Hydrothermal source of radiogenic Sr to Himalayan rivers

Matthew J. Evans   Department of Geological Sciences, Cornell University, Ithaca, New York 14853-1504, USA Louis A. Derry  Suzanne P. Anderson Department of Earth Sciences, University of California, Santa Cruz, California 95064, USA Christian France-Lanord Centre des Recherches Pe´trographiques et Ge´ochimiques, BP 20, Vandouvre-les-Nancy, 54501, France

ABSTRACT ment; formations III and II of the crystallines Hot-spring waters near the Main Central thrust in the Marsyandi River of central are composed of calcic gneisses and marbles have Sr concentrations to 115 ␮M with 87Sr/86Sr to 0.77. Small amounts of hydrothermal as well as pelitic augen gneisses, and forma- water (Յ1% of total river discharge) have a signi®cant impact on the solute chemistry tion I contains mainly quartzo-pelitic gneisses and the budget of radiogenic Sr in the Marsyandi. In the upper Marsyandi, river chem- and some migmatites. The Manaslu leuco- istry re¯ects carbonate weathering, with 87Sr/86Sr Յ 0.72. As the Marsyandi ¯ows across granite is exposed in the northeastern part of the dominantly silicate High Himalayan Crystalline terrane, both 87Sr/86Sr and [Sr] in- the drainage. Variably metamorphosed Pre- crease, associated with increases in the concentration of Na؉,K؉, and Cl؊, all of which cambrian sedimentary rocks make up the are high in the hydrothermal waters. Cation concentrations decrease along the Lesser Lesser Himalaya, pelitic schists and minor do- 13 Himalayan reach of the river. Hot-spring dissolved CO2 has a ␦ C value to ؉5.9½, lomitic carbonates in the upper formation and indicating that metamorphic decarbonation reactions contribute CO2 to the ¯uids. Hy- quartzites and schists in the lower formation. drothermal CO2 is partially neutralized in high-temperature weathering reactions, which Numerous springs are found along the Mar- generate alkalinity and yield abundant radiogenic Sr. Radiogenic hydrothermal carbonate syandi, particularly near the Main Central can form from these solutions and later weather, releasing silicate Sr but imparting car- thrust and the South Tibetan detachment. bonate characteristics to the overall water chemistry. SAMPLING AND ANALYTICAL Keywords: hot spring, alkalinity, Himalaya, strontium, geothermal system. METHODS Main stem, tributary, and spring samples INTRODUCTION Crystalline series, the Main Central thrust, and were collected in April 1997 and April 1999. The Sr ¯ux from Himalayan rivers is an im- the Lesser Himalaya sequence (Colchen et al., All water samples were ®ltered and stored in portant contributor to the Sr isotope mass bal- 1980; Upreti, 1999). The Tethyan Sedimenta- acid-washed polyethylene bottles. Cation sam- ance of the modern oceans, and Himalayan ry Series comprises a Cambrian to early Ter- ples were acidi®ed with concentrated HNO3 rivers are unusual among world rivers in that tiary stable platform sequence dominated by in the ®eld. Cations were determined by in- they have both high concentrations of Sr and shelf carbonates. The High Himalayan Crys- ductively coupled plasma±mass spectrometry high 87Sr/86Sr (Edmond, 1992). Weathering of talline series is high-grade metamorphic base- or emission spectrometry (Cornell University various silicate phases and weathering of ra- diogenic carbonates have been proposed as sources for the radiogenic Sr ¯ux in Himala- yan rivers (Palmer and Edmond, 1989; Krish- naswami et al., 1992; Derry and France- Lanord, 1996; Quade et al., 1997; Singh et al., 1998; Blum et al., 1998; Galy et al., 1999; English et al., 2000). Each of these sources has different implications for the interpretation of the marine Sr record. We have determined solute chemistry for the main stem, all major tributaries, and both hot and cold springs of the Marsyandi River in central Nepal. We ®nd that hot springs contribute signi®cant amounts of radiogenic Sr and other ions to the Mar- syandi River.

