Numerical modeling of arsenic mobility - In-Situ Remediation of Arsenic-Contaminated Sites - page 35

Unraveling observed water quality patterns in aquifers can be a daunting task. The input of
different sources and types of water is the first of factors that adds to this complexity. Sources
include precipitation, surface waters, seawater, ascending deep groundwater and anthropogenic
sources such as wastewater. The overprint of geochemical processes adds to the complexity
since the water composition is altered as it travels through the subsurface. Mineral dissolution
and precipitation, ion exchange and redox reactions are generally the most important chemical
processes that affect groundwater quality. Mixing of different water types through physical pro-
cesses such as advection, dispersion and diffusion also exerts an influence on the composition of
groundwater.
Traditional methods that are applied to interpret large hydrochemical data sets involve plotting
(Fetter, 2001) and classification of samples into groups (e.g., Stuyfzand, 1993) in order to be
able to discern regional trends and to identify chemical processes. As useful as these visual and
statistical methods can be in deriving preliminary conceptual models for the reactions that give rise
to the range of observed groundwater compositions, they are incapable of determining whether the
inferred reactions are thermodynamically feasible. Geochemical and reactive transport models
that quantitatively simulate reactions subject to the laws of thermodynamics are therefore a natural
choice for confirming whether a particular conceptual model is also chemically realistic. Or, as
Lichtner (1996) suggested: 'Quantitative models force the investigator to validate or invalidate
ideas by putting real numbers into an often vague hypothesis and thereby starting the thought
process along a path that may result in acceptance, rejection, or modification of the original
hypothesis'.
The first quantitative geochemical models were originally developed in the 1960s, initially
with the sole objective of calculating the speciation of dissolved ions. These original models were
soon improved and extended to include chemical reactions that alter the water composition such
as mineral equilibria and cation exchange (e.g., Parkhurst, 1980; Plummer, 1983). Geochemi-
cal modeling plays an important role as a stand-alone tool for understanding controls on water
chemistry, but is also a key part of the conceptual model development and testing for coupled
reactive transport models. In the context of groundwater arsenic problems, typical questions and
tasks that can be addressed with geochemical models can include:

• What are the predominant arsenic species (complexes) present in groundwater?
• Is groundwater under- or over-saturated or at equilibrium with respect to specific minerals?
• Which reactant(s) could be added to groundwater of a specific composition to reduce dissolved
arsenic concentrations?
• How does arsenic partition between aqueous and sorbed phases?
• Could changes in water composition and/or pH lead to arsenic release from or uptake by the

aquifer solids?

There are several commonly used public domain and/or commercially available software for
carrying out these tasks including PHREEQC, MINTEQA2 and the Geochemist's Workbench
(see Table 2.1 for more details).
The construction of numerical groundwater flow models on the other hand can provide an
in-depth analysis of the flow system and is a prerequisite for subsequent solute and reactive trans-
port modeling endeavors. Numerical models are necessary especially for simulation of systems
characterized by geological heterogeneities and/or complex transient hydrologic forcings. Typical
questions that can be explored with numerical groundwater flow models include:

• What are the flow velocities in the vicinity and particularly downstream of an arsenic pollution

source?

• Could an arsenic plume be contained by a specific pumping well configuration?

The USGS groundwater model MODFLOW (Harbaugh, 2005; Harbaugh et al. , 2000) is the
most widely used code for simulation of multi-dimensional flow fields, mostly owing to the