly noticeable in the phreatic and unsaturated
zone where perched aquifers exist, mainly at the Aminadav formation with some at
the Soreq and the Kefira formations. Yet, the relatively impermeable aquitards do not
completely prevent water from being transferred between the sub-aquifers. In some
places, water is transferred between the sub-aquifers directly because of a lateral
facies change in which the aquitard is missing. In other places, severe fracturing
breaks the aquiclude continuity, enabling different aquifers to be hydrologically
connected (Peleg and Gvirtzman, 2010; Weiss and Gvirtzman, 2007).
The experimental investigation included monitoring changes in groundwater level,
surface water, raw sewage flow, and assessment of groundwater and runoff quality.
The numerical models were applied to analyze observation results.
The observed groundwater levels were analyzed with a one-dimensional, dual
permeability numerical model for water flow in variably saturated fractured-porous
media. The model was calibrated and used to estimate groundwater recharge in nine
locations. The recharge values exhibit significant spatial and temporal variation with
mean and standard deviation values of 216 and 113 mm/year, respectively. Based on
simulations, relationships were established between precipitation and groundwater
recharge in each of the nine sites studied and compared with similar ones obtained in
earlier regional studies. Simulations show that fast and slow flow path conditions also
influence the annual cumulative groundwater recharge dynamic. In areas where fast
flow paths exist, most of the groundwater recharge occurs during the rainy season
(60–80% from the total recharge for the tested years). In locations with slow flow path
conditions, the recharge rate stays relatively constant with a close to linear pattern and
also continues during the summer season.
A lumped continuous model (HEC-HMS) was applied to simulate runoff flow and
karst aquifer recharge in three sub-basins along the ephemeral Soreq stream. Two
primary infiltration mechanisms were identified during the study. The first is diffuse
groundwater recharge, which occurs at hillslopes over the majority of the study area.
The second infiltration mechanism is direct groundwater recharge, which happens as a
result of surface flow infiltration through highly developed karst conduits along the
Soreq creek. Results of simulations show that such an approach can be used to
describe the spatial distribution of groundwater recharge. The calculated recharge
volumes align with results from previous studies. However, the present methodology
also allows for the quantification of the annual recharge via direct infiltration through
the creek bed. In the highly developed karst system studied, direct infiltration
accounted for 19.2% of the annual recharge volume. Moreover, the present
methodology allows for the assessment of potential infiltration from the creek bed.
Subsequently, the users of the methodology are able to evaluate watershed
management practices that could potentially improve artificial enrichment of the
aquifer.
A quasi 3D dual permeability mathematical model was developed and applied to
simulate the Carbamazepine (CBZ) transport in both the vadose and saturated zones
of the karst aquifer. The results of the simulation show that after the sewage leakage
stopped, significant amounts of CBZ (up to 95%) were retained in the porous matrix
of the unsaturated zone, below the source zone. Water redistribution and slow
recharge during the dry summer season contributed to elevated CBZ concentrations in
the groundwater in the vicinity of the creek and tens of meters downstream. During
fast flow events the aquifer vulnerability increased due to: low solute exchange rate,
lesser diffusion of contaminant into the matrix and a greater mass of CBZ reaching
the groundwater from the unsaturated zone. As a result, there was a larger spread of
CBZ throughout the aquifer.
Finally, the quasi 3D dual permeability mathematical model was used to simulate
transport and attenuation of CBZ and of Caffeine (CAF), as conservative and reactive
tracers, respectively. Most model parameters were estimated by using a CBZ
breakthrough curve from an observation well, while 1st order decay and linear
sorption coefficients were assessed for CAF. The estimated half-life and the partition
coefficients of CAF were 7.6 days and 0.1 L/kg, respectively. The results of the
simulation showed that by the end of the year, significant amounts of CBZ were
retained in the porous matrix of the unsaturated zone below the creek, and tens of
meters downstream in the groundwater; while all the CAF was degraded soon after
the leakage stopped.
The outcomes of this research indicate that this modelling approach can be useful to
describe major mechanisms of flow and transport under the considered conditions.
However, the main limitation of the applied model is the assumption that there is only
vertical flow in the unsaturated zone. In places where low permeability layers in the
unsaturated zone extend significant distances, conditions for developing perched
water bodies and essential lateral flow may limit use of 1D or quasi 3D models.