The effects of different cumulus parameterizations and microphysics schemes in WRF on Typhoon Matmo track after landing

The track of typhoon Matmo (No. 201410) after its landfall and affecting Jiangxi is simulated by WRF V3.5 with different combinations of 7 cumulus parameterization and 13 microphysics schemes (total 91 numerical experiments). The results show that the track is sensitive to the selection of cumulus parameterizations and microphysics schemes. In all 91 experiments, the distance between simulated and observed track ranges from 29.7 to 76.9 km, with a mean deviation of 54.5 km; The tracks simulated by the combinations of Grell-Freitas and WDM5 (N38) or WDM6 (N39) are the most close to observation (deviation about 30km). The combinations of BMJ and SBU-YLin (N14), BMJ and Milbrandt (N22), and BMJ and SBU-YLin (N24) did not result in good accuracy to simulate the track, which deviates eastward over 75 km from observation. For a given cumulus (microphysics) scheme, the skills of the 13 microphysics (7 cumulus) schemes are largely different. Generally, Grell-Freitas and Grell-3 perform the best in the 7 cumulus parameterizations, while WDM5 and WDM6 perform the best in the 13 microphysics schemes. The main reason for the good performance of N38 and N39 is that they accurately simulate the location and evolution of the Western Pacific subtropical high (WPSH). Contrarily, the poor performance of N14, N22 and N24 mainly result from the simulation of the WPSH which deviates eastward from observation.