Verification of a high-resolution mesoscale NWP system

Accurate prediction of low-level winds is an important research area for wind farm forecasting applications. The Centre for Australian Weather and Climate Research has extended its existing numerical weather prediction (NWP) capabilities to develop and implement a mesoscale assimilation and prediction system (named WLAPS for research purposes) which is shown to be all improvement on the Bureau's existing mesoscale model, mesoLAPS (denoted MLAPS). The usefulness of WLAPS as a tool for wind forecasting stems mainly from the fact that it has its own 10 kin resolution data assimilation scheme, in contrast to earlier Bureau of Meteorology mesoscale models which were initialised by interpolation from 37.5 km assimilation analyses. WLAPS performs well against several verification criteria, including: verification scores of gridded output that compare forecasts to model analyses, observation statistics which verify the forecast and analysis against observations, and event-based case studies. Where relevant forecasts are available, the performance of WLAPS is compared with that of its similar to 12.5 km resolution predecessor MLAPS. As part of the WLAPS project, the model has also been extensively verified against observational data from 53 hub height (40-80 m) wind towers across southern Australia. After bias corrections, WLAPS is shown to have a root mean square error (RMSE) of around 2 in s(-1) (depending on site type and season) for 1 to 12-hour forecasts, with bias close to zero. The inclusion of direction and stability information in the bias correction procedure is shown to have a positive impact on the forecasts. Verification against low-level winds is suggested as a useful criterion for mesoscale verification, since we expect the full range of scales of atmospheric motion to be present in low-level wind observations.