GEOLOGIC SETTING The Marsyandi River ¯ows over ϳ150 km from its headwaters north of the Annapurna range to its con¯uence with the Trisuli River, and has a basin area of ϳ4800 km2 (Fig. 1). This major tributary to the River crosses the main tectonostratigraphic units that make up the Nepal Himalaya, and the shear zones that bound them. The Marsyandi Figure 1. Tectonic sketch map of Marsyandi basin, central Nepal. TSSÐTethyan Sed- imentary series; STDSÐSouth Tibetan detachment system; HHCÐHigh Himalayan ¯ows across the Tethyan Sedimentary Series, Crystalline series; MCTÐMain Central thrust; LHÐLesser Himalaya. Sampling loca- the South Tibetan detachment system, forma- tions are marked. Symbols and shading continue in following ®gures. Representative tions III, II, and I of the High Himalayan 87Sr/86Sr values for bedrock are indicated. Fm is formation.

᭧ 2001 Geological Society of America. For permission to copy, contact Copyright Clearance Center at www.copyright.com or (978) 750-8400. Geology; September 2001; v. 29; no. 9; p. 803±806; 4 ®gures; Data Repository item 2001091. 803 Figure 2. Marsyandi main stem, tributary, and spring chemistry. Concentrations in micromoles/kg (␮M). Note scale break on all vertical axes to accom- modate high concentra- tions in hot-spring waters. Shading corresponds to map units in Figure 1 and symbols follow its key. Downstream distance measured from Khangsar Khola. K, Na, and Cl are particularly anomalous in hot springs and clearly af- fect main-stem chemistry down to its con¯uence with Trisuli River.

and University of California, Santa Cruz), and 54 thermal ionization mass spectrometer increase occurs within the High Himalayan anions were determined by ion chromatogra- (Cornell). Spring waters and some river wa- Crystallines, concurrent with the increase in phy (Cornell and Centre des Recherches PeÂ- ters were spiked with an enriched 84Sr tracer [Sr]. Each of the lithologic units in the Mar- trographiques et GeÂochimiques [CRPG]). Sr and [Sr] was determined by isotope dilution. syandi basin has distinct Sr characteristics that isotope ratios were analyzed on a VG Sector Separate samples were reacted under vacu- are re¯ected in tributary streams from mon- um with H3PO4, and the evolved CO2 was olithologic catchments. Tethyan tributaries measured for determination of ␦13Cona have high Sr concentrations (ϳ4.5 ␮M) and modi®ed VG 602D mass spectrometer. The low 87Sr/86Sr, near 0.72. Tributaries from for- 18 87 86 ␦ O and ␦D values from H2O were deter- mations II and III have similar Sr/ Sr at mined for six spring samples by CO2 equil- 0.721, but much lower [Sr] (ϳ0.6 ␮M). For- ibration and reduction over hot uranium, re- mations II and III contain abundant marbles, spectively (CRPG). and as in the Tethyan reach, these carbonates strongly control the riverine Sr in both the RESULTS AND DISCUSSION main stem and tributaries. Formation I tribu- Main Stem and Tributary Chemistry taries are radiogenic (87Sr/86Sr ϭ 0.744) with Within the Tethyan series Ca, Mg, and Sr dilute [Sr] (ഠ0.5␮M). Lesser Himalayan concentrations1 are high in both the tributaries streams have [Sr] of only ϳ0.3 ␮M but very and main stem, re¯ecting carbonate weather- high 87Sr/86Sr ratios (to 0.87). ing (Figs. 2 and 3). Within the High Himala- yan Crystalline and Lesser Himalaya series Spring Chemistry these ions show overall decreases in concen- Five hot (ϳ30±55 ЊC) springs and two cold tration, except for main-stem [Sr], which in- springs along the Marsyandi were sampled. creases in formation I. Levels of K, Na, and Hot springs are Cl rich (Fig. 2) with total dis- Cl in the Marsyandi and its tributaries are rel- solved solid (TDS) values of 800±7000 mg/ atively low upstream of the South Tibetan de- L. Cold springs have higher Ca/Na and Ca/Cl tachment, but increase two to tenfold overall ratios but are much more dilute (TDS ഠ 300± within the High Himalayan Crystallines. [Na], 