Dynamical downscaling for the southwest of Western Australia using the WRF modelling system

Abstract

The southwest of Western Australia (SWWA) is a region of significant cereal production, with the main crops being winter grown wheat and barley. The most important factors influencing wheat growth and production are temperature extremes and precipitation, and hence, it is critical to have an understanding of how these environmental factors have changed in the past, and how they are likely to change in the future. One method of addressing this important research question is by using regional climate models (RCMs) to dynamically downscale re-analysis products and/or output form Global Circulation Models to a fine resolution. One tool which is being increasingly used for this purpose is the Weather Research and Forecasting Model (WRF) Advanced Research (ARW). However, like any modeling system, WRF-ARW requires thorough testing before it is implemented to carry out long-term climate runs. This paper examines the influence of different input data sources, as well as model physics options on simulated precipitation and maximum and minimum temperatures in SWWA by comparing the simulations against an observational gridded dataset. It is found that running WRF3.3 with the 1.0 × 1.0 degree National Center for Environmental Prediction Final analysis (NCEP-FNL), as compared to the 2.5 × 2.5 degree NCEP / National Center for Atmospheric Research (NCEP/NCAR or NNRP) results in much improved simulations of precipitation and temperatures. Using the National Oceanic and Atmospheric Administration 1.0 × 1.0 degree resolution sea surface temperature (SST) dataset does not result in markedly different results as compared to using the NNRP surface skin temperatures as SSTs. Using the Betts-Miller-Jajic (BMJ) scheme for cumulus/convection parameterisation rather than the more widely used Kain Fritsch (KF) scheme results in slightly higher errors for precipitation, and no marked change in temperatures. The latest version of the Rapid Radiative Transfer Model (RRTMG) is found to result in improved simulations of maximum and minimum temperatures, as compared to the RRTM, Community Atmosphere Model (CAM) 3.0, and Dudhia schemes. Use of the Asymmetric Convective Model as the planetary boundary-layer scheme rather than the more widely used Yonsei University scheme results in over-prediction of maximum and minimum temperatures.