The source region of the Yellow River (SRYR) is located at the edge of the Qinghai-Tibet Plateau (QTP), which is completely covered by frozen ground. Due to relatively higher temperatures, the frozen ground in the SRYR is particularly fragile and susceptible to the impacts of global climate change. This study discusses the maximum freeze depth (MFD) of frozen ground in the SRYR, including analysis of measured data at the stations, comparison of simulation models, and projection of future changes. The MFD of frozen ground recorded at nine meteorological stations within the SRYR ranged from a few tens of centimeters to more than two meters. The decreasing trend of MFD was recorded except for a few stations from 1997 to 2017, with a maximum rate of -22.8 cm/10a. The decreasing rate of MFD for the whole SRYR from 1997 to 2017 is -10.8 cm/10a. Furthermore, we assessed the performance of three simulation methods: Stefan equation, multiple linear regression, and BP neural network predicting the MFD using the measured data. The Stefan equation exhibited limited accuracy in simulating the MFD, while the BP neural network demonstrated remarkable performance, with a correlation coefficient R of 0.949. In addition, we evaluated the applicability of different global climate models (GCMs) in the SRYR, identified the optimal model, and combined it with the BP neural network model to predict future MFD change. Among the five climate models, the BCC-CSM2-MR model and ensemble model fit the measured precipitation and air temperature well. The projected results based on the BCC-CSM2-MR model and ensemble model indicate that the MFD of different stations in the SRYR and the whole region will still tend to decrease in the future. Our results contribute to understanding the response of cold region frozen ground to climate change and provide available data.
Keywords: BP neural network; CMIP6 climate model; Climate change; Maximum freeze depth; SRYR.
Copyright © 2023 Elsevier B.V. All rights reserved.