Formation processes of shallow ground ice in permafrost in the Northeastern Qinghai-Tibet Plateau: A stable isotope perspective

Document Type


Publication Date



College of Forest Resources and Environmental Science


The Source Area of the Yellow River (SAYR) on the Northeastern Qinghai-Tibet Plateau (QTP) stores substantial amounts of ground ice, which plays a significant role in understanding the hydrological processes and past permafrost evolution on the QTP. However, little is known about the initial sources and controlling factors of the ground ice in the SAYR. In this study, for the first time, ground ice stable isotope data (δ18O, δD, and d-excess) are presented, along with cryostratigraphic information for nine sites is integrated into three cryostratigraphic units (palsa, thermo-gully, and lake-affected sites) in the central SAYR. The ground ice in the nine sites exhibited diverse structures, ice contents, and stable isotopes due to differences in the initial water sources, ice formation mechanisms, soil types, and climate conditions. All of the freezing lines of ground ice are below those of the precipitation, streams, and lakes in most cases, suggesting the freezing of liquid water. The near-surface ground ice (NSGI) originated from precipitation, active layer water, and precipitation-fed springs. The NSGI was formed by quick freezing at the thermo-gully site (TG-1). In contrast, the formation of the NSGI at the palsa site (Palsa-1) experienced a slow segregation process during the permafrost aggradation. The NSGI was formed by quick freezing at the lake-affected sites under colder climate conditions. Conversely, the deep-layer ground ice (DLGI) at the lake-affected sites was fed by isotopically negative water and lake water occurred during a colder climate period. The DLGI at the TG-1 and Palsa-1 formed via similarly slow segregation of supra-permafrost water (mixed with precipitation), but had opposite water migration directions. The stable isotope compositions of the DLGI at the lake-affected sites became gradually more positive with decreasing distance from WL Lake, emphasizing the large influence of the lake changes on the growth of ice. The two end-member mixing model estimated that the contributions of paleo-lake water to the DLGI ranged from 9.8 % to 63.4 % towards the lake at the lake-affected sites, while the meltwater from past permafrost/ground ice contributed 36.6–90.2 % of the total input. A conceptual diagram of the δ18O trajectories of ground ice was constructed, the water migration patterns and ground ice formation processes between the palsa, thermo-gully, and lake-affected sites were clarified. The results of this study emphasize the influence of lake changes and past permafrost evolution on ground ice growth and improve our understanding of permafrost changes on the QTP.

Publication Title

Science of the Total Environment