Quanzhou Gao　Zhen Tao　Xiakun Huang Ling Nan　Kefu Yu　Zhengang Wang

　　1　Introduction

　　In the chemical weathering of silicate and carbonate rocks, gaseous CO2 was converted into aqueous HCO3- and CO32-, then discharged into the sea where they react to form calcite and dolomite with Ca2+ and Mg2+ and subside as carbonate rocks, which process has effectively weakened the greenhouse effect of the earth system (Walker et al., 1981; Raymo and Ruddiman, 1992; Gaillardet et al., 1999; Galy and France-Lanord, 1999; Liu and Zhao, 2000; Dalai et al., 2002; Picouet et al., 2002; Amiotte-Suchet et al., 2003; Dessert et al., 2003; Raymond and Cole, 2003; Das et al., 2005; Krishnaswami and Singh, 2005). Hence, the chemical weathering of continental rocks constitutes an important carbon sink of the global biogeochemical cycles at different timescales (Berner, 1990; White et al., 1999; Berner and Kothavala, 2001). For carbonate rocks, the weathering process on the continent and the deposition process in the sea build up a close cycle of CO2, which needs a relatively shorter time for its balance. While for silicate rocks, the corresponding timescale of the carbon cycle is much longer than that of carbonate rocks because the weathering-deposition process and related CO2 cycle involves the interior-earth processes (Berner et al., 1983; Gibbs et al., 1999; Wallmann, 2001). The chemical weathering of continental silicate rocks, together with the burial of organic carbon, controls the concentration of atmospheric CO2 at the geological timescale (>106 y) and shapes the nature of the global climate and environment (Raymo, 1994; France-Lanord and Derry, 1997; Gaillardet et al., 1999). Inversely, the nature of the global climate and environment, especially the atmospheric temperature and water cycling rate, determines the chemical weathering rate of continental silicate rocks and its consumption rate of the atmospheric CO2 (Kump et al., 2000). Through the greenhouse effect, the atmospheric CO2 level and chemical weathering rate constitute a particular negative feedback (Walker et al., 1981; Berner et al., 1983), preventing the global climate from evolving toward extremeness at the geological timescale and providing a relatively suitable and stable climatic environment for the emergence and development of life on the earth (Huh, 2003).

　　Rivers integrate the natural and human activities taking place within the drainage basin. Riverine water chemical signatures reflect (to a large extent) the chemical weathering processes of the rocks and sediments within the drainage basin (Dupre et al., 2003; Huh, 2003; Meybeck, 2003). Moreover, a good understanding of upper stream weathering processes will be helpful in explaining the dynamics of the CO2 system in the complicated estuarine zone (Cai et al., 2004; Guo et al., 2008). 

　　Though a number of investigations have been documented on the chemical weathering-atmospheric CO2 consumption in the humid tropic zone (Das et al., 2005; Singh et al., 2006; Moon et al., 2007), a few studies have been published on the major ion chemistry of the Xijiang River (XJR) in south China based on the analysis of the historical hydrological and aqueous chemical records (Chen and He, 1999; Zhang et al., 2007). We seasonally sampled and analyzed the HCO3- concentration and calculated its transporting flux in the years of 1997 and 1998 at two sections of the lower reach of the XJR, however, without analysis of cations and other anions (Gao et al., 2001). No study has yet been devoted to chemical weathering serving as a sink of atmospheric CO2 on the basis of riverine ion chemistry in south China.

　　To investigate the concentration of river-borne substance increasing or decreasing with discharge is very important for precise estimation of the rock-weathering flux within the drainage basin. Here we report the concentrations of major riverine ions (Ca2+, Mg2+, K+, Na+, HCO3-, SO42-, Cl-, NO3-) in the XJR from monthly analysis for 14 consecutive months (March 2005 to April 2006) and from more frequent sampling during a catastrophic flood. Based on the aqueous geochemical data, the CO2 consumption flux from the chemical weathering of silicate and carbonate rocks within the XJR basin were estimated with the aim of detecting the effect of rock chemical weathering processes on the atmospheric CO2 level in the humid subtropical zones.

