Geochimicaet CosmochimicaActa, Vol. 58, No. 23, pp. 5131-5145, 1994
Copyright0 1994ElsevierScience Ltd
Printed in the USA. All rightsreserved
Pergamon
0016-7037/94 $6.00 + .OO
0016-7037(94)00294-O
Dynamic model of the genesis of calcretes replacing silicate rocks in semi-arid regions
YIFENG WANG,
I,* DANIEL
NAHON,’ and ENRIQUE MERINO’
‘Department of Geological Sciences, Indiana University, Bloomington, IN 47405, USA
*Laboratoire de GCosciences de I’Environnement, U.R.A. CNRS No 132, Facultt de St. Jtrbme,
Universitt d’Aix-Marseille 111,13397 Marseille Cedex 13, France
(Received July 15, 1993; accepted in
revisedform July 13, 1994)
Abstract-In
both pedogenic and groundwater
calcretes, calcium carbonate precipitates in voids, or
displacing other grains, or replacing underlying parent silicates. Replacement textures are widespread in
pedogenic calcrete. Many calcretes also contain magnesium layer silicates and minor chert.
We present a reaction-transport
model that accounts for the genesis of replacement
in calcretes and
for their mineralogy. Replacement is difficult to account for geochemically because it requires simultaneous
removal of large amounts of silicates and import of also large amounts of CaC03. In the model the
genesis of replacement
is directly related to seasonally alternating dry-wet climates and to appropriate
groundwater (or circulating soil water) compositions.
In a dry season, water evaporation causes CaC03
and sepiolite (or attapulgite) to precipitate. If groundwater contains enough Mg’+, sepiolite precipitation
by the chemical-divide
mechanism depletes SiOz( aq), resulting in the dissolution of parent silicates. In
the following wet season, sepiolite dissolves fast, and silica and cations are flushed away by rainwater,
making room for CaC03 precipitation
in the next dry season. As climate cycles repeat, CaC03 is accumulated and silicates are removed. The sepiolite (or attapulgite, or Mg-smectite)
serves as a temporary
storage of silica between seasons. If the groundwater
contains too little aqueous Mg then the model
predicts growth of calcium carbonate without removing silicates, thus producing void filling and or
displacive textures instead of replacement.
The model consists of a set of nonlinear partial differential equations taking account of mass conservation,
dispersion, advection, rainwater infiltration, evaporation, and the kinetics of mineral reactions. The hydrodynamics
of unsaturated
media is applied in determining
water flow in calcrete profiles. Wet/dry
seasonal changes are incorporated by alternating the upper boundary conditions. The model successfully
produces the mineral and textural zonation observed in many calcretes (namely, at the bottom of the
profile the parent rock is first replaced by sepiolite, only part of which is in turn replaced by calcite,
whereas at the top of the calcrete the sepiolite is itself completely replaced by the calcite, which appears
to “directly” replace the parent rock). The model can produce calcrete near the Earth’s surface well
above the water table. Low Pco,, intensive evaporation, and long dry seasons all are predicted to produce
thicker calcretes.
Calcretes constitute an effective geochemical tool that, via replacement,
removes parent silicate rocks
and shapes the landscape of semi-arid countries. The model herein provides a mechanism that accounts
for the efficiency of replacement in removing silicates.
INTRODUCTION
carbonate is thought to be eolian in calcretes of the southwestern USA (e.g., MACHETTE, 1985; MARION et al., 1985;
MAYER et al., 1988) and in some calcretes of South Australia
and Victoria (QUADE et al., 1994). But the eolian contribution is negligible in west African calcretes (NAHON and
RUELLAN, 1972; BOULET, 1974; NAHON, 1976; MILLOT et
al., 1977; DUCLOUX et al., 1984). In western Senegal, calcretes
are well developed on marls and limestone but not on kaolinitic sandstones, especially not on sandstones by the ocean
(NAHON, 1976). In the Sahara, the “Sr/‘%r ratio of calcretes
indicates a calcium source from parent schists ( NAHON, 1976,
p. 183).
Two types of calcrete are distinguished,
pedogenic and
groundwater
calcretes (WRIGHT and TUCKER, 1991). The
latter are formed at or near water tables. Calcretes display
void-filling, displacive, and replacive textures (WATTS, 1978;
CHADWICK and NETTLETON, 1990; NAHON, 1976). As we
show below, the replacive texture poses the hardest problem
from a geochemical standpoint.
Calcite accumulation
in soils has been simulated with
compartment
models by MCFADDEN and TINSLEY (1985),
MARION et al. (1985), and MAYER et al. (1988). These
SINCE THE WORK OF GILE ( 196 1) and GILE et al. ( 1966) in
New Mexico, USA, and MILLOT et al. ( 1969) and RUELLAN
( 197 1) in Morocco, it has been well known that calcium
carbonate precipitates in soils and weathering profiles of semiarid regions, forming so called calcretes. Reviews of the genesis
and properties of calcrete can be found in GOUDIE ( 1983),
WRIGHT and TUCKER ( I99 1 ), and MILNES ( 1992). Calcretes
constitute a sharp geochemical tool that shapes the landscape
in semi-arid countries (e.g., NAHON, 1976; REEVES, 1983).
Because of their widespread occurrence through geological
history and their close association with semi-arid climates,
calcretes have been used as paleoclimate
indicators (e.g.,
HUBERT, 1978).
Calcrete formation
involves import of CaC03, whose
source is considered to be eolian or the parent rock or upstream Ca-bearing rocks. In all cases, water is implicitly essential in dissolving and transporting the CaC03. The calcium
* Present address: School of Earth and Atmospheric Sciences,
Georgia Institute of Technology, Atlanta, GA 30332-0340, USA.
5131
5132
Y. Wang, D. Nahon,
models take account of only eolian dust source of calcium
carbonate and only calcite growth, and thus they fail to account for some of the mineralogical
and textural changes
observed in calcrete profiles.
Our purpose in this paper is to propose a geochemical
model of the genesis of calcretes
replacing
underlying
silicate
rocks. In the model, calcrete formation is driven by the wet/
dry seasonal alternations of semi-arid climates. The model
allows us to estimate rates of calcrete formation, and, by
numerical simulation, to predict the response of calcrete formation to climatic and groundwater-composition
changes.
The model incorporates both short and long timescales. We
begin by summarizing descriptions of the mineralogy, texture,
and occurrence of calcretes in order to ascertain the main
features and constraints that need to be included in the model.
SYMBOLS
activity of chemical speciesj
dispersion coefficient, cm’/s, Eqns. 22-25
constant characterizing
rock capillarity, Eqn. 2 I
effective permeability, a function of 8, cm/s, Eqn. 20
effective permeability
for saturated rock, cm/s, Eqn. 20
equilibrium constant of reaction i, Eqns. 9- 12, 14- 17
rate constant of reaction i, mol/cm3 S, Eqns. 9-12
rate constant of reaction i scaled by dispersion, Eqn. 45
depth of water table, cm
molality of chemical component j
molality of component j in rainwater, Eqn. 3 I
molality of component j in the groundwater at Z = L
constant characterizing
rock capillarity, Eqn. 2 I
partial pressure of CO2 in pore air, atm, Eqn. I4
water flux, cm3 of H20/cm2 s, Eqn. I8
evaporation rate in dry seasons, cm3 of H20/cm2 s, Eqn. 28
infiltration rate in wet seasons, cm3 of H20/cm2 s. Eqn. 30
scaled evaporation
rate in dry seasons
scaled infiltration rate in wet seasons
rate of reaction i, mol /cm3 s, Eqns. 9- I2
typical timescale for calcrete formation. Eqn. 47
time. s
volume fraction of mineral i
volume fraction of mineral i in parent rock
Peclet number, characterizing the relative importance of water
advection to dispersion, Eqn. 45
molar volume of mineral i, cm3/mol
distance from the Earth surface, cm
time increment on the scale of days to solve the seasonal variation of the system, Eqn. 48
time increment on the scale of 1000 years to solve mineral
volume changes on large timescale, Eqn. 48
saturation degree, the fraction of pore space occupied by liquid
water, Eqn. 19
small positive constant, Eqns. 9- I2
residual moisture
relative saturation degree, defined as (0 - O,)/( 1 - 0,)
density of water, 10m3 kg/cm3
scaled time, Eqn. 46
porosity
hydraulic head (cm), Eqn. I8
CALCRETES: MACROSCOPIC
DESCRIPTION
Calcretes may reach several meters in thickness,
semi-arid zones with 400-600 (or less) mm annual
and tend to develop preferentially on low-angle fans
iments or glacis ( GOUDIE, 1983). They can form
form in
rainfall,
and pedon rocks
and E.Merino
with little or no calcium ( GILE et al., 1966; RUELLAN, 1971)
in semi-arid regions, but, in more humid areas, they develop
only on Ca-bearing parent rocks ( NAHON and RUELLAN,
1972; DUCLO~X et al., 1984). Calcretes usually display welldefined vertical and lateral sequences. From bottom to top,
and also from uphill to downhill, the calcium carbonate occurs successively as diffuse patches, unindurated nodular accumulations, indurated nodules, plates, and a hard crust; the
general sequence is always from nodular accumulations
toward the bottom (or uphill) to planar precipitates toward
the top (or downhill). See, e.g., REEVES ( 1970) for the calcretes near Lubbock, Texas, USA, RUELLAN ( 197 1) for the
calcretes of northeast Morocco, HAY and REEDER ( 1978 )
for the calcretes of Olduvai Gorge and Ndolanya Beds in
Tanzania, and NAHON ( 199 1, p. 166) for calcretes from
Mauritania. Contacts between adjacent horizons in the sequence are gradual. The orientation of the bunches and plates
may be controlled by the texture of the host rock, especially
ifthe latter is a schist ( WIEDER and YAALON, 1974; WIEDER,
1977) or if it is fractured (MILLOT et al., 1977).
