Abstract |
Past studies with the variational data assimilation systems have shown that assimilating precipitation observations improves the analysis, while the model quickly "forgets" their impacts, and the resulting forecasts are not substantially improved after a few hours. By introducing the Gaussian transformation to the precipitation data, and directly modifying the dynamical variables through an Ensemble Kalman Filter system, Lien et al. (2013) showed with the SPEEDY model that assimilating precipitation can improve both the analysis and medium-range forecast for almost all variables. Consistent positive results were then observed in the follow-up studies using more realistic global models and observations (GFS-LETKF/TMPA, Lien et al. 2016a, 2016b; NICAM-LETKF/GSMaP, Kotsuki et al. 2017). However, it is still unclear whether this methodology is applicable to the mesoscale regional model. Therefore, it is of our interest to investigate whether precipitation assimilation can provide additional forecast benefits to the regional model. During this internship, we implement the precipitation assimilation with the SCALE-LETKF regional data assimilation system, and conduct experiments on Typhoon Chan-hom and Nangka in 2015. The preliminary results show that with the GSMaP precipitation assimilation, the sea level pressure analysis of the typhoon center is more close to the best track data. In addition to the dynamical adjustments, the hydrometeor fields around the typhoon center are also improved when verified against the cloud liquid water and ice water path retrievals from the GPM Microwave Imager. The forecast experiments show that additional precipitation assimilation improves both the track and intensity forecasts.
|