　　2　Study area

　　The Zhujiang (Pearl) River, because of its water discharge, is the second largest river in China. The Zhujiang water system consists of four parts: the XJR, the Beijing River, the Dongjiang River, and the Zhujiang delta water system. The XJR, with 35.31×104 km2 of total drainage area, represents 77.8% of the area and 63.9% of the water discharge of the Zhujiang River basin. The XJR passes through Yunnan, Guizhou, Guangxi, and Guangdong Provinces and discharges into the South China Sea. The XJR basin located at the subtropical monsoon zone in southern China, with the mean annual temperature across the drainage basin ranging from 14 C to 22 C, the mean annual precipitation ranging from 1000 mm to 2200 mm, and precipitation decreasing westward spatially and concentrating on the period from April to September (the wet season) temporally. Average annual water discharge over the years (1950-2005) amounts to 230×109 m3 (7293 m3 s-1). The XJR basin consists of various source rocks from Precambrian metamorphic rocks to Quaternary fluvial sediments. Carbonate rock (mainly of limestone) outcrops occupy about 44% of the XJR basin and are mainly distributed in the upper and the middle parts of the basin (Fig.1) where karst geomorphologic processes have been intensively developed both on the

　　Fig.1　The main river system and the carbonate rocks distribution within the Xijiang River basin. Except for the carbonate rocks, other rocks are assigned to be silicate rocks, which include igneous and metamorphic rocks and siliceous sedimentary rocks and deposits.

　　surface and under ground (Yuan, 1994). While silicate rocks and deposits (including igneous and metamorphic rocks and siliceous sedimentary rocks and deposits) have also been intensively chemically weathered for a long time under humid tropical and subtropical circumstances, and have formed a thick layer of red weathering crust at relatively flat sites. Limestone red earth and red weathering crusts are the main parent materials of soil in the XJR basin (Zhu, 1993; Huang et al., 1998; Wang et al., 1999). The geological and environmental factors from active tectonism near the east flank of the Tibetan Plateau, extensive mountainous and hilly surfaces, and long-term agricultural activities together resulted in 15.1% of the XJR basin water and soil loss up to the 1980s (PRWRC, 1991), which is getting much more serious (Xia, 1999). 

　　The sampling section, Wuzhou hydrological gauge station (the drainage area above the station is 32.97×104 km2 and comprises 93.4% of the total XJR drainage area), is located at the linking point of the middle and lower reaches of the main stream of the XJR system. The XJR basin encountered a catastrophic flood in the last 10 days of June 2005 (namely, “056” catastrophic flooding). Water levels of the Wuzhou station at 11:00, 23 June 2005, was 26.75 m asl, and the corresponding flood discharge was up to 53 000 m3 s-1, which was merely less than the catastrophic flooding in 1915 of the XJR (27.07 m asl and 54 500 m3 s-1, respectively). At its annually averaged water level (about 5.5 m asl), the width of the sampling section is about 520 m.

　　3　 Materials and methods

　　Water samples were collected monthly at the Wuzhou section from March 2005 to April 2006. More frequent sampling was conducted on 28-29 April 2005 and around the “056” catastrophic flooding (Fig.2).

　　Fig.2　Daily mean discharge at Wuzhou station (source: http:∥www.pearlwater.gov.cn/index.jsp). The shadow area denotes the frequent sampling period around the “056” catastrophic flood. The black dots (·) in the curve denote the monthly sampling date, and the open circle (o) represents the frequent sampling time.

　　For monthly sampling, three vertical sampling lines were arranged at Wuzhou sampling section, located at the right channel, middle channel, and left channel. Three sampling spots were chosen in each sampling vertical line: located at 0.5 m under the water surface, at mid-water depth, and at 0.5 m above the river bottom, respectively (Gao et al., 2002). The right and left vertical sampling line is 50 m away from the right and left riversides, respectively. Nine samples were obtained for each monthly survey. For the flood sampling, only one water sample was collected at the depth of 1.0 m below the water surface of the middle channel, since the water body was mixed relatively uniformly during the flood.

　　The concentration of HCO-3 was determined in situ by digital titrator (Digital Burette III, Brand Company, Germany) with 0.05 M HCl and pH meter on 100 mL of prefiltrated sample water (the endpoint is 4.3 pH); the precision was about ± 5  M. Total dissolved solids (TDS), electric conductivity (Cond.), water temperature (T), and pH were measured using UltrameterTM Model 6P (Myron L Company, USA), and the precision was ± 0.01 ppm, ± 0.01 ?s cm-1, ± 0.1 C, and ± 0.01, respectively. River water filled in acid-cleaned 1-L glass bottles was shipped to the laboratory and filtrated through Whatman equipment (Whatman GF/F), then analyzed for ions.

