Variations in the suspended sediment on the Qinghai-Tibet Plateau have important implications for aquatic ecosystems. Although changes in the cryosphere induced by climate change have been shown to increase sedi-ment yields, their impacts on water and sediment dynamics in headwater regions remain poorly investigated. Here, we examined the responses of runoff and suspended sediment dynamics to changes in the climate and ground freeze-thaw cycle in the source region of the Yangtze River (SRYR) from 1964 to 2019. Long-term daily in situ water and sediment observations provided evidence that climate change controlled change in seasonal and annual water-sediment dynamics by regulating air temperature and precipitation. Attribution analysis showed that precipitation (-41.93 %, through driving rainfall splash, overland flow erosion, and mass wasting) and land surface temperature (-30.66 %, through driving freeze-thaw erosion) were the major factors contributing to increasing fluvial sediment fluxes over the past 30 years. We found that freeze-thaw cycles changed the soil erosion patterns by governing the thermal state of the near-surface active layer and driving associated thermal processes. Furthermore, the extension of the thawing duration and the advance of the thawing starting date (at an average rate of 13.5 days/10 yr) exacerbated freeze-thaw erosion, leading to elevated sediment fluxes in the initial thaw and initial freezing periods. This study highlights the need to focus on cryosphere-hydrology ob-servations in terms of sediment dynamics; these findings are critical for soil and ecological protection in alpine headwater regions.

Borehole-measured soil temperatures have been routinely used to calibrate soil parameters in permafrost modeling, although they are sparse in the Qinghai-Tibet Plateau (QTP). A feasible alternative is to calibrate models using land surface temperatures. However, the quantitative impacts of various soil parameterizations on permafrost modeling remain unexplored. To quantify these impacts, two sets of soil parameters (denoted as Psoil and Psurf) were obtained via calibration using borehole temperature measurement and ERA5-Land (the land component of the fifth generation of European Re-Analysis) skin temperature, respectively, and applied to the Geophysical Institute Permafrost Laboratory Version 2 (GIPL 2.0) model. Comparing against the borehole -measured soil temperatures of 4 soil layers, the ERA5-Psurf simulation (with Root Mean Squared Error, i.e., RMSE from 1.4 degrees C to 3.9 degrees C) outperform ERA5-Psoil simulation (RMSE from 1.4 degrees C to 3.9 degrees C) during 2006-2014. The obtained Psoil and Psurf were then utilized as soil parameters in GIPL 2.0 to model permafrost dynamics for a long period from 1983 to 2019, respectively, using ERA5-Land as forcing data. Simulations revealed significant disparities. In comparison to the simulation using Psurf results using Psoil show that the mean annual soil tem-perature at 1 m depth was 2.72 degrees C lower with a 0.01 degrees C/a (50.0%) lower trend; the active layer thickness was 0.81 m (35.7%) less with a 2.16 cm/a (82.1%) lower trend; the duration of the thawing season at 1 m depth was underestimated by about one month, and the zero-curtain period was about 23 days (37.7%) shorter. The change rates of the zero-curtain period, however, were comparable. This study implies that choosing soil parameteri-zations is critical for model evaluation against observations and long-term model prediction.