Linkages between the Madden Julian Oscillation, process-level diagnostics and GCM parameterization behavior in YOTC simulations

Abstract:

The simulation of the Madden Julian Oscillation (MJO) remains a significant challenge in climate models. The primary difficulty lies in relating MJO skill to parameterized physical processes - the main access point for model development. One theory of the MJO relies on scale-interactions from small to large scales. The expectation is that GCMs should reproduce the correct relationships at the smallest resolved scales and this will translate through increasing scales and lead to a skillful simulation of the MJO. So-called 'process-based' diagnostics have recently been applied to simple model fields in order to relate accurate simulation of the MJO to accurate, small-scale process-level relationships (Kim et al., 2014). In this presentation we will take this technique further to provide greater insight into how the underlying physical parameterizations in the Community Atmosphere Model (CAM) conspire to provide the process-level responses in the model, particularly as it relates to precipitation and humidity dependent processes. This provides the potential for a range of dependencies between parameterization tendencies and MJO skill. Furthermore, these dependencies are examined to quantify the effect of model biases. This entails performing the same process-level analysis on simply initialized and nudged CAM simulations that make use of YOTC analysis. These techniques enable diagnosis of the relationship between degrading model simulation (basic state and MJO) and changes in the parameterized response at the process level. In summary, this talk will show the most promising relationships between MJO simulation performance and the fidelity with which the parameterized physics produce observed process-scale relationships.