MARLET Serge (*),VALLÈS Vincent (**),LAFOLIE François (**) CONDOM Nicolas (*)

Transcription

1 Scientific registration n 537 Symposium n 29 Présentation : Oral Hydrogeochemical modeling: a suitable approachto predict the effect of irrigation on soil salinity,sodicity and alkalinity Modélisation hydrogéochimique :une méthode adaptée à la prévision de la salinité, de l alcalinité et de la sodicité des sols sous irrigation MARLT Serge (*),VALLÈS Vincent (**),LAFOLI François (**) CONDOM Nicolas (*) (*) CIRAD, BP 5035, Montpellier Cedex 1 (**) INRA, Station de science du sol, Avignon Cedex 9 After aton [1950] and Richards [1954], irrigation water quality is usally evaluated by the Residual Sodium Carbonates (RSC), the lectrical Conductivity (C) and the Sodium Adsorption Ratio (SAR) and classified with respect to alkalinization, salinization and sodification hazard. This study aims at showing that geochemical modeling can greatly improve the evaluation. More, such geochemical models can be coupled with hydrological models and allows to predict the trend of soil chemical properties according to water quality, soil properties and irrigation and drainage management. GOCHMICAL MODL PRSNTATION We used the model IRRICHM derived from the thermodynamic model GYPSOL [Vallès and Bourgeat, 1988] which accounts for speciation of the major chemical component, precipitation or dissolution of few minerals and ion exchange. This model is sufficiently simplified so that its repeated use is not too time consuming, suitable to conditions prevailing in cultivated soils, but also accurate and reliable for decision making in soils management. Speciation of the major chemical component The major variables of the chemical system are K, Na, Ca, Mg, Cl, SO 4, Si, alkalinity and CO 2. Alkalinity is calculated by assuming electrical neutrality of the solution. Alkalinity ~ (molc.l SUP { -1 } ) ~ = ~ 2Ca + 2Mg + Na + K - Cl - 2SO SUB { 4 } In the solution that are rich in divalent cations and weak acid anions, ion pairs can form and 34 aquaeous species were considered. Their activities were calculated from an ion pair model using the Debye-Hückel s law extended to saline conditions and their dissociation constant (table 1). It has been validated for the ion behavior in natural waters up to a ionic strengh of 1. 1

2 Table1: Dissociation constant of the aqueous species (Vallès et Bourgeat, 1988 Dissociation reactions log K (25 C) Dissociation reactions log K (25 C) H 2 OWH + + OH - HSO 4 - W H + + SO 4 HCO 3 - W H + + CO 3 H 2 CO 3 0 W 2H + + CO 3 H 2 CO 3 0 W CO 2 + H 2 O NaCl 0 W Na + + Cl - NaOH 0 W Na + + OH - NaSO 4 - W Na + + SO 4 Na 2 SO 4 0 W 2Na + + SO 4 NaCO 3 - W Na + + CO 3 Na 2 CO 3 0 W 2Na + + CO 3 NaHCO 3 0 W Na + + HCO KCl 0 W K + + Cl - KSO 4 - W K + + SO 4 KOH 0 W K + + OH - MgCO 3 0 W Mg 2+ + CO 3 MgHCO 3 + W Mg 2+ + HCO 3 - MgOH + W Mg 2+ + OH - MgSO 4 0 W Mg 2+ + SO 4 CaCO 3 0 W Ca 2+ + CO 3 CaHCO 3 + W Ca 2+ + HCO 3 - CaOH + W Ca 2+ + OH - CaSO 4 0 W Ca 2+ + SO 4 H 4 SiO 4 0 W H 3 SiO H Precipitated species When water evaporates, precipitates occur and affect the composition of the soil solution. Common precipitates are: calcite, chert (colloïdal silica, chalcedony), gyspum and clay minerals [Appelo and Postma, 1995]. Mg content is low in calcareous deposit and carbonates do not fully account for the control of Mg molality in solution. The saturation of extracts of soil solution with respect to these minerals were evaluated in Niger [Marlet et al.,1996] and Pakistan [Marlet, 1997]. 2