　　Concentrations of Ca2+, Mg2+, K+, Na+, and dissolved silicon were analyzed by inductively coupled plasma atomic emission spectrometry (ICP-AES) model IRIS Advantage (Therom Jarrell Ash Corporation, USA). The standard geochemical samples, GSD-12 and GSS-6, were simultaneously analyzed using the same method for quality control. Concentrations of Cl-, SO42-, and NO3- were analyzed by ion chromatography (IC) model DX-600 (Dionex Corporation, USA). Before measurement, the sample was kept refrigerated (4 C). Ion analysis was carried out within 48 h after sampling. The average analytical precision is better than 0.5%.

　　The reliability of the water chemical data can be tested in terms of the balance relationship between the inorganic positive and negative electric charges (Dalai et al., 2002). Under normal circumstances, the riverine positive charge is mainly from the cations: TZ+=2(Ca2++Mg2+)+Na+ +K+, and the riverine negative charge is from the anions: TZ- = HCO-3+2SO42-+Cl- (see Huh et al., 1998), then the inorganic electric charge balance index (e) is calculated as follows:

　　e(TZ+-TZ-)/TZ+(1)

　　Stoichiometric analysis would provide some qualitative information for tracing sources of major dissolved ions in weathering-dominated river waters (Stallard and Edmond, 1983). More quantitative calculations on source of solutes can be carried out on the basis of balance approaches. Generally, direct/forward method (Berner et al., 1983; Velbel, 1986; Meybeck, 1987; Amiotte-Suchet and Probst, 1993; Benedetti et al., 2003; Moon et al., 2007) and indirect/inverse method (Allégre and Lewin, 1989; Negrel et al., 1993; Gaillardet et al., 1999; Roy et al., 1999; Moon et al., 2007) are used in budget. Here, the forward method was applied to explain the contributions of carbonate, silicate, and evaporite weathering to the ion contents in the XJR.

　　The forward method of mass balance is a simple budget to allocate the dissolved substance to the corresponding source minerals by a series of steps (Meybeck, 1987). The mass balance equation for the element X in the dissolved load (in molar concentration) can be expressed as

　　［X］riv=［X］cyc+［X］eva+［X］car+［X］sil+［X］anth

　　where riv=river; cyc = cyclic source; eva = evaporite source; car = carbonate source; sil = silicate source; anth = anthropogenic source, consisting of pyrite oxidation (from mining) and acid precipitation (from fossil fuel combustion).

　　4　Results

　　A total of 178 samples, including 126 monthly samples and 52 flooding samples, were analyzed. The major parameters and ion concentrations of monthly samples are listed in Table 1, and the major parameters and ion concentrations of flooding samples are shown in Fig.3 and also listed in the Supplementary Appendix.

　　Fig.3　The variation of hydrochemical parameters around the “056” catastrophic flood. The dashed line and the number under the line denote the averaged value of the 14 sampling months.

　　Table 1　Hydrological and hydrochemical parameters and electric charge balance of the riverine water in the XJR during the sampling period

　　4.1　Chemical composition of the riverine water

　　For all the monthly samples, the riverine major cations can be ranked in terms of their proportions (in turn, Ca2+>Mg2+>Na+ >K+) and the major anions (in turn, HCO-3 >SO2-4 >Cl- >NO-3) (Table 1; Fig. 4). The pH values of river water range from 7.27 to 7.98, with an average of 7.70. Total dissolved solids vary from 150.6 to 212.7 mg·L-1, with an average of 173.53 mg·L-1, 2.67 times the global median of 65 mg·L-1 (Meybeck and Helmer, 1989). Cations Ca2+, Mg2+, and Na+ comprise 69.81%, 15.74%, and 11.36% of the total, respectively (Fig.4A). Concentration of HCO-3 varies from 1738.2 to 2793.3  M, with an average of 2293.49  M, comprises 84.44% of the total anions; the second major anion SO42- (mean value of 226.22  M) accounts for 8.33% of the total anions (Fig.4B). Because the riverine K+ and NO3- are relatively low in the XJR, we neglected the NO3- and K+ in the following budget.

　　Fig.4　(A) Cation and (B) anion ternary diagrams (in charge equivalent). For the flood samples, the anion data are absent.

