loading page

Evaluation of soil moisture in CMIP6 multi-model simulations over conterminous China
  • +1
  • Aihui Wang,
  • Xianghui Kong,
  • Yue Chen,
  • Xin Ma
Aihui Wang
Institute of Atmospheric Physics

Corresponding Author:wangaihui@mail.iap.ac.cn

Author Profile
Xianghui Kong
Nansen-Zhu International Research Centre, Institute of Atmospheric Physics
Author Profile
Yue Chen
Lanzhou University
Author Profile
Xin Ma
Institute of atmospheric Physics, Chinese Academy of Sciences
Author Profile


The Sixth Phase of the Coupled Model Intercomparison Project (CMIP6) provides the long-term soil moisture (SM) products and this study conducts a comprehensive assessment of SM products of multiple CMIP6 model simulations over conterminous China. Both near-surface (0-10 cm) SM simulations from 40 models and root-zone (0-100 cm) SM from 25 models are compared with a set of station measurements in the growing season (April to September) for 1992-2013 in term of magnitude, spatial and temporal variability, and the long-term trend and interannual variability of near-surface SM for 1961-2014 are further evaluated with an offline land surface modeling dataset. Simulations from most models broadly capture the spatial characteristics of observation and the multi-model mean (MME) well reproduces seasonal variations over majority regions regardless of large-spread across models. Models from the same institution likely manifest similar performances and the land surface scheme plays a dominant role in the SM reproduction. The majority of models well simulate the overall drying trend in China as a whole and the signs of SM trend are highly consistent across models, but the areas with significant wetting/drying trends vary with models. The spatial patterns of SM interannual variability are model-dependent in term of spatial patterns. MME is overall superior to the simulations of individual model and may have potential applications in the future research. The heterogeneity SM performances across models reveal the complexity in modeling land surface variables, suggesting the need for improving representations of land surface processes in the coupled models.