3 Pakistani data displays that saturation is achived for calcite (fig.1), sepiolite (fig.2) and illite (fig.3) and that the trend of the data parallels that of the theoritical equilibrium lines. In case of Niger, illite and sepiolite do not achieve saturation. However, the 0 undersaturation is low. Closely related minerals do precipitate. H 4 SiO 4 remains undersaturated with respect to amorphous silica (fig. 2 and 3), abundant in these soils. Slow dissolving of amorphous silica could explain the apparent disequilibrium and H 4 SiO 0 4 should be issued from silicates dissolution. xcept for one soil sample, gypsum does not achieve saturation but could precipitate at higher concentration. In the model, calcite, sepiolite, illite and gypsum precipitation was allowed and the solution was supposed in equilibrium with respect to kaolinite and amorphous silica (table 2). Table 2: Dissolution reactions of minerals - Thermodynamic data " \f D Minerals Disolution Reaction pk Calcite CaCO 3» Ca CO Amorphous silica H 4 SiO 4» H 4 SiO Gypsum CaSO 4.2H 2 O» Ca 2+ +SO 4 +2H 2 O Sepiolite Si 6 Mg 4 O 15 (OH) 2.4H 2 O+8H + +3H 2 O»4Mg ++ +6H 4 SiO Kaolinite Si 2 Al 2 O 5 (OH) 4 +6H +»2Al 3+ +2H 4 SiO 0 4 +H 2 O Illite Si 3.5 Al 2.3 Mg 0.25 O 10 (OH) 2 K H + +3H 2 O»0.6K Al Mg H 4 SiO Cation exchange The cation exchange isotherms are defined by a selectivity coefficient in the Gaines and Thomas convention [Gaines and Thomas, 1953]. For the exchange reaction between a cation A of valence a and a cation B of valence b, adsorbed on the exchanger soil, the selectivity coefficient is modeled according to an indirect method proposed by Rieu et 3

4 al. [1991]. The relation between the exchangeable fraction ratio (FR) and the activity fraction ratio (AFR) is fitted as a power law function which writes : FR = a b b a ( B) = α ( A) a b β = α [ AFR] β where (A) and (B) are the activities, a and b are the valences, A and B are the charge (or equivalent) fractions of adsorbed ions A and B, and " and $ are parameters. The exchange between Ca 2+, Mg 2+, Na + and K + are considered in the model and the sum of the equivalent fraction of the cations adsorbed is equal to the Cationic xchange Capacity (CC). In Niger and Pakistan, the soil samples residue were dried after extraction and the cationic exchange capacity and the adsorbed K, Na, Ca and Mg were determined by the cobaltihexamine chloride method. The relations do not differ significantly between Niger and Pakistan. The results are presented in table 3 and the example of Na-Ca exchange is presented in figure 4. Table 3: Cation exchange: relation between exchangeable fraction ratio (FR) and activity fraction ratio (AFR) - fitted data for Pakistan and Niger Na-Ca exchange Na-Mg exchange Na-K exchange Na Ca = ( Na) ( Ca) Na Mg = ( Na) ( Mg) K Na ( K) = ( ) Na IRRIGATION WATR QUALITY ASSSSMNT Residual alkalinity concept The principle of successive precipitation of minerals has been worked out by aton [1950] and Hardie and ugster [1970] in an evolutionary sequence for evaporating waters. It suggests that the composition of saline waters is a function of the concentration ratios in the starting solution where alkalinity plays an important part. When a solution is concentrated by evaporation up to the point that calcite (CaCO 3 ) precipitates, alkalinity and calcium molality cannot increase together. In case Ca 2+ equivalents exceed alkalinity equivalents in the original solution, alkalinity decreases and calcium molality increases. In a reverse situation, alkalinity increases and calcium 4

