A reactive-transport model for weathering rind formation on basalt

Abstract

Saprolite formation rates influence many important geological and environmental issues ranging from agricultural productivity to landscape evolution. Here we investigate the chemical and physical transformations that occur during weathering by studying small-scale “saprolites” in the form of weathering rinds, which form on rock in soil or saprolite and grow in thickness without physical disturbance with time. We compare detailed observations of weathered basalt clasts from a chronosequence of alluvial terraces in Costa Rica to diffusion-reaction simulations of rind formation using the fully coupled reactive transport model CrunchFlow. The four characteristic features of the weathered basalts which were specifically used as criteria for model comparisons include (1) the mineralogy of weathering products, (2) weathering rind thickness, (3) the coincidence of plagioclase and augite reaction fronts, and (4) the thickness of the zones of mineral reaction, i.e. reaction fronts. Four model scenarios were completed with varying levels of complexity and degrees of success in matching the observations. To fit the model to all four criteria, however, it was necessary to (1) treat diffusivity using a threshold in which it increased once porosity exceeded a critical value of 9%, and (2) treat mineral surface area as a fitting factor. This latter approach was presumably necessary because the mineral-water surface area of the connected (accessible) porosity in the Costa Rica samples is much less than the total porosity (Navarre-Sitchler et al., 2009). The model-fit surface area, here termed reacting surface area, was much smaller than the BET-measured surface area determined for powdered basaltic material. In the parent basalt, reacting surface area and diffusivity are low due to low pore connectivity, and early weathering is therefore transport controlled. However, as pore connectivity increases as a result of weathering, the reacting surface area and diffusivity also increase and weathering becomes controlled by mineral reaction kinetics. The transition point between transport and kinetic control appears to be related to a critical porosity (9%) at which pore connectivity is high enough to allow rapid transport. Based on these simulations, we argue that the rate of weathering front advance is controlled by the rate at which porosity is created in the weathering interface, and that this porosity increases because of mineral dissolution following a rate that is largely surface-reaction controlled.

Introduction

As bedrock weathers in place without disturbance, regolith can form that retains evidence of bedrock structure. This regolith, generally known as saprolite, is more erodible than bedrock. Understanding the rates at which saprolite forms on silicate rocks is necessary to develop basic understanding of processes such as nutrient cycling (Hyman et al., 1998, Chadwick et al., 1999, Derry et al., 2005, Starr and Lindroos, 2006), landform evolution (Dethier, 1986, Pavich, 1986, Pope et al., 1995), and regulation of CO₂ concentration in Earth’s atmosphere over geologic timescales (Walker et al., 1981, Berner et al., 1983).

Basalt is of particular interest with respect to saprolite formation because it is easily weathered and accounts for approximately 30–35% of the atmospheric CO₂ drawdown attributed to silicate weathering (Dessert et al., 2003). A number of researchers have put forth considerable effort to measure basalt weathering rates across watersheds of varying extent (e.g. Gislason et al., 1996, Louvat and Allegre, 1997, Louvat and Allegre, 1998, Dessert et al., 2001, Gaillardet et al., 2003, Das et al., 2005) as well as to quantify weathering rates in laboratory dissolution experiments for basalt glass (e.g. Gislason and Eugster, 1987, Eick et al., 1996, Gislason and Oelkers, 2003, Wolff-Boenisch et al., 2004) and crystalline basalt (Gudbrandson et al., 2008, Hausrath et al., 2009).

To compare rates measured in different systems, weathering rates are typically normalized to mineral surface area. For example, dissolution rates measured in the laboratory are generally normalized to the BET surface area of dissolving minerals. This is determined by adsorbing an inert gas to mineral powder and then assessing the surface area through the use of the Brunauer–Edwards–Teller, i.e. BET, adsorption isotherm (Brunauer et al., 1938). Laboratory dissolution rates normalized by such BET surface areas are often 2–5 orders of magnitude faster than field-derived weathering rates that have been normalized to estimates of BET surface area for the field samples (White and Brantley, 2003). This so-called field-laboratory discrepancy has been observed for many lithologies or mineralogies (Velbel, 1993, Anbeek et al., 1994, Blum et al., 1994, Gislason et al., 1996, Drever and Stillings, 1997, White and Brantley, 2003, Beig and Luttge, 2006, Maher et al., 2006). Part of the problem is that BET surface area is difficult to estimate for field systems. Given this difficulty, most field scale weathering rates are generally reported after normalization by geographic instead of mineral surface area. Navarre-Sitchler and Brantley (2007) argued that much of the discrepancy observed when comparing field- and laboratory-derived weathering rates is related to scaling of surface area, but that weathering rates for basalt can also vary by up to two orders of magnitude even after the surface area scaling problem is corrected (Navarre-Sitchler and Brantley, 2007).

