How the different analyses forcing fields affect the tropical cyclone simulation?

Regional model and detailed cloud-resolving storm simulation system typically require the initial and boundary conditions from analysis data to make tropical cyclone simulation and forecast. Currently there are several different analyses and reanalyses datasets compiled by leading operational centers. Even though all the analyses are intended to represent the observed atmospheric state as close as possible, their differences due to different modeling and data assimilation systems are not negligible. There is a potential impact from these different forcing fields on tropical cyclone track and intensity forecast as well as the dynamically downscaled high-resolution regional climate simulation and the embedded climate and weather extremes. Presently such regional modeling work often use the analyses data that are more readily available or specified in the experiment protocol.

Our study aims to make assessment on how the differences in the analysis/reanalysis data affect the simulation of tropical cyclone track, intensity, and associated rainfall. Seven different analyses/reanalyses (EC-YOTC, ERA-Interim, JRA25, NCEP-CFSR, NCEP-FNL, NCEP R1, and R2) data were used as initial and lateral boundary conditions for a Cloud-Resolving Storm Simulator (CReSS) model for the simulation of Typhoon Morakot over Northwest Paciifc in early August 2009. The track error due to low resolution forcing data, such as NCEP R1, are most prominent and can lead to significant underestimate of the torrential rainfall induced by topography over souther Taiwan. The intensity of model was only reproduced when the model is drived by the high-resolution EC-YOTC analysis. The model simulated precipitation amount associated with the Typhoon Morakot over the open ocean can be more than triple in the run forced by EC-YOTC analysis as compared to the run forced by other analyses.