5 molality decrease. This concept of Residual Alkalinity (RA) has been generalized to the successive precipitation of several minerals [Van Beek and Van Breemen, 1973; Al Droubi et al., 1980]. Residual Alkalinity is calculated adding the cations charges and substracting the anions charges, which are involved in precipitates, to alkalinity. Simulation of irrigation water evaporation In order to illustrate the various geochemical mechanisms according to the residual alkalinity concept, the evaporation of three water qualities issued from the Niger river (Niger), the Indus river and a tubewell (Pakistan) were simulated according to the geochemical model (table 4). A cationic exchange capacity of 0.12 mol c /kg, bulk density of 1.5 and a pco 2 of atm. were chosen for the simulations. Soil solution is assessed in equilibrium with respect to calcite, sepiolite and illite all over the concentration process. Chloride does not interact with its environnement and is used as a chemical tracer. Table 4: Composition and residual alkalinity (RA) of irrigation waters (mmol c /l) Niger Indus Tubewell 1 Tubewell 2 calcium magnesium sodium potassium chloride sulfate alkalinity RA: calcite+sepiolite+illite RA: calcite+sepiolite+illite+gypsum The niger river (fig. 5) is characterized by a positive residual alkalinity (with respect to calcite, sepiolite and illite precipitations). During the concentration process, calcite, sepiolite and illite precipitate continuously. Thus alkalinity increases at a lower rate than the chemical tracer while calcium, magnesium and potassium molalities decrease in soil solution. 5

6 The canal water in Pakistan (fig.7) is characterised by a negative residual alkalinity (with respect to calcite, sepiolite and illite precipitations). At the beginning of the concentration process, calcite, sepiolite and illite precipitate and alkalinity decreases while calcium, magnesium and potassium increase at a lower rate than the chemical tracer. When gypsum precipitates, residual alkalinity (with respect to calcite, sepiolite, illite and gypsum precipitations) increases but remains negative. Thus the rates of decrease in alkalinity and increase in calcium molality slow down while sulfate molality decreases. The tubewell 1 water in Pakistan (fig.7) is characterized by a negative residual alkalinity (with respect to calcite, sepiolite and illite precipitations). At the beginning of the concentration process, calcite, sepiolite and illite precipitate and alkalinity decreases while calcium, magnesium and potassium increase at a lower rate than the chemical tracer. When gypsum precipitates, residual alkalinity (with respect to calcite, sepiolite, illite and gypsum precipitations) becomes positive. Thus alkalinity increases while calcium, magnesium and potassium decrease and sulfate molality increases in the soil solution. In figures 5, 6 and 7, Na molality increases at a lower rate than the chemical tracer because it adsorbs on the exchange complex while calcium, magnesium and potassium desorb and neutralize alkalinity through minerals precipitation. Cation exchange acts as a buffer with respect to alkalinization, more particularly in clayey soils with high cationic exchange capacity. 6

7 HYDROGOCHMICAL MODLLING Residual alkalinity is a good indicator for water quality assessment but water management must be taken into acount for relevant prediction.two examples are presented to illustrate how this geochemical model can be coupled with different hydrological models to predict salinity, sodicity or alkalinity according to water quality and irrigation management. xample 1: Irrigation water quality and management in Pakistan In Pakistan, the geochemical model have been coupled on a simple water balance model [Marlet, 1997]. At each calculation step, the molality of each chemical component is calculated from the solutes added by irrigation and substarcted by leaching. The model was run for a 40-year period and 6 simulations have been realized under: - two assumptions of leaching fraction (LF) of 0.01 and 0.1.We considered an irrigation amount I of 1000 mm/year and an available water content (AW) of 200 mm; - three assumptions of water quality: canal water; a mix of canal water and 10% of the tubewell 1 water (mix1); and a mix of canal water and 10% of the tubewell 2 water (mix2) (table 4). We considerd of pco 2 of , a CC of 0.12 mol c.kg -1 and a bulk density of 1.5. The salinization hazard is due to irrigation management (or to the use of saline water that was not considered in this study) (fig.8). ven if good water quality is used, salinization can occur if the leaching fraction is low (poor water management). The sodification hazard is mainly due to water quality (fig. 9). ven if the irrigation management is good, the use of poor water quality can lead to soil sodicity that become much more a problem than salinity. Furthermore, low tubewell water quality use 7

