CR 67:99-115 (2016)  -  DOI: https://doi.org/10.3354/cr01366

ABSTRACT: The New South Wales (NSW)/Australian Capital Territory (ACT) Regional Climate Modelling (NARCliM) project aims to deliver robust climate change projections for southeast Australia at a scale relevant for decision-making. In the first phase of the project, the Weather Research and Forecasting (WRF) model with 3 physics scheme combinations, driven by NCEP/NCAR reanalysis dataset as ‘perfect’ boundary conditions, was run for a 60 yr period from 1950-2009 to assess the model’s ability to simulate regional climate for southeast Australia. In this study, model results for daily precipitation and maximum and minimum temperatures were compared to gridded observations from the Australian Water Availability Project (AWAP) to evaluate model performance at varying time scales using a number of statistical metrics. Results show that all simulations have good representation of daily, monthly, seasonal, annual, multiannual and decadal variation in precipitation and temperature. However, there is a bias in precipitation in the northwest part of the domain (25-100%) and along the Great Dividing Range (75-150%). The temperatures are systematically underestimated across the domain (2-3°C for maximum temperature and 1-2°C for minimum temperature), suggesting the need for bias correction. The evaluation results indicate that the cumulus scheme is critical to precipitation simulation, and planetary boundary layer and radiation schemes are more important in temperature simulations. The findings from this study give us confidence in the WRF model for long-term regional climate modelling for southeast Australia. They also provide guidance in the parameterisation of the WRF model in providing more reliable precipitation and/or temperature projections.

KEY WORDS: NARCliM · Weather Research and Forecasting · WRF · Precipitation · Maximum and minimum temperatures