MINERALOGY
OF CALCRETES
Calcretes consist of authigenic carbonates, which are dominant, Mg clays, and chert, plus relict minerals of the parent
rock. The proportion of the latter decreases from bottom to
top (or downhill ). The main carbonate is calcite. The earliest
carbonates
are disordered,
hydrated,
or organic-matterbearing calcite ( DUCLOUX et al., 1984; DUPUIS et al., 1984).
and they may include aragonite, but all of them quickly recrystallize into the ordered calcite that makes up most of the
calcrete. The transition occurs over tens of micrometers, and
has been reproduced experimentally
( DUPUIS et al., 1984).
Aragonite has been identified in calcretes from South Africa,
the Mediterranean
region, and West Africa ( GOUDIE, 1973;
NAHON et al., 1980). Magnesian calcite is rarer ( MILLOI- et
al., 1977). Both aragonite and magnesian calcite are always
minor relative to well crystallized calcite.
Sepiolite [ Mg&O,,(
OH)2 - HZ01 and attapulgite (= palygorskite) [ ( Mg,Al,Fe).,SisOzo * nH*O] are common in calcretes (e.g., WATTS, 1980; HAY and WIGGINS, 1980; ARAKEI.,
1986; JONESand GALAN, 1988; MONGER and DAUGHERTY,
199 1b). Progressive growth of calcite is accompanied
by increasing attapulgite, which reaches its peak toward the top
of the nodular horizon, from where it decreases in amount
and disappears upwards ( MILLOT et al., 1969; GOUDIE, 1983,
in Fig. 4.2). In more humid areas smectite grows rather than
attapulgite, and it also disappears in the platy and crust horizons at the top (NAHON et al., 1975; MILLOT et al., 1977).
Petrographic, microprobe, and X-ray diffraction work ( MILLOT et al., 1977) shows that clay growth proceeds as host
mineral grains dissolve.
Minor amorphous silica or chalcedony occur in calcretes,
toward their base ( NAHON, 1976). Microscopically, the chert
occurs adjacent to partly dissolved parent quartz grains ( DUCLOUX et al., 1990, photo 3a). The chert described by DuCLOUX and LAOUINA ( 1988) contains minor magnesium. In
central Australia dolomitic and chalcedonic replacements of
calcrete generally coincide with present drainage channels in
washout areas ( ARAKEL, 1986 ).
Calcrete replacement of silicate rocks
REPLACEMENT OF HOST MINERALS BY CALCITE
Many calcretes, particularly pedogenic ones in semi-arid
countries, display replacement of parent silicate rocks by calcite( NAGTEGAAL, 1969; ARISTARAIN,1971; BOULET, 1974;
CHAPMAN, 1974;NAHONet al., 1975; NAHON, 1976; MILLOT
et al., 1977; DURAND, 1979; BECH et al., 1980; REHEIS, 1988;
DUCLOUX et al., 1990; MONGERand DAUGHERTY,1991a,b).
Microscopic and macroscopic observations show conclusively
that the replacement is pseudomorphic, regardless of the nature of the parent rock (granite, basalt, sandstone, argillite,
shale, schist, quartzite). Textures and structures of the parent
rock are preserved.
Because of its importance and solidity, we summarize here
some evidence. In a Precambrian greenschist of southern
Morocco, a calcrete develops progressively upward along the
joints and schistosity (Fig. 1) ( MILLOT et al., 1977). The
calcite occurs as a millimeter-thick “grid,” discontinuous at
bottom, continuous higher up. The “bars” of the grid separate
volumes of greenschist of progressively smaller size upward.
The schistosity planes and quartz veins can be traced from
volume to neighboring volume. The calcite making up the
bars of the grid contains innumerable tiny relicts of greenschist
that preserve the original orientation, a detail which precludes
the interpretation that the calcite is a fracture filling. The
evidence is conclusive, and the complete description should
be studied in Millot et al. ( 1977): the progressive calcitization
took place without disturbing the original schistosity or the
original bulk volume. Similarly compelling observations also
demonstrating pseudomorphic replacement by calcium carbonate have been carried out on other schists in Mauritania
(NAHON, 1976) and on granites, shales, and quartzites in
Morocco ( MILLOT et al., 1977).
Microscopic evidence also shows pseudomorphic replacement conclusively. In the coarse-grained Ifni granite of
southern Morocco ( MILLOTet al., 1977) calcite microcrystals
grow first along grain contacts and cleavage planes (Fig. 2C).
With progressive growth higher in the profile, the microcrystals form a microgrid that cuts up each grain of feldspar
or quartz into small islands. These islands display optical
continuity, and twin planes can be traced from island to island
(Fig. 2D,F). Biotite grains are replaced by micritic calcite
layer by layer or normally to the cleavage, but leaving in
place, unreplaced, any opaque inclusions contained in the
fresh biotites (Fig. 2E).
Significantly, the replacement exhibits a typical macroscopic zonation (e.g., MILLOT et al., 1977). Toward the bottom of a calcrete profile, silicate minerals are replaced by
attapulgite, which is in turn partly replaced by calcite (Fig.
1C,D,E) , but toward the top of the profile the replacive calcite
occurs adjacent to relict silicate minerals (Fig. 1F) .
QUALITATIVE GEQCHEMICAL MODEL
OF CALCRETE FORMATION
From the above summary of evidence we retain the following basic features and constraints that have significant
geochemical implications for the genesis of calcretes:
1) Calcretes form typically in semi-arid regions, where climate is characterized by alternating dry and humid seasons.
5133
2) Pseudomorphic replacement of underlying parent rocks
of any kind, including silicate rocks with little or no calcium, implies import of a great volume of CaC03 and
simultaneous removal of a huge amount of silicates. If
calcium carbonate were thought to accumulate through
inadequate leaching in semi-arid areas (G~UDIE, 1983,
p. 112), how would parent silicates, being generally less
soluble than calcium carbonates, be leached out? Thus,
any model that only considers carbonate accumulation
without also accounting for silicate removal misses a crucial half of the geochemical problem.
3) Toward the bottom of calcrete profiles, silicate minerals
are replaced by attapulgite (or another Mg clay), which
is in turn partly replaced by calcite. Toward the top, the
replacive calcite is adjacent to the relict parent grains.
4) Precipitated amorphous silica may occur adjacent to halfdissolved quartz grains, in spite of the apparent contradiction implied by the high solubility of the former and
low solubility of the latter.
