Improved historical simulation by enhancing moist physical parameterizations in the climate system model NESM3.0

Abstract

The third version of the Nanjing University of Information Science and Technology (NUIST) Earth System Model (NESM3.0) has been recently developed for sub-seasonal to seasonal climate prediction and projection of future climate change. This requires realistic simulation of both internal modes of climate variability and global energy balance and climate sensitivity. In the historical experiments based on the coupled model intercomparison project (CMIP6) forcings, the original version of NESM3.0 does a reasonably good job in capturing warming trends and climate sensitivity to external forcing, but not in simulating climatology and climate variability. For our project we modified the deep and shallow convective scheme, cloud cover, and cloud microphysics in the atmospheric model. We then conducted hundreds of experiments to test the results in the fully coupled model with comprehensive metrics to ensure that any individual targeted improvement does not affect the overall performance. The modifications in moist physics improved the model’s climatology and internal variability significantly without degrading global energy balance and climate sensitivity. The key was to reduce the model’s warm sea surface temperature (SST) bias and associated excessive precipitation bias in the tropics by reducing convective precipitation and increasing large-scale stable precipitation. This effort leads to more realistic simulation of the zonal mean circulation and temperature structure, global monsoon, and ocean salinity. The El Nino Southern Oscillation (ENSO) simulation was improved by reducing wind stress biases associated with ENSO in the central and eastern Pacific. Better ENSO leads to better teleconnection in mid-latitudes, particularly over the North Pacific and Atlantic Ocean. The eastward propagation of the Madden–Julian Oscillation (MJO) was significantly improved by enhancing the interaction between the boundary layer and lower tropospheric heating. These results suggest that improvement in moist physical parameterizations is an effective way to improve simulation of climatology and major modes of internal variability without degrading the global energy balance and climate sensitivity in the historical run.