Caribbean Conference Short Paper

From GeoMod

Urbano, L., and Thibault, C., (in review). Pattern of saltwater contamination resulting from tsunami inundation of small islands, Proceedings of the 17th Caribbean Geological Conference 2005, San Juan PR.

Abstract

The inundation of coastal regions and small islands by the Indian Ocean tsunami on December 26, 2004 resulted in extensive salinization of shallow groundwater resources. In order to investigate the spatial and temporal consequences of tsunami inundation, we employ an idealized numerical model that simulates the effects of inundation on the freshwater lens of a small island. The model simulates the tsunami run-up and retreat, the storage and gradual seepage of seawater in shallow depressions, and the flushing of the aquifer by fresh recharge. Local depressions in topography collect saline flood waters and allow persistent seepage of saline water into the aquifer. However, the slow seepage of solutes through the unsaturated zone makes the unsaturated zone thickness a major control on the timing of aquifer salinization.

Introduction

Figure 1. Diagram of a freshwater lens and the analytical solution (Strack, 1998) for the maximum thickness (tf) of a freshwater lens beneath a small island. Figure from Urbano (2001).

The primary source of water on many small islands is often the fragile freshwater lens that is formed in the shallow subsurface due to the density difference between salt and fresh water (Figure 1). The thickness of the lens (tf) depends on the size of the island (L) and rate of recharge to the aquifer (R) and the hydraulic properties of the aquifer material (K) (Strack, 1998).

The most common threat to these freshwater sources is from over extraction from wells, which may cause the interface between salt and fresh water to move upward beneath the well in a process called upconing. However, storm surges and tsunamis can also flood large areas of small oceanic islands with saline ocean water that may contaminate groundwater sources from above. The transgression caused by rising sealevel during such storms is also a threat because it reduces the size (L) of the island. Processes that affect the recharge to the aquifer are also important. Climate change for example, in the form of a reduction in precipitation or an increase in evapotranspiration can also reduce the size of the freshwater resevoir (Essink, 2001; IPCC WGII, 2001).

In this paper we investigate the impact of tsunami flooding on contamination of the freshwater lens by introducing saline recharge from the surface. Contamination of fresh shallow groundwater supplies is expected to remain a significant issue for years in many areas affected by the Indian Ocean tsunami of December 2004 (IGRAC, 2006).

Method

The Numerical Model

Figure 2. Oblique view of the small island model, showing the initial freshwater lens (prior to saltwater inundation) and the numerical mesh. Salinity concentration (C) is shown as mass fraction (kg/kg).

Table 1. Model parameters.

The three-dimensional, numerical model SUTRA, published by the United States Geological Survey (Voss and Provost, 2003)\ used in this study is designed to simulate both variable density groundwater flow and continuous flow through saturated and unsaturated porous media. Variable density flow is necessary to simulate the infiltration of saline water into a fresh island aquifer and identify changes in the salt-water freshwater interface beneath the island.

An accurate description of partially saturated flow is essential for determining infiltration into the soil and into the aquifer during and after flooding. SUTRA solves the highly non-linear flow equations using a pressure-based, finite-element and integrated-finite-difference hybrid method. It solves the general equation for saturated and unsaturated flow using hexahedral finite-elements. This provides an accurate solution, but because the governing equations are not specially linearized to provide the most efficient solution for unsaturated flow, unsaturated flow models using SUTRA require a fine spatial discretization and can be very computationally intensive.

SUTRA also solves the solute transport equations using the finite element method, which permits the simulation of a diffuse interface between fresh and salt water, which offers a potentially more accurate solution than the sharp interface models often used to simulate the freshwater/salt water interface. The finite element solution, however, can lead to discretization errors and localized regions of anamolously high solute concentrations in areas with rapid spatial changes in concentration.

Idealized Island Model

The idealized, hypothetical island used to investigate the effects of tsunami inundation is based on the quarter island example described by Voss and Prevost (2003). This model assumes a circular island of radius 500 m. The symmetry of this island shape permits it to be simulated using just a quarter of the island (Figure 2). The model domain extends an additional 300 m offshore, which allows a better approximation of the saltwater/freshwater interface boundary at the edge of the island.

The simulation uses a freshwater recharge rate of 0.75 m/yr and uniform sediment permeability of 5x10-12 m2, which creates a saltwater/freshwater interface at approximately 50 m below the center of the island and allows a model thickness of 100 m. The hydraulic properties of the sediment and the fluid (including the equation of state for density) are shown in Table 1.