We propose here a dynamic model of calcrete formation
that solves the problems listed above. In the model the process
is driven by the alternation of dry and wet seasons, the mechanism of chemical divides (driven by evaporation) plays a
crucial role, and the major geochemical problem involved is
that of supplying and precipitating calcium carbonate while
dissolving and removing silicates. To see how the model
works, consider the replacement of quartzite by calcrete. Assume that the groundwater (or circulating soil water) brings
in dissolved Ca, Mg, and SiO*( aq). In a dry season, evaporation of water concentrates all solutes in the porewater at
the drying front. Calcite and sepiolite begin to precipitate. If
the molality ratio of Mg to SiOZ in the incoming water is
greater than 2 / 3, the precipitation of sepiolite depletes SiOl
in porewater while Mg continues to be concentrated as evaporation proceeds (this works as in the chemical divides of
HARDIE and EUGSTER, 1970). Eventually, SiO*(aq) concentration becomes so low that quartz starts to dissolve. Dissolved silica contributes to form more sepiolite. Note that
quartz dissolution here is unavoidable even if the initial water
was saturated with quartz. In the following wet season,
downward percolating rainwater dissolves sepiolite rapidly,
thus massively removing silica from the system (although
minor amorphous silica or chalcedony may still precipitate
locally due to the high concentration of SiOZ). The net effect
of one year’s climate cycle is that some silica is evacuated
and some calcium carbonate accumulates. After thousands
of yearly climate cycles, a calcrete forms.
Because only a tiny volume of the parent rock is removed
in each climate cycle and the space left is immediately filled
by calcite in the next dry season, the original rock textures
and structures are easily preserved during calcretization. Note,
however, that this pseudomorphic replacement is different
from that caused by crystallization force, a mechanism that
forces the dissolution rate of the replaced mineral to be exactly
equal to the growth rate of the replacing mineral ( MALIVA
and SIEVER, 1988; MERINO et al., 1993). During calcrete
formation, the two rates are certainly not equal on the seasonal scale, but they are roughly equal over the time span of
a year.
5134
Y. Wang. D. Nahon. and E. Merino
Calcrete replacement of silicate rocks
Furthermore, sepiolite precipitated in a dry season may
be completely dissolved in the following wet season near the
top of the profile, where the porewater is close to rainwater
in composition and therefore very undersaturated with sepiolite. As a result, parent rocks appear to be directly replaced
by calcite toward the top of profile, even though the absent
sepiolite still plays a key role there in removing silica.
The proposed mechanism also works for a calcrete replacing aluminosilicate rocks, and then attapulgite (or even
smectite+ if climates are more humid), which contains Al,
would play the same role as sepiolite does in the chemical
divide. In addition, the eolian source of calcium carbonates
is not excluded in the mechanism. As long as eolian sourced
calcium carbonate undergoes dissolution and reprecipitation
in forming calcretes, and if Mg concentration is high enough
in porewater, the proposed mechanism still works. The scenario described above is simulated below with a transportreaction model, in order to further clarify the effect of climate,
hydrology, and chemical kinetics on calcrete formation.
5135
Note that reaction 1 only proceeds to the right, but reaction
4 may go either way. For the heterogeneous
reactions l-4,
we adopt kinetic laws as follows:
Iotherwise
R2 = k2 -
Chemical Reactions
For simplicity, we only consider a calcrete replacing
quartzite.* The following chemical reactions take place:
SiOZ (quartz) + 2Hz0 + H4Si04
(1)
CaCOX (calcite) + H+ G CaZt + HCO;
(2)
MgSi1.503.75(OH)~.5(H~0)1.5 (sepiolite)
+ 2H+ + 0.25H20 * Mg*+ + 1.5 H4Si04
Si02 (amorphous
silica) + 2H20 = H4Si04
CO2 + Hz0 = H+ + HCO,
HCO;
= H+ + CO:-
HzO @H++OHH4Si04 * H+ + H,SiO;.
(3)
(4)
(5)
(6)
(7)
(8)
+ The mechanism may not work so effectively with smectite as
with sepiolite or attapulgite. The latter two minerals are known to
dissolve fastest among silicates (BRADY, 1992).
t This choice of parent rock is made to cut down on the number
of reactions below. There is no problem in adding more reactions
needed for other parent rocks to the model.
0
(30 +
(3
l-
a&+aHCO5
an+&
(
)
(10)
(11)
(12)
R4=k4&(1-y)
and they must satisfy
Ri < 0
TRANSPORT-REACTION MODEL OF
CALCRETE FORMATION
0
when
V, = 0,
i = l-4,
(13)
where R, is the rate of reaction i (mol/cm3 s); ki and Ki
are, respectively, the rate and equilibrium
constants of reaction i; uj is the activity of chemical speciesj; 0 is saturation
degree, the fraction of pore space occupied by liquid water;
0,, is a small positive constant; V, is the volume fraction of
mineral i.
The reactive surface area of a mineral may not necessarily
decrease as the mineral dissolves, because new reactive surface
area may be created by disaggregating large aggregates into
small pieces (Figs. 1F and 2D). Secondary minerals, usually
forming coatings around parent mineral aggregates, may have
large reactive surface areas even where a small amount of
them is present. For simplicity, reactive surface areas are taken
to be constant in Eqns. 9-12 and are buried in ki values.
Equation 13 simply means that no dissolution is allowed for
a mineral which is not present. The factor 0/( 0. + 0) (with
0 = fraction of pore space occupied by liquid water) in Eqns.
9- 12 represents the degree of contact between liquid water
and minerals. This factor is needed because minerals can
only dissolve or grow to the extent that there is liquid water
in the pores.
According to the Ostwald ripening rule (e.g., MORSE and
CASEY, 1988), an unstable phase is more favored to form
than a stable one from solution in a low-temperature
environment.
If the latter grows too slowly compared to the
former. Therefore, in our model quartz precipitation
is not
FIG. 1. (A) Calcrete replacing a Precambrian greenschist (Anzi Series, 30 km east of Tiznit, southern Morocco).
Schistosity dips 15’ off the vertical. The rock is densely jointed across the schistosity. The calcrete develops progressively
from bottom to top of a 5-m profile. In the lower part of the profile, the original schistosity is preserved by a ‘grid” of
calcium carbonate (light), and the relict islands (dark) of greenschist retain their original orientation, indicating that
the reolacement is macroscopically pseudomorphic. (See Plate I- 1, MILLOTet al., 1977.) ( B) Detail of the top of the
calcrete profile (hard crust):.in the indurated calcrete (c), the original structure of greenschist (gs) is still preserved.
Relicts of greenschist are “floating” on massive calcite matrix and still keep the original orientation, with no displacement.
(C) Base of the calcrete profile on greenschists. Early stage of pseudomorphic replacement of the greenschist (gs) by
attapulgite (a) from microfissures (f); (crossed nicols). (See Plate 11-5, MILLOTet al., 1977.) (D) Detail of Fig. 1C.
Attapulgite (a) is in turn replaced by a small amount of calcite (c) along microfissures. (See Plate 11-7, MILLOTet al.,
1977.) (E) Higher in the same calcrete profile, calcite increases and attapulgite decreases in amount. But still the
greenschist (g.. is first replaced by attapulgite (a), which is then replaced by calcite; (crossed nicols) . (See Plate 11-9,
MILLOTet al., 1977.) (F) Top portion of the calcrete profile (hard crust). The greenschist (gs) is “directly” replaced
by calcite (c); (crossed nicols). (See Plate 11-10,MILLOTet al., 1977.)
5136
Y. Wang. D.
Nahon. and E. Merino
FIG. 2. (A) Calcrete developed on the Messti granite, 5 km from Ifni, southern Morocco. At the base of the profile,
calcrete (c) replaces granite(g); note the apparently sharp reaction front. (B) The calcite formed by replacement, not
by fracture filling, as shown by the preservation of the crystalline-granular granite texture in the (white) calcrete. Also,
note that the front is not as sharp as it looks in (a). (See Plate I-3, MILLOTet al., 1977.) (C) Incipient replacement of
a plagioclase ( pf) by calcite (c) with no displacement of the twins. (See Plate III- I 1,MILLOTet al., 1977.) (D) Advanced
stage of p~udomo~h~c replacement of a plagioctase by &cite. with no displacement of twin planes. (See Pfate fit12, MILLOTet al.. 1977.) (E) Pseudomo~hic replacement of biotite (b); ghosts of cleavages are still preserved (arrow)
near the reaction front. (F) The fragments of the original grain still maintain optical continuity, indicating that the
replacement took place without any displacement. ( MILLOTet al.. 1977).
5137
Calcrete replacement of silicate rocks
as is implied in Eqn. 9, whereas amorphous silica
can precipitate or dissolve.
