Evaluation of CMIP6 Models in Simulating Australian Monsoon Over Northern Australia

This study provides a comprehensive evaluation of the latest generation of climate models in simulating the dynamics of the Australian summer monsoon over northern Australia. The analysis focuses on both spatial and temporal characteristics of precipitation, low-level circulation, monsoon onset and retreat as well as the representation of El-Niño Southern Oscillations (ENSO) teleconnections. Using multiple observational datasets and standard performance metrics, including correlation, root mean square error, standard deviation and bias, models are assessed and ranked based on their ability to simulate observed monsoon features. The results reveal persistent model biases, including overestimated precipitation, underestimated wind intensity, delayed monsoon onset and retreat (by up to 2 weeks in some models), a weak or overly uniform simulation of ENSO influence and in some cases interannual rainfall variability is more than double the observed value. These limitations affect the models' ability to capture the seasonal cycle and interannual variability of the monsoon. Importantly, performance varies significantly even among models from the same modelling institution, underscoring the need for individual model evaluation. Using a subset of high-performing models that accurately capture key monsoon features leads to more consistent and robust projections. These high-performing models simulate increased rainfall in northeast Australia (+15 to +20%) by the late 21st century under a high emissions scenario. In contrast, low-performing models simulate much weaker increases (0 to +5%). This study highlights both the progress and ongoing challenges in current-generation climate models, contributing to improved understanding and model selection for future monsoon projections across northern Australia.

Keywords: Australian monsoon | climate models | future changes | model evaluation