　　 For the flooding samples, the pH values of river water range from 7.25 (15∶00 29 June 2005) to 8.18 (8∶00 29 June 2005), with an average of 7.65. Total dissolved solids vary from 125.8 to 175.3 mg·L-1, with an average of 156.57 mg·L-1 being less than that of the monthly samples. Cations Ca2+, Mg2+, and Na+ comprise 75.09%, 13.49%, and 8.27% of the total, respectively (Fig.4A), which is characteristic of the increasing Ca2+ share compared with the monthly samples.

　　Generally, the concentrations of Ca2+ and Mg2+ are much higher for karst water than other nonkarst water. For example, the concentration of (Ca2++Mg2+) is 1895 and 2201  M in Guilin and Liuzhou cities (karst area), Guangxi Province, respectively (Yuan, 1994). However, the concentrations of (Ca2++Mg2+) in the surface water of granite hills in Zhuhai city (Fig.1), Guangdong Province, are mostly < 100  M (29~122  M, averaged at 53  M; Gao et al., unpublished data; Fig.5). Evidently, the higher contents of HCO3- and (Ca2++Mg2+) in the river water indicate that the XJR is characterized by a typical carbonate river: the total riverine alkalinity (Alk.) is mainly determined by HCO3-.

　　Fig.5　The relationship between the content of Ca2++Mg2+ and HCO-3 content in the Xijiang River.

　　Dissolved silicon occurring in the river water is sourced from the chemical weathering processes of the silicate rocks, and the regolith and soils covered the rocks within the drainage basin. Seemingly, the concentration of dissolved silicon is correlated positively with the atmospheric temperature (Fig.6). However, in the monsoon areas, the atmospheric temperature and the precipitation vary in a synchronal pattern. Generally, the dilution effect of the increasing discharge (coming from the precipitation) will exceed the effect of increase in temperature, which will accelerate the release of dissolved silicon from weathering. Another mechanism can be invoked to explain the fact mentioned above (Fig.6): the sources of dissolved silicon in the riverine water are changeable seasonally.

　　Fig.6　The relationship between water temperature and concentration of dissolved silicon in the Xijiang River.

　　In the dry season (from October to March), groundwater (mostly karst water) that is depleted in dissolved silicon dominates the river water; while in the wet season, especially in the flooding period, large amounts of surface water that flow through the surface siliceous soil layers and regolith and bring large amounts of dissolved silicon dominates the river water. Hence, the dissolved silicon is concentrated during the flooding period (with a higher atmospheric temperature) than that in the dry season (with a lower atmospheric temperature).

　　On the Gibbs sketch (Gibbs, 1970), the riverine ions data in present study are located on the middle part of the Y axis and the most left part of the X axis (the figure is omitted). The very low ratios of Cl- to (Cl-+HCO-3) (0.02) and Na+ to (Na++Ca2+) (0.11), and relatively moderate TDS concentrations (162.54 mg L-1) are evidence of the fact that XJR is a typical river dominated by rock chemical weathering on the basis of the Gibbs sketch (Gibbs, 1970).

　　4.2　Variation of dissolved substances during the “056” catastrophic flooding

　　During the “056” catastrophic flooding period, the concentrations of the riverine dissolved substances varied within broad limits: from higher than the averaged value during the sampling year for dissolved silicon, to lower than the averaged value during the sampling year for Ca2+, Mg2+, Na+, K+, HCO-3, and TDS (Fig.3).

　　The concentrations of Ca2+, Mg2+, Na+, K+, HCO-3, and TDS were all higher in the dry season than those in the wet season (Table1; Fig.3), which are consistent with the pattern of dissolved solute variation with increasing discharge in other humid tropical and subtropical rivers (Mortatti and Probst, 2003; Rai and Singh, 2007). This phenomenon can be explained by the dilution effect of the increasing water discharge. However, the dilution is not in 1:1 proportion to the increase in water discharge; the concentrations of various elements decrease by a factor of only about two compared to an increase in the water discharge by about an order of magnitude (see Rai and Singh, 2007). Evidently, the riverine solute source is increased in spite of its relatively less concentration. For example, the concentration of TDS in the Changjiang (Yangtze) River is lower in the wet season than that in the dry season, while the TDS transporting flux in the wet season is about twice higher than that in the dry season (Chen et al., 2002). In the dry season, a significant fraction of the exposed surface in river drainage may remain out of contact with river water and hence be unavailable for interactions. In contrast, during the flooding period, most of the drainage basin comes in contact with river water and thus provides more surface area for chemical weathering (Rai and Singh, 2007). Further, in the wet season, increased physical weathering also promotes more chemical weathering; for example, typhoon activity can enhance silicate weathering rates on Taiwan Island (Goldsmith et al., 2008).

