Composite Hydrological Recharge Sources of Calcium Chloride-Type Brine in Dongling Lake, Northern Qaidam Basin: Evidence from Hydrochemistry, Multi-Isotopes, and Dynamic Monitoring

Abstract

Dongling Lake, located on the northern margin of Qarhan Salt Lake (QSL), is a representative calcium chloride-type surface brine lake in the northern Qaidam Basin. Hydrochemical data, H-O-B-Sr isotopes, water-soluble experiments on borehole sediments, and surface water-groundwater monitoring were integrated with a recharge-runoff-discharge (RRD) model to constrain whether modern terminal infiltration contributes to the Dongling Lake brine system in addition to deep Ca-Cl recharge. Dongling Lake brine is characterized by high salinity and a Ca-Cl hydrochemical facies, with TDS ranging from 148.49 to 534.12 g/L; the two measured surface brine samples have the highest TDS and contain higher Li and B concentrations than most intercrystalline brines. H-O isotopes indicate an atmospheric-precipitation and snowmelt-related water-source background, whereas high δ¹¹B values in sediment water-soluble phases, elevated Sr contents, and ⁸⁷Sr/⁸⁶Sr ratios indicate that the solute budget is additionally affected by formation-water dissolution, water-rock interaction, and deep fault-related fluids. Dynamic monitoring and RRD modeling suggest a seasonal hydrological connection between Dongling Lake and the terminal Golmud River: during the flood season, 22%-28% of surface water is transformed into groundwater, of which 18%-22% may further infiltrate toward Dongling Lake; during the dry season, the conversion ratio decreases to 5%-8% and lateral groundwater recharge becomes dominant. Model validation gives R² = 0.81-0.86 and NSE = 0.72-0.78; relative errors for major-ion and isotope fitting are generally <5% after isotope-coupled calibration. These percentages should therefore be interpreted as model-constrained ranges rather than fixed values. Because the surface-brine dataset is small (n = 2), isotope coverage is limited to selected boreholes, and the RRD framework simplifies a complex fault-controlled system into key endmembers, the recharge partitioning and Li-B transport interpretation remain constrained but uncertain.