An intercomparison of WRF-ARW and JMA-NHM performance in prediction of tropical cyclones over the South China Sea in 2008

The Japan Meteorological Agency Non-Hydrostatic Model (JMA-NHM) (Saito et al. 2006) is an operational non-hydrostatic mesoscale model currently in use at JMA and the Hong Kong Observatory (HKO). An intercomparison of the Advanced Research WRF (WRF-ARW) (Skamarock et al. 2005) and JMA-NHM has been conducted to examine their performance in tropical cyclone prediction for the tropical cyclones over the South China Sea. With a horizontal grid spacing of 10 km and 50 vertical levels using the same domain configurations and similar physics schemes, 72-hour integrations of the models are run to simulate six tropical cyclones which affected Hong Kong in 2008. The initial and lateral boundary conditions are extracted from the 0.5°×0.5° outputs of the JMA Global Spectral Model (GSM), along with the global high resolution sea surface temperature analyses produced by NCEP.

The performance of both models in terms of tropical cyclone track and intensity predictions has been verified against the best track dataset from HKO. Both models exhibit similar skills in track prediction, with JMA-NHM exhibiting a smaller average track error over the first 36 hours of prediction. On the other hand, WRF-ARW produces more accurate intensity forecasts due to a more efficient spin-up of tropical cyclone circulation than JMA-NHM. Both models have achieved appreciable improvement over the corresponding predictions from JMA GSM. The quantitative precipitation forecasts from both models have also been verified against the precipitation analysis based on passive microwave satellites. Judging from the bias score, equitable threat score and root mean squared error, WRF-ARW generally performs better than JMA-NHM in the quantitative precipitation forecasts of daily rainfall amounts, though the former tends to over-predict the occurrence of light rain. In terms of computational efficiency, WRF-ARW displays an advantage over JMA-NHM.