Abstract
Numerical weather predictions of three heavy rainfall events in the northeast monsoon causing floods and damage in southern Thailand are reported in this paper. The Pennsylvania State University/National Center for Atmospheric Research (PSU/NCAR) mesoscale model (MM5) was used with the Betts-Miller (BM), Grell (GR), and new Kain-Fritsch (KF2) convective parameterization (CP) schemes at 5 km resolution, and also with an unparameterized or explicit (EX) scheme, to look for a promising method for precipitation forecasting. The accumulated precipitation amounts predicted by the model were evaluated qualitatively by comparison with the Tropical Rainfall Measuring Mission (TRMM) maps, and quantitatively by calculating statistical score indices for different rainfall thresholds. The simulations were evaluated by comparing the synoptic near surface fields with the NCEP FNL objective analysis fields and also with the vertical profiles of wind speed, temperature and mixing ratio at Songkhla station.
The simulations gave generally satisfactory predictions of the synoptic fields, except for an overprediction of the relative humidity. The heavy rainfall was associated in each case with a small vortex where the northeasterly wind interacted with a warm humid southerly wind. The model underestimated the heavy rainfall amounts, and widely different rainfall patterns were produced by the different schemes used. The KF2 and EX schemes generally gave better results than the BM and GR schemes.