The dependence of quartz reaction rates on concentrations
is based on DOVE and CRERAR ( 1990) and RIMSTIDT and
BARNES( 1980); calcite rates are based on MORSE (1983)
and INSKEEPand BLOOM( 1985). Because of the lack of appropriate kinetic data, the exponents on concentrations in
the rate law of sepiolite dissolution are chosen to equal stoichiometric coefficients in reaction 3.
Reactions 5-8 are assumed to be fast and thus in equilibrium. The equilibrium condition is imposed by
allowed,
e =
- (1 - el/m)q2
[l +(hl*l”)]-”
I 1
(20)
for
9G
0
for
\k > 0
(21)
with m = 1 - 1 /n. Here K( 1) is the effective permeability
for saturated rock; n and h are two constants characterizing
rock capillarity.
(14)
Continuity Equations for Solutes
= &aHcos
(15)
Solute concentrations in porewater satisfy the following
mass continuity equations:
=
aH+aoH- = K,
aH+aH$ioi
K(0) = K( l)e”.5[i
K5 PCO,
an +aHCOT
&+a&
(s). The effective permeability K and hydraulic head * are
related to the relative moisture 0 by (VANGENUCHTEN, 1980)
=
(16)
(17)
&an4sio4,
where Ki is the equilibrium constant of the ith reaction;
PC,, is the partial pressure of CO*, which is assumed to be
constant in the whole system.
Continuity Equations for Water
The one dimensional system to be modelled is shown in
Fig. 3. The coordinate is located at the Earth surface, with
positive direction downward. Water movement in the profile
is governed by Darcy’s law (RICHTER, 1987):
(18)
a(@)
,=-&[bK($-
l)],
(19)
where Q is water flux (cm/s); * is hydraulic head (cm); 4
is porosity; 0 is the relative saturation degree, defined as (0
- a,)/( 1 - O,), where 0, is residual moisture: K( 0) is the
effective permeability (cm/second), which is a function of
0: Z is the distance from the Earth surface (cm); t is time
dry season
wet
season
infiltration
----
I
water
----a
..
FIG. 3. Model system for calcrete formation. The origin of the
coordinate pointing downward is located at the surface. The bottom
of the system is set at the depth, 15, where a water table is located.
~=-.-t(,&!!&%$!d,
(25)
where pw denotes the density of water ( 10m3kg/cm3); Mj is
the molality of the subscripted chemical component; D denotes dispersion coefficient (cm2/s), which is taken to be
constant for all aqueous species in our model;
MZSi
M,c, =
MHcor
+
=
~Mco:-
Wi4sIo, +
+
-
MOH-
+
Mti&Oi
(26)
mH$iOi
2MM,2- - 2Mc,1+ - MH+.
(27)
MA takes account of the effect of the other ions, which are
not explicitly considered in the model, in maintaining electroneutrality of solution ( HARDIEand EUGSTER, 1970). The
first terms on the right sides of Eqns. 22-25 represent the
dispersion effect, and the second terms characterize the solute
transport by water advection due to infiltration or evaporation.
Alternating Boundary Conditions
One of the key elements in the mode1 is that the seasonal
alternation of dry-wet climate changes in semi-arid regions
is explicitly incorporated as upper boundary (Z = 0) conditions. The lower boundary (Z = L) is set at water table.
The boundary conditions are imposed as follows:
at Z = 0:
in dry seasons:
@QD $!$
+ QeMj = 0
Q=-Qe
(28)
(j = Ca’+, Mg*+, ZSi, A);
(29)
5138
Y. Wang, D. Nahon, and E. Merino
in wet seasons:
(30)
Q = Qinr
(,j = Ca’+. Mg”,
.M, = M,”
ZSi, A);
(31)
at Z = L:
in dry seasons:
*=o
M, = M,f
(32)
(j = Ca’+, Mg’+. 2%. A);
*=o
az
(j = Ca”,
DIMENSIONLESS
PARAMETERS
DUAL TIMESCALES
(33)
in wet seasons:
aM,_
-0
Because they are a set of steady-state equations, Eqns. 3943 do not require initial conditions. We should, however, be
aware that + and M, in those equations still vary with time
because the boundary conditions and mineral volume fractions change with time. So Eqns. 39-43 are not steady-state
equations in a strict sense. The approximation
employed is
called quasistationary
state approximation
( LICHTNER,
1988).
(34)
Mg*+, ZSi, A),
(35)
The parameters in the model can be grouped into a least
number of independent
and dimensionless
parameters. To
achieve this. first insert the following scaled dimensionless
variables into Eqns. 9- 12 and 18-25:
where Qe is evaporation rate in dry seasons; Q,,fis infiltration
rate in wet seasons; My represents rainwater composition:
IV:_ represents groundwater composition.
MINERAL
VOLUME
CHANGES
Let 1; (i = I-4) denote the volume fractions of quartz,
calcite, sepiolite, and amorphous
silica, respectively. The
changes in mineral volume are described by
q=K(l)
at
Vz = 0,
t = 0: V, = Vy.
vj = 0,
Vd = 0,
(37)
where V ‘f is the volume fraction of quartz in fresh quartzite.
Rock porosity 4 is related to mineral volume fractions by
4 + SV, = I.
QUASI-STATIONARY
STATE
(38)
APPROXIMATION
Assuming that transient periods of the system are much
shorter than season duration (i.e., assuming that Eqns. 19
and 22-25 approach steady state much faster than seasonal
climate changes), we can drop off all terms on the left sides
of those equations, and we then obtain
-
a(QMcd+) + 3
az
PW
= ()
1
-
a(QMW+)
az
+
R, + 1.5R3 + R4 zz
0,
PW
+ & = o
Pw
,
(40)
(41)
(42)
and
-&@&+~=O.
(43)
Q
y
I
-CR
(44)
’
P~MD
and then group factors so as to end up with as few coefficients
as possible in each equation. A? is the typical concentration
of aqueous species. The resulting independent dimensionless
parameters are
*4_
(36)
where v,” is molar volume of mineral i. The initial conditions
for Eqn. 36 are
AND
k’(l)L:
D
k’
=
’
L2
-k,;
h’ = hL.
P~MD
(45)
where 24is Peclet number, characterizing the relative importance of water advection to hydrodynamic
dispersion; and
ki is the rate constant of chemical reaction scaled relative to
the dispersion coefficient.
According to the proposed mechanism,
the replacement
of a quartzite by a calcrete is limited by the dissolution of
quartz. Thus, we can estimate, from the kinetic data of quartz
dissolution, the rate of the calcrete formation. Let us first
scale Eqn. 36 for quartz into
7=-
;
(46)
with
(47)
where T is a typical time scale for calcrete formation; it is
only on this scale that significant changes in mineral-volume
fractions can take place.
From DOVE and CRERAR ( 1990, Fig. 8 ), the rate constant
of quartz dissolution in 25°C electrolyte solutions is 1O-”
mol/m2 s. By choosing a specific surface area of 60 cm2/cm3
(which corresponds to grain sizes of 1 mm), k, is 6 X lo-“’
mol/cm3 s. Since molar volume ofquartz, v,, is 22.45 cm3/
mol, the timescale of calcrete formation is estimated from
Eqn. 47 to be 0.75 X lOI s, i.e., 24,000 years. Also from
DOVE and CRERAR ( 1990), the rate constant of quartz dissolution in pure water at 25°C is 10-‘2.4 mol/m2 s, and for
the same specific surface as before, k, is 2.4 X lo-l5 mol/
cm3 s, which again with Eqn. 47 yields a timescale of 560,000
years. According to the proposed mechanism, quartz dissolution takes place in dry seasons, when cations are concen-
Calcrete replacement of silicate rocks
trated in porewater by evaporation. Therefore, we think that
the actual timescale of calcrete formation falls between the
two estimates and probably is closer to the shorter one.
The estimated timescale seems comparable with measured
timescales for various calcretes, including calcic soils: HAY
and REEDER ( 1978) found that the Olduvai calcrete in Tanzania formed rapidly, perhaps over several thousand years.
HAWLEY et al. ( 1976) have claimed that at least 100,000
years are required to form the calcrete of the Jornada surface
in New Mexico, USA. CARLISLE(1983) inferred that valley
calcretes formation began less than 0.5 myr ago, perhaps
during a major arid phase starting 25,000 myr ago. The UTh dating carried out by SZABO et al. ( 198 1) shows that the
age of calcretes in southern Nevada, USA, ranges from 5,000
to 345,000 years.
