Abstract:
Weather radar observations provide one of the most spatially and temporarily complete high resolution datasets describing atmospheric state at sub-kilometre resolutions. New Zealand’s national radar network infers both rainfall intensity and wind speed by measuring the reflectivity and Doppler velocity signals of falling rain. There is significant interest amongst the forecasting community in making optimal use of such weather radar observations however to date little progress had been made in New Zealand. The process of including radar observations in weather prediction models through a process known as data assimilation is complicated by their complex and non-trivial relationship with the model variables. This thesis makes progress in the better specification of a Doppler velocity operator for the open source Weather Research and Forecasting data assimilation system (WRFDA). The operator better accounts for the physical properties of the radar system and is demonstrated to improve the data assimilation process in a case study with a large number of members. Radar Doppler velocity assimilation is also tested in two higher resolution studies. The first is for the passage of tropical cyclone Wilma in 2011 where improvements in the vortex structure and track forecasts are realised. Secondly a pair of experiments is carried out to test a high resolution ensemble forecasting system for a small catchment and the resulting rainfall predictions are verified against observations from the Atmospheric Physics Group’s high resolution radar.