Variability of historical tropical cyclone best track databases and their impact on the developed stochastic track models and estimated wind hazard for mainland China

Historical tropical cyclone (TC) best track databases are used as the basis to develop the stochastic TC track models, which are essential for TC hazard assessment. However, different databases may report different numbers of TCs or inconsistent parameters for the same TC event. The impact of such inconsistencies on the developed stochastic models and estimated TC wind hazards has not been systematically quantified. In this study, we investigated the impact of using different best track databases on statistical characteristics such as TC occurrence rate, translation velocity, translation direction, and central pressure difference for landfalling TC events along the coastline of mainland China. The comparison of the TC track variables for landfalling TCs showed notable differences in statistics among different databases. We developed stochastic beta-advection TC track models for TC events originating from the western North Pacific basin using five TC best track databases provided by different agencies or organizations. We used the developed TC track models to assess TC wind hazards for sites in the southeast coastal regions of mainland China. We presented a quantitative comparison of the T-year return period values of the annual maximum TC wind speed, vA-T, estimated by using the developed stochastic track models as well as using the historical tracks directly. The results indicated that the estimated vA-T for T = 50 and 500 years for some sites can be sensitive to the TC track database used to develop the stochastic track model.