dc.description.abstract | This study focused on investigating the impacts of assimilating Global Telecommunication System (GTS) and/or GPS Radio Occultation (GPSRO) data using two assimilation systems 3-Dimension variational (3DVAR) and Ensemble Adjustment Kalman Filter (EAKF) on the realtime forecast for Typhoon Morakot (2009) using a 18-km grid. From statistical verifications of simulated rainfall with dense ground base rain guage data, we found that assimilation GTS and GPSRO has improved the 24-48 h rainfall forecast because the initial vortex is better resolved with data assimilation. However, beyond 48 h model runs, there is no significant improvement in rainfall forecast skill. We also found that model initialization using the EAKF assimilation system produces better rainfall forecast as compared with the 3DVAR data assimilation system simply because the EAKF data assimilation system provides better initial hurricane vortex as compared with the 3DVAR data assimilation scheme.
We conduct a series of sensitivity tests by nest down the initial vortex after data assimilation to a nested domain with a 2-km grid. We found that the lateral boundary conditions for the 2-km convection-allowing model provided by National Centers for Environmental Prediction (NCEP) Global Forecast System (GFS) produces better rainfall forecast after 24 h of model integration than the lateral boundary conditions from both the EAKF and 3DVAR runs with a 18-km grid. It is apparent that with different initial vortex (GFS, EAKF and GFS) rainfall forecast within the first 24 h is mainly affected by the strength and structure of the initial vortex. After 24-48 h model integration, the influences of lateral boundary conditions on model forecasts become important. It is apparent that the track and rainfall forecasts after 24-48 h model integration are dominated by environment flow.
We also examine 18 storms (2004-2013) over the western Pacific using TC method developed by Nguyen and Chen (2011). The preliminary results of this study show that the environment has a significant effect on the initial storm structure. During the early season, storms embedded within the southwesterly monsoon flow have a tendency to exhibit a “9” type asymmetric structure with an upper level outflow channel extending southwestward from the southeastern quadrate of the storm. At low levels, the convergence area between the storm circulation and the southwesterly flow is a favorable location for the development of spiral rainbands. Late season storms have a tendency to produce a “6” type storm structure with an outflow channel extending northeastward from the northwestern part of the eyewall, especially when an upper-level cold low or trough is present to the northwest of the storm. At low levels, the convergence of the northeasterly monsoon flow and the cyclonic circulation of the storm are favorable for the occurrences of spiral rainbands. For intense storms that underwent an eye-wall replacement cycle, the NC2011 scheme also shows considerable skill in reproducing the double eye-wall structure in the model initial conditions. So, finally we will combine NC2011 and 3DVAR data assimilation methods to study TC Jelawat (2012) and make some discussion about the forecasting results. Our preliminary results show that combining NC2011 and 3DVAR data assimilation method is feasible, however, because the track forecast affected by large-scale circulation is too significant, so the results did not fully achieve the desired effect, but still showing some characteristics close to satellite observations. | en_US |