600 mg/L) and closely resemble nearby trib- [Cl], and [K] in Lesser Himalayan tributaries utaries. Sr in the hot springs is consistently are mostly relatively dilute. Measured and cal- more radiogenic and concentrated than in both culated alkalinities are similar and range from the main stem and tributary values at the same ഠ1500 to 1800 ␮M for the main stem; the position along the Marsyandi; the hot springs highest values are in the Tethyan series. de®ne a Sr source distinct from the tributaries, The 87Sr/86Sr values for the main stem of with 87Sr/86Sr ranging from 0.736 to 0.768 Figure 3. A: Sr concentrations in waters. the Marsyandi increase downstream from Note vertical axis break for hot-spring sam- and [Sr] from 15 to 115 ␮M (Fig. 3). The two ples. Numbers give ␦13C (in per mil) of dis- 0.719 to 0.733 (Fig. 3) and the majority of this cold springs sampled near the South Tibetan 87 solved CO2 in hot-spring samples. B: Isoto- detachment are much less radiogenic ( Sr/ pic ratio of dissolved Sr, symbols as in 1GSA Data Repository item 2001091, Water 86Sr ϭ 0.712) and concentrated ([Sr] ϭ 6±3 Figure 1. Compared to tributaries, hot chemistry data, is available on request from Docu- springs in formation I of High Himalayan ments Secretary, GSA, P.O. Box 9140, Boulder, CO ␮M) than their warmer counterparts and re- 13 Crystallines have ϳ100 times more Sr, and 80301-9140, [email protected], or at www. semble Tethyan tributaries. The ␦ C of dis- are more radiogenic. geosociety.org/pubs/ft2001.htm. solved inorganic carbon in the sampled

804 GEOLOGY, September 2001 tion of hot-spring alkalinity that is derived from carbonate versus silicate sources. The fraction of hot-spring alkalinity present as Na* and K necessarily represents a minimum es- timate of the silicate-derived alkalinity ¯ux from the springs. The positive ␦13C values (ϩ0.48½ to ϩ5.9½) in spring water CO2 imply that some of the dissolved CO2 (alkalinity) is derived from metamorphic decarbonation reactions (Shieh and Taylor, 1969; Friedman and O'Neil, 1977). Metamorphic CO2 has been identi®ed in hot springs from the nearby Mus- tang graben (Galy and France-Lanord, 1999). The high alkalinities of the hot springs indi- cate that a signi®cant quantity of metamorphic CO2 has already been neutralized by reactions with base cations.

Source of Sr in Marsyandi River The hot-spring ¯ux has a signi®cant impact Figure 4. Mixing diagram for Marsyandi and its sources within High Himalayan on the isotopic and mass balance of Sr in the Crystallines. Solid arrow shows expected mixing line between main stem and Marsyandi. Within the Tethyan series and for- formation I tributaries. Open arrow shows expected mixing line between main mations II and III, the 87Sr/86Sr ratios of the stem and hot-spring source. Tributaries are radiogenic but dilute, and cannot cause increase in Sr concentration observed within formation I. Hot springs are Marsyandi are similar to those of its tributar- both radiogenic and highly concentrated and drive main stem to higher [Sr] and ies; formation II and III tributaries only serve 87Sr/86Sr. to slightly dilute the main stem. Formation I tributaries are also dilute, and if they are the springs ranges from Ϫ3.1½ to ϩ5.9½ (Fig. Marsyandi despite the fact that they contribute only new source of solutes to the main stem, 3). The springs with the highest Cl concentra- only a small amount of the water ¯ux. then the Marsyandi should evolve toward tions (to ϳ57 000 ␮M) also have the highest more radiogenic and dilute values (Fig. 4). [Sr] and 87Sr/86Sr values (ϳ0.77) and tend to Hydrothermal Solutes However, Sr in the Marsyandi River becomes have positive ␦13C. Calculated alkalinities for The high [Cl] and [Na] suggest that basinal both more radiogenic and concentrated in for- the highest-Cl hot springs range from 36 000 brines or buried evaporites could be sources mation I of the High Himalayan