Initial Assessment of the Vertical Resolution Sensitivity of the ACME Atmosphere Model

Abstract

We have constructed several vertical grids for the ACME model to assess the model behavior with higher vertical resolution, thinner surface layer, and higher model top. We find that some evidence for model sensitivity to the vertical grids. Some features  of the changed resolution (with no tuning changes) are common in all model configurations examined (e.g. the different convection schemes) while others differ from configuration to configuration. For example, increasing vertical resolution reduces clouds (as well as the cloud forcing) in the tropics and increases global mean precipitation with all three convection configurations, while some longstanding climate model biases such as the southern ocean surface wind stress and double ITCZ appear to be reduced by thinning the surface layer and increasing the vertical resolution with the CAM5 and UNICON schemes, but not the CLUBB scheme. The 72-level model (with a higher model top) further improves the inter-annual variability while produces the same time averaged climate as the 64-level model, and will be used for future model testing, tuning, and evaluation.