Model Simulation

Tsunami run-up, retreat and flushing are simulated in a three step process (Table 2). The first step is the initial inundation of the island by marine water during the tsunami. This step consists of one simulation: the inundation of the island for one hour to a depth of 1.8 meters. The second step is the simulation of seepage of ponded saline water collected in shallow depressions. This step involves two simulation runs. In the first simulation twenty five percent of the island remains flooded by marine water for two weeks. The amount of flooding is reduced to 10 percent for the second simulation which also lasts for two weeks. The third step is flushing of the system by freshwater recharge. This step consists of four two-week simulations and three three-month simulations. Normal recharge rates (Table 1) are used for each of the simulations which continue until the aquifer is flushed of saline water. The aquifer is considered flushed of saline water when concentrations of a selected group of nodes are all below a critical value of 500 ppm, which is the U.S Environmental Protection Agency's upper limit for drinking water.

In order to evaluate changes in salinity the same set of nodes were sampled for each simulation. The nodes were selected at a point close to the center of the island (X = 40 m and Y = 40 m) to extend from the surface at the site of a saline pond, to the point of saturation prior to tsunami inundation.

Table 2. Tsunami inundation simulation series.

Results

Temporal trends in salinization

Figure 3. Salinity change over time from the land-surface to the water-table (-4 m), for the high permeability simulation.

The model indicates that for the initial run salinity values in ponded regions remain near seawater concentrations (35,700 ppm) while the area is ponded. Values near the surface reach a high of approximately 22,000 ppm then moderately decline to approximately twenty percent of seawater concentration at three months (Figure 3). Beneath the land surface to the depth of the water-table salinity values plateau at values ranging from 8,000 ppm to 12,000 ppm before gradually diminishing. The lowest node slowly increases from 4,800 ppm at two weeks to 6,000 ppm at 3 months then slowly decreases to 3,000 ppm after 1 year.

Spatial trends in salinization

Figure 4. Map view of salinity in the saturated zone (at sea-level elevation beneath the island) after draining of all surface water (end of simulation 2b). Salinity concentration (C) is shown as mass fraction (kg/kg).

Figure 5. Map view of salinity in the saturated zone 9 months after tsunami (end of simulation 2b) at sea-level elevation beneath the island. Salinity concentration (C) is shown as mass fraction (kg/kg).

Salinity from the initial seawater inundation propogates through the vadose zone at a relatively constant rate, so the water-table is first affected by salinization at the island margins, where the water-table is closest to the surface (Figure 4). Salinization then propogates inland as the thickness of the vadose zone increases, such that after 9 months the coastal regions have been flushed of salinity while the water-table in inland areas is still receiving saline water drained from the unsaturated zone (Figure 5).

Discussion

In our simulation, the primary control on vadose zone residence time is the the thickness of the unsaturated zone, although it is to be expected that the permeability of the soil is another significant controlling factor. The model also indicates that the near-surface unsaturated zone is affected more intensely but recovers quickly compared to water deeper in the aquifer. The near-surface salinity begins to fall immediately after the retreat of the saline surface water and drops to 50% of the maximum within 2 weeks. At 3 meters beneath the surface the maximum salinization occurs 2 weeks after the end of saline surface recharge and the drop to 50% takes over 7.5 months. These timescales assume constant recharge of fresh water after the effects of the tsunami have passed, thus local climatic conditions such as wet and dry seasons will have a significant impact on salinity leaching times.

Addressing the impact

Addressing the impacts of tsunami inundation is a long term issue, however the slow supply of salinity to the aquifers suggests that amelioration strategies, such as switching from water-table to deeper wells, may prove useful on small islands. Wells screened beneath the water-table are likely remain free of salinization for longer than water-table wells. Although pumping may focus salinity toward the wells, the dilution of the salinity pulse as it mixes into the saturated zone would likely reduce the impact of salinization on deeper wells. In addition, the relatively low salinity of water reaching the water-table as demonstrated by the model suggests that it is unlikely that the salinity at the water-table will substantially impact the upconing of saline water beneath the well.

Conclusions

The model presented confirms that large inundations of saline water following events such as a tsunami can have long-term impacts on water quality. Water-table salinization propagates from the coastline inward and is primarily controlled by the residence time of salt in the unsaturated zone. The slow leaching of water into the aquifer, however, does suggest that the temporary use of deeper wells may serve to supplement water supplies until salinity has been flushed from the system.

References

Essink, G., 2001. Improving fresh groundwater supply- problems and solutions, Ocean and Coastal Management, v.44, p.429-449.

IGRAC, 2006. Impact of the 26-12-04 tsunami on groundwater systems and groundwater based water supplies. International Groundwater Resources Assessment Centre Special Report, http://igrac.nitg.tno.nl/tsunami1.html (February 6th, 2006).

IPCC WGII, 2001. Climate Change 2001: Impacts, Adaptation and Vulnerability, Cambridge University Press, Cambridge.

Strack, O., Groundwater Mechanics, Prentice Hall, New Jersey, U.S.A., 1988, 245 p.

Voss, C. and Provost, 2003, A model for saturated-unsaturated, variable-density ground-water flow with solute or energy transport. U.S. Geological Survey Water-Resources Investigations Report 84-4369, 260 p.

Urbano, L., 2001, Ground-water and climate change [PhD Thesis]: University of Minnesota, 163 p.