A DRP-4DVar data assimilation scheme for typhoon initialization using sea level pressure data

In this study, a new data assimilation system based on a dimension-reduced projection (DRP) technique was developed for the fifth-generation Pennsylvania State University-NCAR Mesoscale Model (MM5) modeling system. As an initial step to test the newly developed system, observing system simulation experiments (OSSEs) were conducted using a simulated sea level pressure (SLP) field as "observations" and assimilation experiments using a specified SLP field to evaluate the effects of the new DRP-four-dimensional variational data assimilation (4DVar) method, initialization, and simulation of a tropical storm-Typhoon Bilis (2006) over the western North Pacific. In the OSSEs, the "nature" run, which was assumed to represent the "true" atmosphere, was simulated by the MM5 model, which was initialized with the 1.0° × 1.0° NCEP final global tropospheric analyses and integrated for 120 h. The simulated SLP field was then used as the observations in the data assimilation. It is shown that the MM5 DRP-4DVar system can successfully assimilate the (simulated) model output (used as observations) because the OSSEs resulted in improved stormtrack forecasts. In addition, compared with an experiment that assimilated the SLP data fixed at the end of a 6-h assimilation window, the experiment that assimilated the SLP data every 3 min in a 30-min assimilation window further improved the typhoon-track forecasts, especially in terms of the initial vortex location and landfall location. Finally, the assimilation experiments with a specified SLP field have demonstrated the effectiveness of the new method.