　　5　Discussion

　　5.1　Balance of electric charge in the riverine water

　　The e values of the monthly samples varied from -3.23% to 1.27% (Table 1), with an average of -0.27%, which is close to 1.56% of the averaged e values from 1954 to 2002 at Wuzhou station (Zhang et al., 2007), suggesting that the inorganic electric charge in the XJR is essentially in balance.

　　Because the dissolved organic carbon (DOC) concentration (varying from 0.47 to 2.05 mg·L-1 and with an average of 1.24 mg·L-1; Gao et al., in review) in the XJR is significantly lower than that of other rivers in the world, organic acids can be ignored in the electric charge balance, which coincides with that the contribution of organic ligands is not obvious to the charge balance in the Nanpanjiang and Beipanjiang rivers (see Fig.1), the upper reaches of the XJR (Xu and Liu, 2007). Hence, the riverine DOC is an unimportant medium for rock chemical weathering within the XJR basin.

　　5.2　The contribution of carbonic acid and sulfur acid to the riverine chemical composition

　　Riverine water chemistry is controlled by the chemical weathering of terrestrial rocks, evaporation-crystallization processes, atmospheric precipitation processes, and anthropogenic input (Moon et al., 2007). Terrestrial rock chemical weathering processes mainly produce the cations Ca2+, Mg2+, Na+, and K+; the anions HCO-3 and SO42-; and dissolved silicon. Atmospheric precipitation contributes Na+, Cl-, NO3-, and SO42-. Human-produced effluents bring Na+, Cl-, NO3-, and K+ to the river water in the study area, which makes it more complex to identify the sources of the major riverine ions.

　　On the basis of the theory of rock chemical weathering, Eq.(2) can be simplified as follows:

　　［Ca2+］river=［Ca2+］CO2+SO2+FeS2carbonate+［Ca2+］CO2+SO2+FeS2silicate［Ca2+］gypsum(3)

　　［Mg2+］river［Mg2+］CO2+SO2+FeS2carbonate+［Mg2+］CO2+SO2+FeS2silicate(4)

　　［K+］river=［K+］CO2+SO2+Fes2silicate|［K+］fertilizes(5)

　　［Na+］river=［Na+］CO2+SO2+FeS2silicate+［Na+］sea-salt(6)

　　where the superscript CO2 represents carbonic acid, SO2 is sulfuric acid sourced from atmospheric acid precipitation, and FeS2 stands for sulfuric acid from the oxidation of pyrite. A blank superscript denotes the dissolution process. The subscripts denote the sources of ions. The brace ［ ］ denotes the molar concentration.

　　In the course of calculation, the molar ratio of Cl- to Na+ is served as 1.17 (sea water) for the sea salts. Minor gypsum mineral deposits are scattered within the XJR basin, and the salt-bearing stratum has not been recorded (Geological Bureau of Guangxi Zhuang Autonomous Region, 1985). Hence, part of the riverine Ca2+ and SO2-4 is from the dissolution of gypsums within the drainage basin (Zhang et al., 2007). The equivalent of Ca2+ from the dissolution of gypsum is equal to SO42-, with the assumption that the drainage basin is free of atmospheric acid pollution (Stallard and Edmond, 1983). However, the SO2-4 concentration varied from 3.0 to 8.0 mg L-1, with an average of 5.64 mg·L-1 from 1954 to 1984 for the Zhujiang River (Chen and He, 1999). The increased SO2-4 concentration (21.72 mg·L-1) then in this study results from atmospheric acid precipitation (Larssen et al., 2006) and oxidation of pyrite (the sum of the two items is the human activities induced sulfuric acid). The atmospheric acid precipitation enhanced the concentration of Ca2+ and SO42- in the XJR, while the total alkalinity decreased since the middle of the1980s (Zhang and Chen, 2000). The contribution to riverine K+ from fertilize usage is neglected because of its lower concentration mentioned above (see section 4.1).

　　If the SO42- concentration in the river water of the XJR (sourced from the dissolution of gypsum) is 5.64 mg·L-1 (Chen and He, 1999), then, the SO42- concentration from the pyrite oxidation and atmospheric acid precipitation in the total drainage basin is equal to the measured riverine SO42- concentration minus the SO42- concentration from dissolution of gypsum in numerical value.

