Applications of a New Tropical Cyclone Initialization Scheme on Improving TC Intensity, Structure and Track Forecasts

The tropical cyclone (TC) initialization scheme developed by Nguyen and Chen (2011, 2014) was used to produce the initial TC structure and intensity in the model for 18 TCs (2004-2013) over the western Pacific using the Weather and Research Forecast Model (WRF). For all these storms, the storm structure and intensity in the model initial conditions agrees well with the best track data and satellite observations, including Vmax, Pmin, symmetric vs. asymmetric structure, and single vs. double eyewall. These results attest that the environment, including SST, in which the storm is embedded has a significant impact on the structure and intensity of these well-developed TCs. With better initial conditions, the new TC initialization scheme results in significantly improved track and intensity forecasts, especially the intensity forecast during the first 48 hours for four landfalling storms from the open ocean between the Philippines and Vietnam in 2006. Recently, the scheme was used to initialize the model throughout the entire life cycle of Super Typhoon Jelawat (2012) over the western Pacific, which underwent a rapid intensification (RI) stage. The new scheme was also tested in real-time experiment settings for a Category 1 tropical storm (Ana 2014) over the Hawaiian Islands. Our model predictions were compared with the performance of the Hurricane WRF model (HWRF) and the Coupled Ocean/Atmosphere Mesoscale Prediction System (COAMPS) with encouraging results. Additional tests were made for a few other storms including Haiyan (2013) over the western Pacific and Iniki (1992) over the Hawaiian Islands. For all cases considered, the scheme works well in predicting intensity, structure, and track for storms with different intensities at different stages of their life cycle and under different large-scale settings. This is because at the model initial time, initial storm intensity, and structure are well adjusted to the environmental conditions in which it is embedded and well adapted to the model employed.