Evaluation and uncertainty analysis of the land surface hydrology in LS3MIP models over China

The attached is the dataset assoicated with the paper titled "Evaluation and uncertainty analysis of the land surface hydrology in LS3MIP models over China" which was submitted to Journal of Earth and Space Science.
The Land Surface, Snow and Soil moisture Model Intercomparison Project (LS3MIP) offers valuable land surface hydrology products from the land modules of current Earth System Models (ESMs). In this paper, historical LS3MIP hydrological variables including precipitation (PR), evapotranspiration (ET), soil moisture (SM), total runoff (Ro), and snow cover fraction (SCF) were extensively evaluated with various high-quality reference datasets over Chinese mainland. The six ESMs in LS3MIP were driven by four meteorological forcing datasets. The results indicated that the LS3MIP multi-model means (MMEs) of most variables are underestimated overall, while they show high spatial consistency in term of linear trends, with the percentage area ranging 56% ~ 85% between simulations and reference datasets. After computing and ranking multi statistical metrics (bias, correlation coefficient, normalized standard deviation, and unbiased root-mean-square biases), it is found that the CESM2 model produces the best performance of land surface hydrological variables, while as the meteorological forcing dataset GSWP3 exhibits the highest quality. Furthermore, the analysis of variance method (ANOVA) was then used to trace sources of the uncertainty of the LS3MIP hydrological variables for 1900–2012 (1948–2012 for Ro). In ANOVA, the simulation uncertainties may be decomposed into three sources: model, atmospheric forcing datasets and their interactions. In LS3MIP historical hydrological variables over China, model uncertainty is the dominant factor overall although it shows regional differences, and the dependence of uncertainty on the model differs among hydrological regimes. This highlights the urgent requirements to improve the land surface model representation in future research.