　　The amount of cations weathered by H2SO4 (by equivalent, the following is the same) is

　　［TZ+H2SO4］=2×(［SO2-4］river-［Ca2+］gypsum)(7)

　　The amount of cations produced by carbonic acid is

　　［TZ+H2CO3］=2×(［Ca2-4］river+［Mg2+］

　　river)+［K+］river+［Na+］river-［Na+］sea-salts-2×［Ca2+］gypsum-TZ+H2SO4(7)

　　The amount of cations dissolved from gypsum and sea salt is

　　［TZ+evaporite］=［Na+］sea-salt+2×［Ca2+］gypsum(9)

　　The calculated results are listed in Table 2.

　　Table 2　Share of electric charge of the cations sourced from H2SO4 weathering, H2CO3 weathering, and dissolution of the evaporate rocks within the XJR basin

　　Table 2 shows that 81.20% of cations in the XJR originate from the rock chemical weathering caused by carbonic acid and that 11.32% of the riverine cations originate from rock chemical weathering induced by sulfuric acid. The rest of the riverine cations come from the dissolution of gypsum and atmospheric precipitation of sea salts and comprise about 7.48%. The amount of cations produced by carbonic acid is more than 7 times the amount of cations induced by sulfuric acid. Clearly, carbonic acid is a major agent of rock chemical weathering within the XJR basin.

　　5.3　Chemical weathering rates of silicate and carbonate rocks

　　The cycling mechanism of CO2 in the processes of weathering, erosion, transportation, and sedimentation of the carbonate rocks is different from that of silicate rocks. It is important to distinguish the share of atmospheric CO2 consumption in carbonate rock chemical weathering and silicate rock chemical weathering (Dalai et al., 2002; Picouet et al., 2002; Hren et al., 2007).

　　In the chemical weathering of the silicate rocks (basalts), elements Ca and Mg are nearly congruently released (Das et al., 2005). On the basis of the facts that the mass portion of Ca and Mg in igneous rocks of China is 3.24% and 2.22%, respectively (Li and Rao, 1963), and that the ion erosion modulus of catchments dominated by silicate rocks (mainly of granite) in the Zhujiang River basin is about 80.59 t km-2 y-1 (Chen and He, 1999), we calculate that the erosion modulus of Ca2+ and Mg2+ in the silicate rocks region of the XJR basin is 2.61 and 1.79 t km-2 y-1, respectively.

　　For the carbonate rocks region, according to the mass portion of Ca and Mg in dolostone in Guizhou Province being 23.71% and 12.18%, respectively (Ji et al., 2000), and the mass portion of Ca and Mg in limestone being theoretically 40% and 0%, respectively, and the dolostone area representing 10% of the carbonate rocks distribution area in the XJR basin, then the mass portion of Ca and Mg in carbonate rocks in the XJR basin is assigned as 38.37% and 1.22%, respectively. Ion erosion module in the catchments outcropped mainly with carbonate rocks or limestone within the Zhujiang River basin is 269.79 t km-2 y-1 (Chen and He, 1999). According to these facts mentioned above, the erosion module of Ca2+ and Mg2+ in the carbonate rocks region of the XJR basin is 103.52 and 3.29 t km-2 y-1, respectively.

　　As was noted above, the area covered with carbonate rocks comprises 44% of the total drainage basin, and the rest (56%) of the drainage area is covered with silicate rocks and siliceous deposits (for convenience, the siliceous deposits are included into the silicate rocks hereafter). The sources of riverine Ca2+ and Mg2+ in the XJR can be assigned as follows:

　　The share of Ca2+ from carbonate rock chemical weathering is 96.89% (103.52 0.44 /(103.52 0.44+2.61 0.56)), and the remaining 3.11% is from the chemical weathering of silicate rocks.

　　The share of Mg2+ from carbonate rock chemical weathering is 59.09% (3.29 0.44/ (3.29 0.44+1.79 0.56)), and the remaining 40.91% originated from the chemical weathering of silicate rocks.

　　5.4　Estimation of atmospheric CO2 consumed by rock chemical  weathering

　　As shown in Table 2, the total of the positive electric charges originating from chemical weathering driven by carbonic acid is averaged at 2373.4  Eq, which can be assigned to the four main riverine cations (Ca2+, Mg2+, Na+, and K+) according to their proportion in the original records (Table 1) then can be expressed as (Ca2+)calibrated, (Mg2+)calibrated, (Na+)calibrated, and (K+)calibrated, respectively.

