Application of digital filter weak constraint four dimensional variational data assimilation on typhoon initialization I. Numerical experiments

Abstract

Traditional four-dimensional variational (4D-Var) bogus data assimilation (BDA) will upset the internal dynamical balance of wind and mass fields, which will generate spurious fast oscillations in the prediction and cause the numerical simulation failure. How to effectively suppress the spurious fast oscillations is very important. In this paper (part I), the digital filter weak constraint 4D-Var is tested in BDA experiments. Initialization and simulation experiments are conducted for typhoon Chaba (2010) using this approach. Results show that the initial noise property is different for typhoon case and rainfall case, and the difference in surface pressure tendency is small in assimilation time and large at the initial stage in different digital filter weight setting BDA experiments. The vertical velocity field of 700 hPa and divergence field of 850 hPa in typhoon areas are both weakened in digital filter weak constraint 4D-Var experiments. Compared with the traditional 4D-Var experiment, the track and intensity predictions are improved obviously in digital filter weak constraint 4D-Var experiment with a proper digital filter weight setting, and the track prediction is apparently improved. Meanwhile, the optimal digital filter weight for different typhoon cases is different, and how to optimally select the digital filter weight is very important. In Part Ⅱ, a digital filter weight selection method will be proposed for BDA experiment.