In contrast to the timescale of mineral-volume changes,
the concentrations of aqueous species (Mj), water flow ( Q),
and reaction rates (Ri) vary seasonally, on a timescale of
months or even less. That is, calcrete formation inherently
involves dual timescales. The long-term mineral-volume
changes are directed controlled by the short-term variation
of climates. This dual timescale brings about a difficulty for
numerical modeling, because it would be very time consuming to simulate a geochemical process of IO5 years with a
time resolution less than a month. To overcome this difficulty,
we design an averaging technique to solve the model on the
two timescales but still in a modest computing time.
NUMERICAL SIMULATION
The model is solved by a finite difference method and
Newton-Raphson iteration ( HUYAKORNand PINDER, 1983 ) .
The model is solved on two different timescales by using the
following strategy: Let A7 denote the time increment to solve
for mineral volume changes on the large timescale; thus AT
can be 100 years. And let AZ represent the time increment
to solve for the seasonal variation of the system; At must be
less than a month, say, a half month. The long-term variation
of mineral volume is calculated by
V’~(T + AT) = Vi(T) + AT
C
,
z
At.
(48)
Equation 48 is based on the assumption that annual mineralvolume changes vary smoothly over the large timescale 7.
In the numerical simulations presented here, as a first order
approximation, activity coefficients are taken to be 1. Numerical simulations show that calcretes can form near the
Earth’s surface well above the water table (Fig. 4a,b,c). Figure
4a,b show that sepiolite accumulates below a calcite-rich horizon,” the upper part of which is almost exclusively pure
calcite. At the bottom of the quartz dissolution zone, quartz
is replaced only by sepiolite. Higher in the profile, both calcite
9The occurrence of a very thin layer of sepiolite at the top of the
simulated profilesis probably caused by the oversimplifiedtreatment
of water evaporation at the upper boundary. In the model, all evaporation takes place at the Earth surface,though the actual evaporation
front may be located somewherebelow it. A more rigorous treatment
of the evaporation process can be found in SURBAJI( 1994).
5139
and sepiolite accumulate. Sepiolite grows at expense of quartz
through the chemical divide mechanism in the dry season
(Fig. 4d,e); some of the precipitated sepiolite in turn dissolves
in the following wet season (Fig. 4f,g), making room for
calcite accumulation. Thus, this zone would display a texture
of parent rocks replaced by sepiolite, which is in turn partly
replaced by calcite. Near the top of the calcrete, because there
is no net sepiolite accumulated over each climate cycle, parent
rocks appear to be “directly” replaced by calcite: the sepiolite,
though all gone, still has played a key role in removing silica.
Furthermore, Fig. 4a,b show that the porosity in the calcrete
reaches its minimum value over most of the profile. We set
this minimum value at 0.15 to avoid unrealistic zero porosity.
This implies that in these intervals of the profile the total
(tiny) volume of quartz dissolved can be effectively refilled
with calcite and sepiolite over each climate cycle, leaving the
original textures and structures preserved. That is, the replacement is pseudomorphic. Also, the porosity decrease may
contribute to calcrete induration. All these predicted details
just nicely coincide with the observations summarized above.
In addition, the model suggest that a high porosity zone could
be developed just under the pure calcite zone.
Figures 4 and 5 show that the spatial sequence of calcite
and sepiolite accumulation is insensitive to the reaction-rateconstant ratio of sepiolite to calcite, provided that the rate
constants (k$ and k;) of the two minerals are high enough
relative to that (k’, ) of quartz. Experimental data do show
that the reaction rates for both calcite and sepiolite are higher
than that for quartz (DOVE and CRERAR, 1990; RIMSTIDT
and BARNES,1980; MORSE, 1983; INSKEEP~~~BLOOM,1985;
BRADY, 1992). This insensitivity results from the asymmetry
of reaction rate laws for dissolution vs. growth and from the
proposed chemical reaction path. The reaction rate of a mineral, R, is roughly related to the saturation index, 9, by R
= k( 1 - Q), where k is a rate constant (R > 0 means dissolution and R < 0 means precipitation). R is not a symmetric
function of 0 around fi = 1. As fi approaches 0, R reaches
its maximum value, k. That is, there is a upper limit on
mineral dissolution rate. In contrast, if mass transport is fast
enough, there is no upper limit on the precipitation rate. As
a result, the amount of calcite that can be dissolved in a wet
season is limited, but calcite can in principle precipitate in
any amount during a dry season, if the groundwater contains
enough Ca and evaporation is strong enough. It is this asymmetry that causes calcite to accumulate in most of the profile
over each climate cycle. This is, however, not the case for
sepiolite. If the concentration ratio of silica to Mg is low in
groundwater, most of the silica needed for sepiolite precipitation is supplied by quartz dissolution, and thus sepiolite
precipitation in a dry season is limited by slow dissolution
of quartz. As a result, near the top of the profile, sepiolite
precipitated in a dry season would be effectively removed by
its fast dissolution in the following wet season. This is why
sepiolite accumulates below the calcite-rich zone.
The chemical divide effect ( HARDIE and EUGSTER, 1970)
between Mg and SiOz can be seen clearly in Fig. 4d. It is
caused by sepiolite precipitation and driven by evaporation.
It requires the mole concentration ratio of Mg to Si02 in the
groundwater water to be greater than 2/3 in the case simulated here. Otherwise, the evaporation cannot deplete silica
5140
Y. Wang. D. Nahon,
volume
(4
0_
0.2
0
0.4
and E. Merino
fraction
0.6
W
0.8
1
volume
0
0
0.2
0.4
fraction
0.6
0.8
0.2
5
i
i
i
i
i
i porosity
i
i
i
i
a
0.8
1
%
1
g 0.6
quartz
I
I
0.4
i
i
i
quartz
0.6
0.8
I
1
I/
degree
(d)
0.2
I
Iporosity
0.8
water saturation
@I
0.4
8
u
1
concentration
(1 OM4
M)
1.2
0.2
r
‘;
0”
-0
2
E
In
0.8
0.4
0.6
0.8
1
8
PH
FIG. 4. Simulated profile ofa calcrete replacing quartzite. (a) Mineral volume fractions and porosity vs. scaled depth
at 10’ years after the calcrete started to form. Quartz has dissolved in the upper half of the profile: both sepiolite and
calcite have accumulated.
Sepiolite accumulates below the calcite-rich zone. (b) Mineral volume fractions and porosity
versus scaled depth at 2 X IO5 years. (c) Saturation degree 0 (moisture) profile at 2 X IO5 years. Water table is located
at the bottom of the profile. (d) Concentration
profile of aqueous species at the end of the dry season at 2 X 10’ years.
Because of the chemical divide imposed by growth of sepiolite, Mg2+ concentration
increases while Siq (aq) concentration
decreases toward the top of the profile. It is this SiO, depletion that forces quartz to dissolve. (e) Mineral dissolution
5141
Calcrete replacement of silicate rocks
scaled dissolution
(e)
rate
(f)
- 5
-10
O
0
I#
F
L
(1 0m4 M)
I...,___
.111.1......
----.
concentration
10
5
‘.
-.
calcite..
i
I
i
0.2
I
:
0.2
\
I
i
:
2+
@I
i
i:
Ca2+
1
iP”
:
:.
1
:
0.8
0.8
~--~-.-.-.........__.__~~,
___....1.1.1.111--.----~
1
I
L
1
:
:
m..
I
:
i
..I
1
I
7
9
8
PH
scaled dissolution
(cl1
0
,
0
/
I 20/ I,’
.
rate
80
I 40,
I #60,
2
“...=--_“.......--sepiolite
c
,
0.2
c 0.4
z
0”
u
u
z 0.6
fn
0.8
1
rates vs. depth at the end of the dry season at 2 X lo5 years; negative values represent growth rates. (f) Concentration
profile of aqueous species at the beginning of the wet season at 2 X IO’ years. (g) Mineral dissolution rates vs. depth
at the beginning of the wet season at 2 X lo5 years. Parameter values in this simulation are chosen as follows: dry
season duration = 0.8 years, A? = 10m4molality, m&++ = 3, m&++ = 3, WZ&~~,= 1, rni = -1.5, me.++ = 0, m&,++
= 0, m&o, = 0, m”, = 0, PCQ = lO-35 atm, u = 10, qe = 0.01, qinf= 0.01, V$,,, = 0.85, kbuaru= 5, k&,, = 30,
= 5, h’ = 2, n = 2.
kL,,,,,,c = 80, k:,,,owhhausai,iea
10
5142
Y. Wang, D. Nahon,
volume
0
0.2
0.4
fraction
0.6
0.8
1
7
0’
0’
I’
i
;
i
i
0.8
1
and E. Merino
Increasing wet season duration or infiltration rate shifts
sepiolite accumulation
downward and significantly reduces
the thickness of the calcite zone at the top of the profile (Figs.