　　The cations Ca2+ and Mg2+ produced from carbonate rock chemical weathering induced by carbonic acid are calculated as follows:

　　［TZ+］carbonate=96.89%×［Ca2+］calibrated+59.09%×［Mg2+］calibrated(10)

　　The amount of Ca2+, Mg2+, Na+, and K+ produced in silicate rock chemical weathering caused by carbonic acid is:

　　［TZ+］silicate=3.11%×［Ca2+］calibrated+40.91%×［］calibrated+［K+］calibrated+［Na+］calibrated(11)

　　Given that no bicarbonate is produced in carbonate and silicate rock chemical weathering caused by sulfuric acid (the weathering product is CO2 instead of HCO-3), the total riverine alkalinity is entirely from carbonate and silicate rock chemical weathering induced by carbonic acid.

　　The calculated results (Table 3) show that total cation electric charges balanced basically with the total alkalinity charges, with the difference about 3.38%. Hence, the parameters TZ+carbonate and TZ+silicate (Table 3) can be employed for estimating the amount of atmospheric CO2 consumed by the chemical weathering of carbonate and silicate rocks in the XJR basin, respectively.

　　Total annual amount of the cation electric charges and total alkalinity charge resulting from the rock chemical weathering is calculated by

　　tatal=∑365i=1pi×Qi(12)

　　where total is the total amount of riverine dissolved substance, pi is the daily mean concentration of the dissolved substance, Qi is daily mean discharge. We assumed that the concentration of riverine dissolved substance was uniform in the same sampling month.

　　Table 3　Comparison between the cation charge and alkalinity sourced from the rock chemical weathering induced by CO2 ( Eq)

　　a　Note: Balance index (%) = (［Total TZ+］-［alkalinity］)/［Total TZ+］ 100%.

　　We calculated that TZ+carbonate= 3.46×1011 Eq; TZ+silicate = 0.64×1011 Eq; and Alkalinity = 3.98×1011 Eq.

　　The total cation electric charges (4.10×1011 Eq) from rock chemical weathering within the XJR basin is basically balanced with the total alkalinity charge, with a difference of 3.38%.

　　Then, the flux of cation electric charges for the carbonate and silicate rock areas is, respectively:  TZ+carbonate=23.83 ×105 Eq km-2 y-1 and  TZ+silicate = 3.44×105 Eq km-2 y-1.

　　For the chemical weathering of carbonate rocks, half of the bicarbonate originates from the atmospheric CO2 and another half from the weathering products of the carbonate rocks (Amiotte-Suchet and Probst, 1995). Consumption flux of atmospheric CO2 induced by the chemical weathering processes of the carbonate and silicate rocks in the XJR basin is calculated as 11.92×105 mol km-2 y-1 (or 1.73×1011 mol y-1) and 3.44×105 mol km-2 y-1 (or 0.64×1011 mol y-1), respectively. The total consumption flux of atmospheric CO2 within the XJR basin is about 7.17×105 mol km-2 y-1 (or 2.37×1011 mol y-1). From 18 June 2005 to 26 June 2005 (9 days), the Wuzhou hydrological station was reported as at the warning water level, during which CO2 consumption was 0.38×1011 mol and was comprised of 15.94% of the annual CO2 consumption from rock chemical weathering in the XJR basin.

　　5.5　Comparison

　　Rock chemical weathering rates are determined mainly by atmospheric temperature, precipitation, rock types, geomorphic features, plant coverage, and characteristics of human activities within the drainage basin (White and Blum, 1995; Ludwig et al., 1999; White et al., 1999). Riverine bicarbonate transporting flux from 1997 to 1998 was 11.03×105 mol km-2 y-1 for the XJR basin and 12.89×105 mol km-2 y-1 for the Beijiang River basin. However, the sources of the bicarbonate were not partitioned (Gao et al., 2001). The uptake rate of atmospheric CO2 caused by the chemical weathering of silicate and carbonate rocks in the Nanpanjiang-Beipanjiang river basin (see Fig.1) is (0.72~1.3)×105 mol km-2 y-1 and (6.14~7.43)×105 mol km-2 y-1, respectively (Xu and Liu, 2007), which are lower than the results surveyed at the Wuzhou station in this study. Much less precipitation (800~1200 mm y-1) and lower atmospheric temperature (10~15℃) can be employed to explain the lower chemical weathering rate and atmospheric CO2 uptake rate of the Nanpanjiang-Beipanjiang river basin.

