dc.description.abstract |
The initial condition of a model plays an important role in typhoon forecasts. In this study, the initial typhoon vortex is replaced by a dynamically adjusted typhoon vortex, which is derived from cycling runs of one-hour MPAS model integration. In this study, the MPAS variable-resolution of 60-15 km is used, and the region with 15-km resolution covers the East Asia and North Western Pacific. Three experiments are conducted for each typhoon case, i.e., Typhoon Soudelor (2015), Dujuan (2015), Nepartak (2016) and Megi (2016). All of the initial conditions come from National Centers for Environmental Prediction (NCEP) Final (FNL) analysis but in different resolutions, i.e., 1 degree (CTRL), 0.25 degree (CTRH), and cycling runs with resolution of 0.25 degree (NT).
After about 40-80 cycles for typhoon Soudelor, the typhoon structure is greatly improved than the NCEP global analysis by providing a more organized eyewall. The results show that NT has improvements in early track predictions for typhoons Soudelor, Nepatak and Megi, but in the later simulation for typhoon Dujuan. The NT run for typhoon Soudelor obtains a consistent rainfall pattern with a double-peak precipitation, which in a better agreement with the observation compared to the no-bogussing run, even though the extreme at the northeast side is slightly southward. For the rainfall verification, NT shows better score with extreme (smaller and larger) thresholds in equitable threat score, bias score and probability of detection, but the CTRH has higher score in the middle thresholds. As for spatial correlation coefficient, the accumulated rainfall in different periods is better for CTRH. In general, all the spatial correlation coefficients are very similar and more than 0.5. Generally speaking, NT has the best simulation in the 72-h mean track error, minimum pressure, and maximum wind speed. Therefore, the dynamically adjusted vortex has a positive impact on the typhoon simulations, especially in the intensity and structure. | en_US |