Crystallines, to 47 000 ␮M. The hot springs are supersat- of solutes to the springs. Evaporites are known following a mixing line toward sampled hot urated with both calcite and quartz, and trav- from possibly correlative Lesser Himalayan springs. Mixing with the radiogenic and dilute ertine deposits are common near the springs. strata in Pakistan; however, no evaporites have Lesser Himalayan tributaries causes a de- 87 86 The ␦18O and ␦D range from Ϫ10.3 to Ϫ14.4 been described in the Lesser Himalaya of cen- crease in [Sr] and a slight increase in Sr/ Sr. and from Ϫ67.5 to Ϫ101.9, respectively. The tral Nepal. We make the simplifying assump- Approximately 30% of the Sr ¯ux of the Mar- ␦18O and ␦D values are very close to the local tion that all Cl was originally present as NaCl. syandi at its con¯uence with the Trisuli is de- meteoric water line (Galy, 1999) and imply Na in excess of Cl is de®ned as [Na*] ϭ [Na] livered from hydrothermal sources, a ®gure that the water in the hydrothermal system is Ϫ [Cl]. The hot springs near the Main Central consistent with both the Sr isotope budget and locally derived meteoric water. thrust have 8000±9000 ␮M [Na*], 6000± the major ion mass balances. 10 000 ␮M [K], and to 2200 ␮M [Si], all of As in a number of other Himalayan rivers, Spring Flux which re¯ect alteration of silicate minerals. Of the Marsyandi has Sr/Ca ratios higher than ex- Direct measurement of the overall water the total alkalinity in the springs, ϳ40% is pected from simple carbonate dissolution. The ¯ux from the Marsyandi springs is dif®cult composed of Na* and K. The balance is pre- causes of the excess Sr probably include precip- because they frequently do not ¯ow from dis- sent as Ca and Mg, which could be derived itation of secondary carbonate with low Sr/Ca crete point sources and diffuse ¯ow occurs in from dolomitic carbonates in the upper Lesser from supersaturated stream waters; however, the stream bed. Because chloride is anoma- Himalayan section. However, two observa- the overall importance of this process is not lously high in the spring waters, and behaves tions suggest that a signi®cant part of this re- known with certainty (Galy et al., 1999; En- conservatively, we can use the Cl mass bal- maining alkalinity (Ca, Mg) is also derived glish et al., 2000). The Marsyandi hot springs ance to estimate the water ¯ux from the from silicate alteration. First, formation I have very high Sr/Ca ratios (6.9±16.4 ␮mol/ springs. The average [Cl] of the tributaries is gneisses are rich in biotite, and biotite alter- mmol) and can elevate Sr/Ca in the river by 72 ␮M and that of the hot springs is 33 025 ation almost certainly provides some Mg to simple mixing. Distribution coef®cients for Sr ␮M; thus a spring ¯ux of slightly Ͻ1% yields the hydrothermal ¯uids. Second, the Ca/Mg between water and calcite vary from kSr ഠ the observed [Cl] of 366 ␮M at the Marsyan- ratio of the hot-spring ¯uids ranges from 3 to 0.03 to 0.08 over a temperature range of 40± di-Trisuli con¯uence. The discharge of the 15, whereas streams draining Lesser Himala- 200 ЊC (Holland et al., 1964; Humphrey and Marsyandi is 6.61 ϫ 109 m3/yr (Sharma, yan dolomites have a Ca/Mg ratio near 2. Howell, 1999; Malone and Baker, 1999). With 1997), yielding a water-¯ux estimate from the Some Ca in the hot springs is likely derived the stated kSr range, Sr/Ca in calcite precipi- springs within the High Himalayan Crystal- from alteration of oligoclase present in the tated from these hydrothermal ¯uids would lines of ϳ6 ϫ 107 m3/yr. The springs have a High Himalayan Crystallines. However, at range from 1.31 to 0.21 ␮mol/mmol. The Sr/ signi®cant impact on the chemistry of the present we cannot accurately quantify the frac- Ca in the remaining ¯uid would increase in