　　The uptake rate of atmospheric CO2 in the humid tropical watershed located in northern Okinawa Island, Japan (annual mean atmospheric temperature and precipitation is 22.2 C and 2000 mm, respectively), with silicate rocks being the dominating bedrocks, is (3.34~4.71)×105 mol km-2 y-1 (Vuai and Tokuyama, 2007). For a basalt watershed in the Lesser Antilles (annual mean atmospheric temperature and precipitation is 24~28 C and 2400~4600 mm, respectively), the uptake rate of CO2 in chemical weathering is (11~14)   105 mol km-2 y-1 (Rad et al., 2006). For the Yamuna drainage basin, located at the southern slope of the Himalaya, the rate of atmospheric CO2 uptake by chemical weathering is (2.0~4.0)   105 mol km-2 y-1 (Singh et al., 2005). By comparison, we concluded that the calculated results on the flux of atmospheric CO2 consumed by rock chemical weathering in the XJR basin are reasonable.

　　The CO2 flux consumed by rock chemical weathering within the XJR basin is 3.5% of the CO2 released from volcanism and metamorphism (67×1011 mol y-1) and 4% of the net flux of atmospheric CO2 consumed by global silicate weathering (58×1011 mol y-1) (Gaillardet et al., 1999). Hence, the higher CO2 consumption caused by chemical weathering of rocks in humid subtropical zones regulates the atmospheric CO2 level and constitutes a significant part of the global carbon budget. Furthermore, responding to global warming, the rates of rock chemical weathering will be increased, and the carbon sink potential of the rock chemical weathering process in the humid subtropical zones cannot be neglected in the global carbon budget.

　　6　Conclusion

　　Aqueous pH values in the XJR vary from 7.27 to 7.98, with an average of 7.70, and reveal that the riverine total alkalinity is mainly composed of bicarbonate alkalinity. The concentration variation of the dissolved substances in the XJR responds to the dilution effect of the water discharge, with the exception of dissolved silicon. Riverine inorganic electric charges of the XJR basin are basically balanced.

　　Dissolved chemical substances in the XJR mainly originated from the chemical weathering of carbonate and silicate rocks within the drainage basin. Calculations show that 81.20% of cations in the XJR originated from rock chemical weathering caused by carbonic acid and that 11.32% of the riverine cations originated from rock chemical weathering induced by sulfuric acid. The remaining riverine cations came from the dissolution of gypsum and atmospheric precipitation of sea salts, comprising about 7.48%.

　　The total CO2 flux consumed by rock chemical weathering within the XJR basin is 2.37×1011 mol y-1, of which 0.64×1011 mol y-1 resulted from silicate rock chemical weathering and 1.73×1011 mol y-1 originated from carbonate rock chemical weathering. The CO2 consumption is 0.38×1011 mol during the “056” catastrophic flooding and comprises 15.94% of the annual CO2 consumption from rock chemical weathering in the XJR basin. The CO2 consumption flux within the XJR basin is 3.5% of the CO2 released from volcanism and metamorphism and 4% of the net flux of atmospheric CO2 consumed by global silicate weathering.

　　The CO2 consumption caused by rock chemical weathering in humid subtropical zones regulates the atmospheric CO2 level and constitutes a significant part of the global carbon budget. Responding to global warming, the rates of rock chemical weathering will be increased and the carbon sink potential of the rock chemical weathering process in humid subtropical zones deserves extra attention in the global carbon budget.

　　Acknowledgements

　　This study was supported by the Natural Science Foundation of China (No. 40671027, 40871143), the open funds of China Institute of Water Sources and Hydropower Research, and the Guangdong Provincial Natural Science Foundation of China (7003669). We would like to thank four anonymous reviewers for their constructive reviews and comments. We would like to thank Cui Kunyan, Lai Zhihui, Liu Shuifu, and Feng Shunqing, the Instrumentation Analysis and Research Center of Sun Yat-sen University, for their analysis of the ion contents.

　　Appendix

　　Table A. Hydrological and hydrochemical parameters during the 2005 catastrophic flood in the Wuzhou sampling section of the Xijiang River, southern China.

　　原載：Geomorphlolgy,2009,106:324-332.