Rainfall in Queensland: Part 4: the ability of HiGEM to simulate Queensland's rainfall variability and its drivers

Stakeholders and policymakers are seeking detailed information on the impacts of climate change and variability, especially on rainfall, at the local and regional levels. This information can be delivered by only those global climate models that represent the atmosphere and ocean at fine resolution, to robustly simulate the weather systems that produce rainfall. These models provide us with a better understanding of the key meteorological phenomena that affect Queensland. That knowledge can then be used to improve the simulation of these phenomena in lowerresolution climate models. Implementing these improvements will provide more accurate predictions on weekly to seasonal and decadal timescales, as well as more robust predictions of the impacts of climate change on these phenomena. High-resolution Global Environment Model, version 1.1 (HiGEM) is a global, coupled climate model that was developed by the U.K. academic community. It is based on the U.K. Hadley Centre's HadGEM1 model, but HiGEM has considerably higher resolution: 90 km in the atmosphere and 30 km in the ocean. HiGEM has been used in this research as its increased resolution may allow the model to better represent regional climate variability and change in Queensland. In this research, a 150 year control simulation of HiGEM was assessed to evalute the ability of the model to simulate Queensland's rainfall and its inter-annual and decadal variability. HiGEM was also assessed for its ability to produce the observed Empirical Orthogonal Teleconnection (EOT) patterns of rainfall avariability obtained from the SILO gridded rainfall dataset. In the mean, HiGEM produces less rainfall over Queensland than observed, particularly in the north of the state. Most of this dry bias occurs because the model simulates a weaker Australian summer monsoon than is observed. However, HiGEM represents well the relationship between the El Niño Southern Oscillation (ENSO) and Queensland rainfall, on annual and seasonal timescales. The model even captured the observed asymmetric correlation between the ENSO and Queensland rainfall: stronger La Niña events cause stronger flood years in Queensland, but stronger El Niño events do not cause stronger droughts. The research found that HiGEM lacks the ability to model decadal variations in Queensland rainfall and in the teleconnection between the ENSO and rainfall. This is likely due to the model's inability to simulate the Interdecadal Pacific Oscillation (IPO), which has been identified as the key driver of these variations. In relation to the generation of tropical cyclones, HiGEM captures the observed regions of tropical-cyclone formation and the correct distributions of tropical cyclone tracks, but simulates too many tropical cyclones in the Southwest Pacific. When EOT analysis is applied to HiGEM and the results are compared with the EOT patterns computed using observed rainfall, HiGEM performs well for those EOTs related to the ENSO in summer, winter and spring. HiGEM also represents the relationship between Southeast Queensland rainfall and onshore easterly winds, including the decadal variations in the winds' strength and moisture content. Futher, HiGEM correctly simulates the observed association between the frequency of tropical cyclones and summer rainfall in Cape York. The success of HiGEM at reproducing many of the observed EOTs, particularly in summer, increases our confidence in the model's ability to predict the impact of climate change on Queensland's rainfall and its drivers