Analysis of uncertainties in forecasts of typhoon soudelor (2015) from ensemble prediction models

Abstract

Using data from nine ensemble prediction systems (EPSs), we analyze uncertainties in forecasted tropical cyclone TC track (TCT), TC intensity (TCI) and relevant heavy rainfall (TCHR) for Typhoon Soudelor (2015) as it affected the Taiwan Strait and surrounding regions. The largest uncertainties in track predictions occurred when Soudelor traversed Taiwan and when it recurved northeastward after making landfall in mainland China. These large uncertainties seem to be ascribed to the topography of Taiwan and the spread of the perturbed steering flows, respectively. TCI spread was stronger before rather than after the Soudelor made landfall, with regional EPSs having stronger spread than global EPSs. This TCI spread showed high correlation with the evolution of the spread of vertical wind shear at the location of TC center. Large spread in 24-h TCHR during Soudelor's landfall correlated with low-level jets and convergences in most EPSs, and TC track variation had played important role in TCHR uncertainty. At last, the spread-skill relationships among different groups are explored.