7, 8). If wet season duration is long enough and infiltration
rate is high, rainwater can leach away any calcite accumulated
in dry seasons, and then calcretes cannot form. On the other
hand, if wet season duration is too short and infiltration rate
is too low, there is not enough rainwater to dissolve sepiolite
(or attapulgite) and to flush away the dissolved components
effectively, and then little space is created for calcite precipitation by removing parent rocks. Therefore, under certain
climate conditions, the replacement of silicate rocks by calcretes would proceed most effectively.
Figures 8 and 9 demonstrate
that an increase in Pro, in
calcrete profiles dramatically reduces both calcite and sepiolite
accumulation.
This is consistent with the idea suggested by
SALOMONS and MOOK ( 1986) that the degassing of CO2 is
another factor causing calcite precipitation.
iporosity
I
i
i
i
CONCLUSIONS
1-L.
Rc;. 5. Calcrete profile at 2 X IO’ years. All parameter values used
here are the same as those in Fig. 4 except for: k:,,,,,, = 30. kk,,,,,,,
= 20. Compared with Fig. 4, this tigure shows that the spatial sequence
of sepiolite accumulated
below the calcite-rich zone is insensitive to
the relative reaction rate of sepiolite to calcite, provided that the
reaction rates of both minerals are high enough relative to that of
quartz.
in water, and quartz can not dissolve in dry seasons. Calcite
then grows mainly in the original pore space, and no replacement takes place, as shown in Fig. 6. The requirement
of a
high concentration
of Mg relative to Si in groundwater
is
consistent with reported water compositions
in calcrete regions ( ARAKEL, 1986, Table I; GOUDIE, 197 1, Table 5) and
the occurrence of dolomite and Mg-rich calcite in calcretes
(ARAKEL, 1986; WATTS, 1980; HAY and REEDER, 1978).
Some textures indicate that the low-Mg calcite in calcretes
may be transformed from original high-Mg calcite (WATTS,
1980). ARAKEL ( 1986) reports for the water from the calcrete
area in central Australia that SiOl (aq ) becomes depleted as
water salinity increases. Also, MANN and DEUTSCHER( 1978)
report concurrent Mg increase and Si decrease with Cl in a
calcrete aquifer in Western Australia. In this aquifer the water
is supersaturated
with sepiolite but undersaturated
with
amorphous silica. All this evidence supports the idea that the
chemical-divide
mechanism operates in calcrete formation.
Furthermore,
Figs. 4 and 6 demonstrate the effect of Mg/
Si ratio in groundwater on the intensity of calcretization.
If
the upstream groundwater contains high enough Mg, water
evaporation causes Mg and Mg/Si in groundwater to become
higher down stream, and therefore increases the intensity of
calcretization. Thus, our model predicts that calcretes would
increase in their thickness and Mg content downslope. The
prediction agrees with observations ( NAHON, 1976; CARLISLE,
1983; ARAKEL, 1986; Fig. 7c; WRIGHT and TUCKER, 199 I,
Fig. 6b).
In the model herein the genesis of replacement is directly
related to seasonally alternating dry-wet climates and to appropriate groundwater
(or circulating soil water) compositions. In a dry season, water evaporation causes CaCOi and
sepiolite (or attapulgite) to precipitate. If groundwater contains enough Mg2+. the sepiolite precipitation by the chemical-divide mechanism
depletes Si02(aq),
resulting in the
dissolution of parent silicates. In the following wet season.
sepiolite dissolves fast, silica and cations are flushed away by
volume
fraction
1
1
yy+J?$k5
alcite
a;
I
i
i
i
i
i
i
quart z
i porosity
i
i
0.8
1
_
i
i
i
i
i.
i ,
_.A
I
L
FK. 6. Mineral-volume
fractions vs. scaled depth ; at 2 x IO’
years. All parameter values used here are that same as those in Fig,
4 except form&++ = I. mu,,,,, = 2, rn: = -0.5.
Because Mg/Si in
initial groundwater is less than 2/3, sepiolite can not replace quartzite.
and thus calcite precipitates mainly in original pores.
Calcrete replacement of silicate rocks
volume
fraction
0
r,
/
._.-.----
0.2
I
0.4
e
u
u
3.l
3 0.6
c
i
i
i
quart: L
i porosity
i
i
I
0.8
1
.
w
I
I
4
FIG. 7. Calcrete profiles at 2 x IO’ years. All parameter values
used here are the same as those in Fig. 4 except for qinr = 0.03 and
dry season duration = 0.73 year. Compared with Fig. 8, this figure
shows that increasing wet season duration or infiltration rate shifts
sepiolite accumulation downward and significantly reduces the
thickness of the calcite zone at the top of the profile.
in the next
dry season. As climate cycles repeat, CaC03 is accumulated
and silicates are removed. The sepiolite (or attapulgite, or
Mg-smectite) serves as a temporary storage of silica between
seasons. If the groundwater contains too little aqueous Mg
then the model predicts growth of calcium carbonate without
removing silicates, thus producing void filling and or displacive textures instead of replacement.
We simulate the replacement of quartzite by calcrete with
a transport-reaction model, which consists of nonlinear partial
differential equations taking account of mass conservation,
dispersion, advection, evaporation, and the kinetics of mineral
reactions. The hydrodynamics of unsaturated medium is applied in determining water flow in calcrete profiles. Seasonal
climate changes are explicitly incorporated as boundary conditions. The model is solved numerically on two different
timescales, one of I month, the other 1000 years, both of
which are essential for the overall phenomenon. The model
successfully produces the mineral and textural zonation observed in many calcretes (namely, at the bottom of the profile
the parent rock is first replaced by sepiolite, only part of which
is in turn replaced by calcite, whereas at the top of the calcrete
the sepiolite is itself completely replaced by the calcite, which
appears to “directly” replace the parent rock). The model
can produce calcrete near the Earth’s surface well above the
water table. Low Pco,, intensive evaporation, and long dry
seasons all are predicted to produce thicker calcretes.
Calcretes constitute an effective geochemical tool that, via
replacement, removes parent silicate rocks and shapes the
rainwater,
making room for CaC03 precipitation
5143
landscape of semi-arid countries. Our model provides a
mechanism that accounts for the high efficiency of replacement in removing silicates. This high efficiency is attributed
to two factors. First, quartz (the most resistant silicate)
dissolution in dry seasons is accelerated by high cation concentrations in porewater. Second, in dry seasons, by the
chemical-divide mechanism, water evaporation (through the
precipitation of attapulgite or sepiolite) makes porewater very
undersaturated with silicates, thus also accelerating silicate
dissolutions.
In the numerical simulations shown here, silica concentrations are high enough for chacedony or opal-CT to form
(Fig. 4f). This coincides with the occurrence of minor chalcedony in calcretes (GOUDIE, 197 I; NAHON, 1976; ARAKEL,
1986). Because the precipitated silica directly comes from
the dissolution of sepiolite, the predicted rock texture would
show that sepiolite is replaced by the precipitated silica. This
texture is indeed observed in the calcretes of central Australia
( ARAKEL, 1986) and in the Argus Range of California, USA
(HAY and WIGGINS, 1980). Furthermore, also because this
precipitated silica is derived from Mg-phyllosilicates, they
should contain traces of Mg, as was reported by DUCLOUX
and LAOUINA( 1988). Silicification is often associated with
calcretization in semi-arid regions ( GOUDIE, I97 I; WATTS,
1980). The proposed model thus gives a genetic explanation
for the association of sepiolite-palygorskite with silica (e.g.,
JONES and GALAN, 1988). In addition, because the dissolved
silica is carried downward or downslope in the rainy seasons,
silicification is thus most likely to occur at the bottom of the
profile ( REEVES, 1970; WATTS, 1980) or far down-dip (ARAKEL, 1986). As calcretization proceeds further, this authi-
genie silica may also be removed.
0
0.2
r,
0.4
8
u
-0
2
i 0.6
0.8
1
FIG. 8. Calcrete profile at 2 X 10’ years with longer dry season
duration ( = 0.8 year) and lower infiltration rate (qinf= 0.01). All
other parameter values are as in Fig. 7.
