[1] The augmented Noah land surface model described in the first part of the two-part series was evaluated here over global river basins. Across various climate zones, global-scale tests can reveal a model's weaknesses and strengths that a local-scale testing cannot. In addition, global-scale tests are more challenging than local- and catchment-scale tests. Given constant model parameters (e. g., runoff parameters) across global river basins, global-scale tests are more stringent. We assessed model performance against various satellite and ground-based observations over global river basins through six experiments that mimic a transition from the original Noah LSM to the fully augmented version. The model shows transitional improvements in modeling runoff, soil moisture, snow, and skin temperature, despite considerable increase in computational time by the fully augmented Noah-MP version compared to the original Noah LSM. The dynamic vegetation model favorably captures seasonal and spatial variability of leaf area index and green vegetation fraction. We also conducted 36 ensemble experiments with 36 combinations of optional schemes for runoff, leaf dynamics, stomatal resistance, and the β factor. Runoff schemes play a dominant and different role in controlling soil moisture and its relationship with evapotranspiration compared to ecological processes such as the β factor, vegetation dynamics, and stomatal resistance. The 36-member ensemble mean of runoff performs better than any single member over the world's 50 largest river basins, suggesting a great potential of land-based ensemble simulations for climate prediction.

The community Noah land surface model with multiparameterization options (Noah-MP): 2. Evaluation over global river basins