GEOLOGY, September 2001 805 proportion to the fraction of Ca precipitated. In developing an isotopic budget for the Sr processes in the Ganges-Brahmaputra basin The calculated precipitate values are very sim- ¯ux in Himalayan rivers, we argue that it is and the riverine alkalinity budget: Chemical Geology, v. 159, p. 31±60. ilar to Sr/Ca in hydrothermal calcite veinlets important to distinguish between dissolution Galy, A., France-Lanord, C., and Derry, L.A., 1999, at Nanga Parbat (Jacobson and Blum, 2000), of sedimentary marine carbonate units and hy- The strontium isotopic budget of Himalayan and as at Nanga Parbat, Sr in calcite formed drothermal calcite. Weathering of altered lime- rivers in Nepal and Bangladesh: Geochimica from Marsyandi hydrothermal ¯uids will be stones can be a signi®cant source of radiogen- et Cosmochimica Acta, v. 63, p. 1905±1925. Holland, H.D., Holland, H.J., and Munoz, J.L., radiogenic. ic Sr (English et al., 2000) but has no net 1964, The coprecipitation of cations with Silicate rocks of both the High Himalayan effect on the atmospheric CO budget. How- ϩ2 2 CaCO3ÐII. The coprecipitation of Sr with Crystallines and Lesser Himalaya, and Lesser ever, some part of the Sr and Ca in hydro- calcite between 90Њ and 100 ЊC: Geochimica Himalayan carbonates are all potential sources thermal calcite precipitated from the Marsyan- et Cosmochimica Acta, v. 28, p. 1287±1301. of radiogenic Sr in the hot springs. Silicate- di springs is derived from silicates. Although Humphrey, J.D., and Howell, R.P., 1999, Effect of differential stress on strontium partitioning in derived alkalinity and high Sr/Ca ratios in the weathering of such precipitates will yield the calcite: Journal of Sedimentary Research, hot-spring ¯uids are consistent with a silicate solute signature of carbonate weathering, it is v. 69, p. 208±215. source. Sr/Ca in formation I silicate rocks is actually just an intermediate step in introduc- Jacobson, A.D., and Blum, J.D., 2000, Ca/Sr and 87 86 ϳ11 ␮mol/mmol, generally similar to the hy- ing the products of silicate alteration by CO Sr/ Sr geochemistry of disseminated calcite 2 in Himalayan silicate rocks from Nanga Par- drothermal ¯uids. Lesser Himalayan carbon- into the river system. In terms of the global bat: In¯uence on river-water chemistry: Ge- ates in central Nepal have very low Sr con- cycles of Ca, Sr, and CO2, hydrothermal al- ology, v. 28, p. 463±466. tents with Sr/Ca ratios ϳ0.1 ␮mol/mmol teration in the Marsyandi springs can be Krishnaswami, S., Trivedi, J.R., Sarin, M.M., Ra- (Galy et al., 1999), making it unlikely that viewed as high-temperature silicate weather- mesh, R., and Sharma, K.K., 1992, Strontium simple carbonate dissolution or dissolution- ing. Geothermal activity is common along the isotopes and rubidium in the Ganga-Brahma- putra river system: Weathering in the Hima- precipitation could be the major source for the Himalayan front, and a better understanding laya, ¯uxes to the and contri- hydrothermal Sr. of hot-spring ¯uxes should help resolve some butions to the evolution of oceanic 87Sr/86Sr: of the current ambiguities in the solute and Earth and Planetary Science Letters, v. 109, isotopic budgets of Himalayan rivers, and p. 243±253. CONCLUSIONS AND IMPLICATIONS Malone, M.J., and Baker, P.A., 1999, Temperature The hot springs along the Marsyandi are a clarify the ways in which the process of oro- dependence of the strontium distribution co- major source of solutes and, despite compos- genesis interacts with the carbon cycle. ef®cient in calcite: An experimental study ing Յ1% of the water budget, deliver up to from 40 degrees to 200 degrees C and appli- 30% of the river's Sr ¯ux. Active tectonic pro- ACKNOWLEDGMENTS cation to natural diagenetic calcites: Journal of We thank Jay Quade, Nathan English, and Page Sedimentary Research, v. 69, p. 216±223. cesses beneath the Main Central thrust provide Chamberlain for thoughtful reviews. Palmer, M.R., and Edmond, J.M., 1989, The stron- both the heat and some of the CO2 to these tium isotope budget of the modern