Spatial and Temporal Scales of Precipitation Variability and their Relation to MJO Representation in CMIP6 Models

Abstract:

Biases in mean-state moisture distributions have emerged as one reason for poor simulation of the Madden-Julian Oscillation (MJO) in general circulation models (GCMs), since these biases cause errors in the anomalous moisture advection by the MJO winds that drives eastward MJO propagation. GCMs sometimes also poorly represent the observed scales of precipitation, particularly in the tropics where simulated daily rainfall is too light, too frequent and too persistent. We seek to understand whether process-level biases affect the MJO more through their direct interaction with the disturbance, or more through their indirect effect on the background state. Specifically, we investigate how tropical air-sea feedbacks modify the precipitation organization and intensity in the GCMs participating in the Sixth Coupled Model Intercomparison Project (CMIP6). We apply a set of process-oriented diagnostics to analyse scales of spatial and temporal variability in precipitation and relate them to the fidelity of the simulated MJO intensity and propagation. The metrics are used to evaluate model performance relative to satellite-derived observations and reanalyses. Diagnosing the processes responsible for related biases, including SST-surface-flux-convection phase relationships, column moistening, surface pressure east of MJO convection, SST gradients and low-level convergence, will help to pinpoint the improvements most likely to advance MJO simulation.