On discrimination between carbonate and silicate inputs to Himalayan rivers
Résumé
We review new and published analyses of river waters, bedloads and
their constituent minerals from the Dhauli Ganga and Alaknanda, headwaters of the
Ganges in Garhwal, and the Marsyandi in Nepal and their tributaries. These data are
used to discriminate between the inputs of major cations and Sr from silicate and
carbonate sources. Methods of estimating the proportion of the carbonate and silicate
inputs to river waters using mixing arrays in Sr-Ca-Mg-Na-K-87
Sr/86 Sr space are shown
to suffer from systematic correlations between the magnitude of the precipitation of
secondary calcite and the fraction of the silicate component. This results in factor-of-
two overestimates of the fractions of silicate-derived Ca, Mg and Sr. To correct for this
the magnitude of secondary calcite precipitated and relative fractions of silicate and
carbonate-derived cations are instead calculated by modeling the displacement of
water compositions from the compositions of the carbonate and silicate components
of the bedload in subsets of Sr-Ca-Mg-Na-K-87 Sr/86 Sr space. The compositions of the
carbonate and silicate end-members in the bedload are determined by sequential
leaching. The results of this modeling are compared with modeling of the modal
mineral inputs to waters where mineral compositions are derived from electron-
microprobe analyses of the minerals in the bedload. In the upper Marsyandi catch-
ment, which drains low-grade Tethyan Sedimentary Series formations, a set of mainstem
samples collected over a two-year period define tight correlations in Sr-Ca-Mg-Na-K-
87
Sr/86
Sr space. Modeling of the magnitude of secondary carbonate precipitation and
fractions of silicate-derived Ca, Mg and Sr in Sr-Ca-Mg-87 Sr/86 Sr space gives self-
consistent results that are compatible with both the calculations of mineral modes and
published Mg-isotopic compositions, if the ratio of chlorite to biotite weathering is
high or if there is another silicate source of Mg. These calculations imply that between
12 and 31 percent of the Sr and 44 and 72 percent of the Mg is derived from silicate
minerals where the range reflects the seasonal change in the ratio of silicate-derived to
carbonate-derived cations. Modeling in Sr-Ca-Na and/or K space is inconsistent with
the Sr-isotopic and Mg-isotopic constraints and we conclude that in this catchment
dissolution of Na and K are incongruent relative to Sr-Ca-Mg. Potassium is preferen-
tially retained in micas whereas the controls on Na are unclear. Modeling of the
catchments underlain by High Himalayan Crystalline and Lesser Himalayan Series in
Garhwal is complicated by the presence of dolomite as well as calcite in the carbonate
and the results imply that dolomite dissolves faster in the acetic acid leaches than in
nature. Up to 60 percent of the Sr in the catchment on High Himalayan Crystalline
Series and 20 to 30 percent of Sr in the catchments on Lesser Himalayan Series are
estimated to be derived from silicates. However it should be noted that the element
budgets are not all self-consistent and the use of bedrock-element ratios to model the
sources of chemical inputs to river waters remains subject to uncertainties.