5144
Y. Wang, D. Nahon.
volume
and E. Merino
reviewer for their constructive
fraction
suggestions, and Quade and Phillips
for unpublished manuscripts,
We also thank R. Dassulle, J.-J. Motte,
J.-P. Ambrosi, and Marc Benedetti, all of Marseille, and Lorie Canada
of Bloomington,
for their help.
1
e
I
i
i
iporosity
.
/
j
0.8
1
I
i
i
l
I
/
/
I
I
-1
FIG. 9. Calcrete profile at 2 X IO5 years with higher Pco, ( = IO-”
atm) than that in Fig. 8. All other parameter values are as in Fig. 8.
It is shown here that an increase in P co2 in calcrete profiles dramatically reduces both calcite and sepiolite accumulation.
REFERENCES
ARAK~L. A. V. ( 1986)Evolution of calcrete in palaeodrainages
of
the Lake Napperby area. central Australia. Puluro~,qwgr. Pdueodim Pulueoc~~~l. 54, 283-303.
ARISI-ARAIN L. F. ( 197 I ) Clay minerals in caliche deposits ofeastern
New Mexico. J. C&o/. 79, 75-90.
BECI-I J., NAH~N D.. PAQU~:T H.. RUELLAN A., and MIL.LOT G.
( 1980) Sur l’extension geographique et climatique des phenomtnes
d’tpigenie par la calcite dam les encroutements
calcaires. Exemple
de la Catalogne. C. R. .-ld.
Scr furi.s 291, 371-376.
BOULET R. ( 1974) Toposequences
de sols tropicaux en Haute Volta:
Equilibres dynamiques
et bioclimats. These Sciences Naturelles.
Universitt de Strasbourg.
BRADY P. V. ( 1992) Surface complcxation
and mineral growth: Sepiolite. In Mb/w-Rod
Inkruction
(ed. Y. K. KHARAKA and
A. S. MAESI-). pp. 85-8X. Balkema.
CAKI.ISE D. ( 1983) Concentration
of uranium and vanadium in calcretes and gypscretes. In Rcsidrrul Dcyosi/s Gd
SK (ed. R. C.
WILSON). pp. 185-195.
CtiADWICK0. A. and Nr:r I i.t IONW. D. ( 1990) Micromorphological
evidence of adhesive and cohesive forces in soil cementation.
In
Soil Mlc,romorpho(o.~~,: .I Rusk uncl :lpplicd Scicwe Devc~lopmcvt/nl.c
in Sod S~icwx~ No. 19 (ed. L. A. DOIJC~LAS),pp. 207-2 12. Elsevier.
CHAPMAN R. W. ( 1974) Calcareous duricrust in Al-Hasa. Saudi
Arabia. Bll//. C;CJO/.S’oc~..Inwr. 85. I l9- 130.
DOVF. P. M. and CRERAR D. A. ( 1990) Kinetics ofquartz dissolution
in electrolyte solutions using a hydrothermal
mixed flow reactor.
Gcwhim.
are expected
to precipitate in dry seasons, because Mg2+ is concentrated
by evaporation. In the rainy seasons, porewater
becomes
diluted, and the high-Mg carbonate may transform into lowMg calcite. as some textures indicate (WATTS. 1980). This
transformation
should take place more completely at the top
of a calcrete than at the base, because Mg*’ concentration
in rainy seasons is increased downward by the dissolution of
Mg-phyllosilicates.
Therefore, the amount of high-Mg carbonates such as dolomite may increase toward the base of a
calcrete, just as observed by PHILLIPS and MILNES ( 1988).
Figure 2.4 indicates that the reaction front of calcretization
is rather sharp. This sharp front may be caused by the reduction of the mass-transport
rate in the water unsaturated
zone, where dispersion can be significantly reduced as water
saturation degree decreases, so that the whole chemical process may become transport-limited,
and the thickness of the
reaction front approaches 0 ( ORTOLEVA et al., 1987 ).
The results presented here demonstrate the importance of
short-term climate variations to long-term Earth surface processes. Actual climates may be of multiple timescales ranging
from as little as hours to thousand years. We think that realistic geochemical
models for the Earth surface processes
should take into account this inherent timescale variability
of climates. The approach we employ here may have an implication to geochemical modeling of other weathering processes.
According
to the model, high-Mg carbonates
A~,l;no,l,/~~~rnPnls-We
are grateful to the U.S. NSF (grants EAR
9003633 and EAR 9104198)
and the CNRS of France (grant
MRT90L0712)
for partial support of this research. We thank Fred
M. Phillips as associate editor. J. Quade. A. White. and an anonymous
(h.smochitn
.k/u
54, 955-969.
DUCLO~JXJ. and LAOUIIUA A. ( 1988) La glaebulation
calcaire: un
stade de I’alteration des granites et de I’encroltement
carbonate
des sols en climat semi-aride: exemple pris dans le Maroc nordoriental. C’ R. .Acad. Sci. Puri.s 307, 1567-1572.
D~CUXJX J.. D~PUIS T.. B~ITEI. P., and NAHON D. ( 1984) Carbonates de calcium amorphe et cristallises dans les encrodtements
calcaires des milieux ternperes. Comparison
des sequences min&ales naturelles et experimentales.
C. R. .kud. SC. Puris 298(4).
147-149.
D~CLO~JX J., LAOIJINA A.. CHAK~K M.. and DINEI H. ( 1990) Carbonate accumulation
and weathering of the granitic substratum
of Tanecherfi Basin. Northeastern
Morocco. Curcnu 17.493-508.
DLJPUIS T., DU~LOUX J., BU~EL P.. and NAHON D. (1984) Etude
par spectrographic
infrarouge d’un encroutement
calcaire sous
galet. Mise en evidence et modtlisation
experimentale
d’une suite
minerale Cvolutive a partir de carbonate de calcium amorphe. C’lcr)~
hfincd
19, 605-6 14.
DLIRANI) R. ( 1979) La pcdogenese en pays calcaire dans le NordEst de la France. M&r?. .R~r (;C;ol. 55, I - 198.
GILE L. H. ( 196 I ) A classification of Ca horizons in soils of a desert
region. Doiia Ana County, NM. Soil S?i. .Arwr. Proc 25, 52-6 I
GII.E L. H.. PETERSON F. F., and GROSSMAN R. B. (1966) Morphology and genetic sequences of carbonate accumulation
in desert
soils. Soil Sci. 101, 347-360.
GO~J~IE A. ( I07 I ) The chemistry ofworld calcrete deposits. ./. (;&.
80,449-463.
GO~J~IE A. ( 1983) Calcrete. In C‘/J~I)I~& Sedmrmrs rind Gcwrnorp/70~0~J’ (ed. A. S. GOtJmr and P. Pvr), pp. 93-l 3 I. Academic
Press.
HARr)IE L. A. and ELrc;Sr t.R [I. P. ( 1970) The evolution of closedbasin brines. Mincrul. .%Ic. .~lnwr. Spc Pupw No. 3. 373-390.
HAWLEY J. H.. BACHMAN G. O., and MANLEY K. (1976) Quatemary
stratigraphy
in the basin and Range and Great Plains provinces,
New Mexico and western Texas. In @aternary
Strufigruph~~ cd
North Americu (ed. W. C. MAHANEY ). pp. 235-274.
Dowden.
Hutchinson.
and Rose.
Calcrete replacement of silicate rocks
HAY
R. L. and REEDERR. J. ( 1978) Calcretes of Olduvai Gorge and
the Ndolanya Beds of Northern Tanzania. Sedimentology 25,649673.
HAY
R. L. and WIGGINSB. ( 1980) Pellets, ooids, sepiolite and silica
in three calcretes ofthe south-western United States. Sedimentology
27,559-576.
HUBERT J. F. (1978) Paleosol caliche in the New Haven Arkose,
Newark Group, Connecticut. Palaeogeogr. Palaeoclim. Palaeoecol.
24, 151-168.
HUYAKORNP. S. and PINDERG. F. ( 1983) Computational Methods
in SubsuTface Flow. Princeton Univ. Press.
INSKEEPW. P. and BLCK)MP. R. ( 1985) An evaluation rate equation
for calcite precipitation kinetics at Pco2 less than 0.0 I atm and pH
greater than 8. Geochim. Cosmochim. Acta 49,2 165-2 180.