8 without mixing can lead to very fast soil degradation even if the soil is leached with fresh water during the next irrigation. xample 2 : Time-changes and location of sodicity in Niger After Simunek and Suarez [1994] and Hutson and Wagenet [1995], the geochemical model was coupled on a solute transport model based on Richard s equation and the convection dispersion equation under the assumption of both physical and chemical equilibrium [Marlet, 1996]. Our purpose is to predict the long term trend of soil chemical properties under the influence of irrigation within a context of positive residual alkalinity. We considered a medium-type scenario that mimics the furrow irrigation practice and cropping system in Lossa s irrigated scheme. The model was run for a 27-year period at the Lossa site (Niger). Among the numerous modifications including the soil solution, the exchange complex or the minerals, we focused on the time changes and distribution of sodicity. Sodium adsorption is closely related with the concentration of the soil solution which depends on both the rate of evaporation and root water uptake which decreases from the soil surface down to the rooting depth, and solutes transport by the water flow (fig. 10). These opposite phenomena explains that the highest sodium adsorption is observed at 40 cm in depth. At lower depths, the sodium adsorption increases at lower rate. In the upper layer, the sodification is at first faster then tends towards an equilibrium. The soil sodification is slow; for the simulation period, SP increases from 1.34% to 3.50% at the soil surface and to 11.2% at the maximum concentration while it does not vary consistently at the bottom of the profile. References: Al Droubi, A., B. Fritz, J.Y. Gac, and Y. Tardy, Generalized residual alkalinity concept: application to prediction of the chemical evolution of natural waters by evaporation, Am. J. Sci.,280, , Appelo, C.A.J., and D. Postma, Geochemistry, Groundwater and Pollution, Balkema, Rotterdam, 1994 aton, F.M., Signifiance of carbonates in irrigation waters, Soil science,69, ,1950. Gaines, G.L., and H.C. Thomas, Adsorption studies on clay minerals: II. A formulation of the thermodynamics of exchange adsorption, J. Chem. Phys., 21, , Hardie, L.A., and H.P. ugster, The evolution of closed-basin brines, Miner. Soc. Am. Spec. Publ., 3, , 1970 Hutson, J.L., and R.J. Wagenet, An Overview of LACHM: A process based model of water and solute movement, transformationsn, palnt uptake and chemical reactions in the unsaturated zone, Chemical quilibrium and Reaction Models, SSSAJ Special Publication 42, , Lafolie, F., Modelling water flow, nitrogen transport and root uptake including physical non-equilibrium and optimization of the root water potential, Fertilizer research, 27, , Marlet, S., Alcalinisation des sols dans la vallée du fleuve Niger (Niger): Modélisation des processus physico-chimiques et évolution des sols sous irrigation, Doct. Thesis, NSAM, Montpellier, France,

9 Marlet, S., V. Vallès, and L. Barbiero, Field study and simulation of geochemical mechanisms of soil alkalinization in the sahelian zone of Niger, Arid Soil Research and Rehabilitation, 10: , 1996 Marlet, S., Salinization of the irrigated soils in the Punjab (Pakistan), IIMI Pakistan, 32 p., 1997 Richards, L.A., Diagnosis and improvement of saline and alkali soils, United States Salinity Laboratory, Agriculture handbook n 60, USDA, 160 p., 1954 Rieu, M., J. Touma, and H.R. Gheyi, Sodium-calcium exchange on brazilian soils: modeling the variation of selectivity coefficients, Soil Sci. Soc. Am. J., 55, , Simunek, J., and D.A. Suarez, Two-dimensional transport model for variably saturated porous media with major ion chemistry, Water Resour. Res., 30(4), , Vallès, V., and F. Bourgeat, Geochemical determination of the gypsum requirement of cultivated sodic soils. I. Development of the thermodynamic model GYPSOL simulating the irrigation water-soil chemical interactions, Arid Soil Research and Rehabilitation, 2, , Van Beek, C.G.., and N. Van Breemen, The alkalinity of alcali soils, Journal of soil science,24 (1), , Key words : salinization, alkalinization, sodification, irrigation, irrigation water quality, geochimical, model, residual alkalinity, Niger, Pakistan Mots clés : salinisation, alcalinisation, sodification, irrigation, qualité de l eau d irrigation, modèle géochimique, alcalinité résiduelle, Niger, Pakistan 9