ocean: systems, and the hot-spring water and solute REFERENCES CITED Earth and Planetary Science Letters, v. 92, ¯uxes directly re¯ect current tectonic activity. Blum, J.D., Gazis, C.A., Jacobson, A.D., and Cham- p. 11±26. berlain, C.P., 1998, Carbonate versus silicate Quade, J., Roe, L., DeCelles, P.G., and Ojha, T.P., Hydrothermal interaction is probably common 87 86 weathering in the Raikhot watershed within 1997, The late Neogene Sr/ Sr record of in collisional orogens, driven by high uplift the High Himalayan Crystalline series: Geol- lowland Himalayan rivers: Science, v. 276, and denudation rates (e.g., Templeton et al., ogy, v. 26, p. 411±414. p. 1828±1831. 1998). Colchen, M., Le Fort, P., and PeÃcher, A., 1980, Carte Sharma, C.K., 1997, A treatise on water resources of geÂologique Annapurna-Manaslu-Ganesh, Him- Nepal: Kathmandu, Sangeeta Sharma, 493 p. The hydrothermal ¯uids contain a signi®- Shieh, Y.N., and Taylor, H.P., 1969, Oxygen and cant component of CO derived from decar- alaya du NeÂpal: Paris, Centre National de la 2 Recherche Scienti®que, scale 1:200 000. carbon isotope studies of contact metamor- bonation reactions and are therefore a source Derry, L.A., and France-Lanord, C., 1996, Neogene phism of carbonate rocks: Journal of Petrolo- gy, v. 10, p. 307±331. of CO2 in the global carbon cycle. However, Himalayan weathering history and river 87Sr/ the radiogenic Sr and high silicate-derived al- 86Sr: Impact on the marine Sr record: Earth Singh, S.K., Trivedi, J.R., Pande, K., Ramesh, R., and Krishnaswami, S., 1998, Chemical and kalinities in the spring ¯uids imply that hy- and Planetary Science Letters, v. 142, p. 59±74. strontium, oxygen, and carbon isotopic com- drothermal alteration of silicate rocks have Edmond, J.M., 1992, Himalayan tectonics, weath- positions of carbonates from the Lesser Him- alaya: Implications to the strontium isotope neutralized a fraction of this CO2 ¯ux. The ering processes, and the strontium isotope re- alteration of silicates from the High Himala- cord in marine limestones: Science, v. 258, composition of the source waters of the Gan- ga, , and the Indus Rivers: Geochim- p. 1594±1597. yan Crystallines at moderate hydrothermal ica et Cosmochimica Acta, v. 62, p. 743±755. English, N.B., Quade, J., and DeCelles, P.G., 2000, temperatures is analogous to low-temperature Templeton, A.S., Chamberlain, C.P., Koons, P.O., Geologic control of Sr and major element and Craw, D., 1998, Stable isotopic evidence silicate weathering reactions in that CO2 is chemistry in Himalayan rivers, Nepal: Geo- for mixing between metamorphic ¯uids and neutralized by reactions with silicate-derived chimica et Cosmochimica Acta, v. 64, surface-derived waters during recent uplift of p. 2549±2566. base cations. Whereas much of the alkalinity the Southern Alps, New Zealand: Earth and Friedman, I., and O'Neil, J.R., 1977, Compilation of in the Marsyandi is derived from carbonate Planetary Science Letters, v. 154, p. 73±92. stable isotope fractionation factors of geochem- Upreti, B.N., 1999, An overview of the stratigraphy weathering (Galy and France-Lanord, 1999), a ical interest, in Fleischer, M., ed., Data of geo- signi®cant proportion of the hot-spring alka- and tectonics of the Nepal Himalaya: Journal chemistry: U.S. Geological Survey, v. 440-KK, of Asian Earth Sciences, v. 17, p. 577±606. linity is derived from silicate sources (Ն40% p. 1±114. in the hot springs near the Main Central Galy, A., 1999, Etude geochimique de l'erosion ac- Manuscript received January 2, 2001 tuelle de la chaine himalayenne [Ph.D. thesis]: thrust). Thus the hot springs are an important Revised manuscript received April 23, 2001 Nancy, France, Institut National Polytechnique Manuscript accepted May 11, 2001 source of silicate-derived alkalinity to the de Lorraine, 464 p. river. Galy, A., and France-Lanord, C., 1999, Weathering Printed in USA

806 GEOLOGY, September 2001