Errors in tropical cyclone intensity forecast by RSMC Tokyo and statistical correction using environmental parameters

Abstract

Tropical cyclone (TC) intensity forecasts issued by the Regional Specialized Meteorological Centre (RSMC) Tokyo - Typhoon Center are systematically compiled to analyze the long-term behavior of errors and to explore the potential for improvement in the forecast accuracy using a statistical correction approach. In this study, a dataset is constructed from annual statistics and every single forecast listed on annual reports on the activities of the RSMC Tokyo. This study found that (1) the accuracy of annual mean forecast has not improved over 26 years and that (2) forecast errors tend to be larger in the rapidly developing TCs. Further analysis reveals that recent forecast output (2008-2014) contains biases associated with the magnitude of the vertical shear of horizontal wind, convective available potential energy, upper ocean temperature, maximum potential intensity (MPI) and ocean coupling potential intensity (OC_PI). To evaluate the adverse effect of such biases in the current forecast system, a simple statistical correction is applied. It improved TC intensity forecast by 7.8-16.9% when an OC_PI is employed.