This remaining discrepancy has been attributed to biological effects such as the presence or absence of organic acids (Drever and Stillings, 1997), the exposure age or residence time of weathered material (White and Brantley, 2003, Maher et al., 2004), differences in hydrologic regime (Velbel, 1993, Pacheco and Alencoao, 2006), armoring of natural surfaces (Nugent, 1998), or affinity effects that can sometimes be related to secondary precipitates (White and Brantley, 2003, Zhu, 2005, Maher et al., 2006, Maher et al., 2009, Steefel and Maher, 2009). Additionally, large-scale field systems are often impacted by physical erosion that removes weathered material and may influence the interpretation of chemical weathering rates (Bluth and Kump, 1994, Millot et al., 2002). In order to understand weathering of natural systems, it is necessary to examine end members where primary processes that control chemical and physical weathering can be considered separately.

One example of weathering that is relatively uncomplicated by erosion and perhaps biological effects is an alteration rind that has formed on a weathered clast at depth within a soil (Cernohouz and Solc, 1966, Chinn, 1981, Colman and Pierce, 1981, Gellatly, 1984, Oguchi and Matsukura, 1999). Weathering rinds, defined as permeable crusts enriched in immobile oxides that envelop cores of unweathered rock material (Colman, 1982), are often used as indicators of relative ages of sediment deposits in glacial (Porter, 1975, Colman and Pierce, 1981) and alluvial (Ricker et al., 1993, Fisher et al., 1998, Sak et al., 2004) systems. Rinds have even been analyzed on Martian rocks (Hausrath et al., 2008). In effect, weathering rinds can be conceptualized as small-scale “saprolites” that allow detailed observations of the chemical and physical transformations that occur during weathering (Sak et al., 2004, Pelt et al., 2008, Sak et al., 2010).

Here we investigate rinds developed on weathered clasts in terraces of different ages but within the same climate and weathering environment. According to IUGS classification using total alkalis (Na₂O + K₂O weight%) and silica (SiO₂ weight%) reported in Sak et al. (2004) the clasts range in composition between basalt and basaltic andesite (Raymond, 1995). Nonetheless, all of the clasts were referred to as basalts in previous studies (Sak et al., 2004, Hausrath et al., 2008, Navarre-Sitchler et al., 2009) and we will follow the same convention here. The weathered basalt clasts from the Costa Rica chronosequence provide a time series of observations of the geometry of the reaction front and its position over time, i.e. the weathering advance rate. A reaction front is defined as the transition zone across which a mineral reaction occurs. The basalt clasts were deposited in three fill-type alluvial terraces (Qt1, Qt2, and Qt3) along the central Pacific coast of Costa Rica (Fisher et al., 1998, Sak et al., 2004, Fisher and Ryan, 2006) and they weathered in an environment where they were surrounded by terrace material (Sak et al., 2004). Therefore, chemical weathering processes can be studied in these clasts without the additional complications associated with the removal of weathered material by physical erosion.

The terraces are preserved in the lower reaches of the Rios Parrita and Barranca on the Pacific side of Costa Rica (Fig. 1). The basalt clasts are characterized by a central core of relatively unweathered material surrounded by a rind dominated by secondary minerals with a few residual grains of ilmenite and apatite. Rinds increase in thickness as weathering advances into the clasts with terrace age. Rinds on basalt clasts were measured by Sak et al. (2004) to be 12 ± 1, 29 ± 1, and 69 ± 6 mm thick for Qt3, Qt2, and Qt1, respectively. The mean annual temperature and precipitation are 27.3 °C and 3085 mm/y, respectively, for the time period from 1941–1982 (Instituto Meterologico Nacional de Costa Rica, 1992, see Sak et al., 2004). While temperature is relatively constant seasonally, precipitation varies from a low of 28.5 mm in March to a high of 570 mm in October (Sak et al., 2004). Additional detailed descriptions of the terraces are available (Sak, 1999, Marshall, 2000, Sak et al., 2004).

In this study, we collected detailed chemical data across the core/rind interface of five clasts to delineate the weathering advance rate through quantification of the rind thickness and reaction front thickness through time. We present full results of numerical simulation efforts using the multicomponent software CrunchFlow (Giambalvo et al., 2002, Maher et al., 2009) which we used to develop a fully-coupled reactive transport model of weathering rind formation. Model predictions were compared to observations from the basalt clasts. The numerical model successfully predicts weathering advance rates and reaction front thicknesses. Such a model is a first step in providing insight into how to numerically model weathering advance or saprolite formation at larger scales.