Tropical cyclone track prediction using a large-area WRF model at the Hong Kong Observatory

Abstract

Tropical cyclone (TC) track predictions of the 10-km resolution WRF (provisionally named "AAMC-WRF") of the Hong Kong Observatory (HKO), spanning (20⁰S - 60⁰N, 45⁰E − 160⁰E) is studied for a 1-year period from April 2018 to Mar 2019. Real-time predictions, up to 4 times a day and T+48 h ahead, are verified against operational analysis positions of HKO for storms over the South China Sea (SCS) and Western North Pacific (WNP); and of the New Delhi Regional Specialised Meteorological Centre (RSMC) for storms over the North Indian Ocean basin (NIO; including the Bay of Bengal). Out of 21 named TCs over SCS and WNP, mean positional errors of the AAMC-WRF are 33 km (T+0), 63 km (T+24), and 107 km (T+48) based on 209, 178 and 142 forecasts. The AAMC-WRF outperformed Meso-NHM, also run in real-time at HKO, with mean error reduction up to 34 km or 24%. Mean positional errors for 13 NIO storms are 38 km (T+0), 69 km (T+24) and 107 km (T+48) based on 183, 131 and 85 forecasts. This is the first study in which TC predictions of a regional model are simultaneously examined over the SCS, WNP and NIO basins through real-time experiments.