JONES B. F. and GALAN E. ( 1988) Sepiolite and palygorskite. In
Hydrous Phyllosilicates (exclusive of micas); Rev. Mineral. 19 (ed.
S. W. BAILEY), pp. 63 l-674.
LICHTNERP. C. ( 1988) The quasi-stationary state approximation to
coupled mass transport and fluid-rock interaction in a porous medium. Geochim. Cosmochim. Acta 52, 143-165.
MACHETTEM. N. ( 1985) Calcic soils of the Southwestern United
States. In Soil and Quaternary Geology ofthe Southwestern United
States: Geol. Sot. Amer. Spec. Paper 203 (ed. D. WEIDE), pp. lMALIVAR. G. and SIEVERR. (1988) Diagenetic replacement controlled by force of crystallization. Geology 16, 688-69 I.
MANN A. W. and DEUTSCHERR. L. ( 1978) Hydrogeochemistry of
a calcrete-containing aquifer near Lake Way, Western Australia.
J. Hydrology 38, 357-377.
MARIONG. M., SCHLESINGERW. H., and FONTEYNP. J. (1985)
Caldep: a regional model for soil CaC03 (caliche) deposition in
south western deserts. Soil Sri. 139, 468-48 I.
MAYER L., MCFADDEN L. D., and HARDENJ. W. (1988) Distribution of calcium carbonate in desert soils: a model. Geology 16,
303-306.
MCFADDENL. D. and TINSLEYJ. ( 1985) Rate and depth of pedogenie-carbonate accumulation in soils: Formulation and testing of
a compartment model. In Soil and Quaternary Geology of the
Southwestern UnitedStates; Geol. Sot. Amer. Spec. Paper No. 203
(ed. S. W. WEIDE), pp. 23-42.
MERINO E., NAHON D., and WANG Y. ( 1993) Kinetics and mass
transfer of pseudomorphic replacement: Application to replacement of parent minerals and kaolinite by Al, Fe, and Mn oxides
during weathering. Amer. J. Sci. 293, 135- 155.
MILLOT G.. PAOUETH.. and RUELLANA. ( 1969) Neoformation
d’attapulgite dins les sols et carapaces calcaires de la basse Moulouya (Maroc Oriental). C. R. Acad. Sci. Paris 268, 177 l-2774.
MILLOTG., NAHON D., PAQUETH., RUELLANA., and TARDYY.
( 1977) L’tpigknie calcaire des roches silicatCes dans les encrofitements carbonat& en pays subaride Antiatlas, Maroc. Sci. Gbol.
Bull. 30(3), 129-152.
MILNESA. R. ( 1992) Calcrete. In Weathering Soils and Paleosoils
(ed. 1. P. MARTINI and W. CHESWORTH),pp. 309-347. Elsevier.
MONGERH. C. and DAUGHERTYL. A. ( 1991a) Pressure solution:
possible mechanism for silicate grain dissolution in a petrocalcic
horizon. Soil Sci. Amer. J. 55, 1625- 1629.
MONGERH. C. and DAUGHERTYL. A. ( 199lb) Neoformation of
palygorskite in a southern New Mexico aridisol. Soil Sci. Sot. Amer.
J. 55, 1646-1650.
MORSEJ. W. ( 1983) The kinetics of calcium carbonate dissolution
and precipitation. In Carbonates: Mineralogy and Chemistry; Rev.
Mineral. 11 (ed. R. J. REEDER), pp. 227-264.
MORSEJ. W. and CASEYW. ( 1988) Ostwald processes and mineral
paragenesis in sediments. Amer. J. Sci. 288, 537-560.
NAGTEGAALP. J. C. ( 1969) Microtextures in recent and fossil caliches. Leid. Geol. Meded. 42, 13 1- 142.
NAHON D. ( 1976) Cuirasses ferrugineuses et encrofitements calcaires
5145
au StnCgal occidental et en Mauritanie. Systtmes Cvolutifs: Gtochimie, structures, relais et coexistence. Sci. Giol., Mbm. 44.
NAHON D. ( 199I ) Introduction to the Petrology of Soils and Chemical
Weathering. Wiley.
NAHON D. and RUELLANA. (1972) EncroQtements calcaires et
cuirasses ferrugineuses dans I’ouest du S&&al et de la Mauritanie.
C. R. Acad. Sci., Paris 274, 509-512.
NAHON D., PAQUETH., RUELLANA., and MILLOTG. (1975) Encrofitements calcaires dans les alttrations des marnes &c&es de
la falaise de Thi& (SC&gal): Organisation et morphologie. Sci.
G&ol. Bull. 28, 29-46.
NAHON D., DUCLOUXJ., BUTEL P., AUGAS C., and PAQUETH.
( 1980) Ntoformations d’aragonite, premitre &taped’une suite mintralogique Cvolutive dans les encrotitements calcaires. C. R. Acad.
Sci. Paris 291(9),
725-727.
ORTOLEVAP., CHADAMJ., MERINO E., and SEN A. ( 1987) Geochemical self-organization, 11:The reactive-infiltration instability.
Amer. J. Sci. 287, 1008-1040.
PHILLIPSS. E. and MILNESA. R. ( 1988) The Pleistocene terrestrial
carbonate mantle on the southeastern margin of the St Vincent
Basin, South Australia. Australian J. Earth Sci. 35,463-48 I.
QUADEJ., CHIVASA. R., and MCCULLOCHM. T. ( 1994) Strontium
and carbon isotope tracers and the origins of soil carbonate in
South Australia and Victoria. Palaeogeogr. Palaeoclim. Palaeocol.
(in press).
REEVESC. C. ( 1970) Origin, classification and geological history of
caliche on the southern High Plains, Texas and Eastern New Mexico. J. Geol. 78, 352-362.
REEVESC. C. ( 1983) Pliocene channel calcrete and suspenparallel
drainage in West Texas and New Mexico. In Residual Deporitr,
Geol. ioc. London Spec. Publ. No. 1 I (ed. R. C. L. WILSON),-;;:
179-183.
REHEISM. C. ( 1988) Pedogenic replacement of aluminosilicate grains
by CaC03 in Ustollic Haplargids, South-Central Montana, U.S.A.
Geoderma 41,243-26 I.
RICHTERJ. ( 1987) The Soil as a Reactor: Modelling Processes in
the Soil. Catena Verlag.
R~MSTIDT
J. D. and BARNESH. L. ( 1980) The kinetics ofsilica-water
reactions. Geochim. Cosmochim. Acta 44, 1683- 1699.
RUELLANA. ( I97 I ) Les sols B profil calcaire differencik des plaines
de la Basse Moulouya, Maroc Oriental. M&m. Og Rech. Sci. Tech.
Outre-Mer, (ORSTOM)
54.
SALOMONSW. and MOOK W. G. (1986) Isotope geochemistry of
carbonates in the weathering zone. In Handbook ofEnvironmental
Isotope Geochemistry, Vol. 2 (ed. P. FRITZ and J. CH. FONTES),
pp. 239-269. Elsevier.
SURBAJIABDEL-RAHMANM. (1994) The Contribution of H2180
‘HHO, and Cl- to the Modeling of Water Movement in the Unsaturated Zone. Ph.D dissertation, New Mexico Institute of Mining
and Technology.
SZABOB. J., CARRW. J., and GOTTSCHALL
W. C. ( 1981) UraniumThorium dating of Quaternary carbonate accumulations in the
Nevada test site region, Southern Nevada. USGS Open-File Report
81-119.
GENUCHTENM. TH ( 1980) A closed-form equation for predicting the hydraulic conductivity of unsaturated soils. Soil Sci.
VAN
Sot. Amer. J. 44, 892-898.
WATTS N. L. ( 1978) Displacive calcite: evidence from recent and
ancient calcretes. Geology 6, 699-703.
WA?TS N. L. ( 1980) Quaternary pedogenic calcretes from Kalahari
(southern Africa): Mineralogy, genesis and diagenesis. Sedimentolgy 27, 66 I-686.
WIEDERM. ( 1977) Occurrence and genesis of carbonate nodules in
soil. Ph.D. thesis, Hebrew Univ., Jerusalem.
WEEDERM. and YAALOND. H. ( 1974) Effect of matrix composition
on carbonate nodule crystallization. Geoderma 11, 95- I2 1.
WRIGHTV. P. and TUCKERM. E. ( 199 I ) Calcretes: an introduction.
In Calcretes (ed. V. P. WRIGHT and M. E. TUCKER), pp. l-22.