Simulation of equatorial wind field patterns with TAPM during the 1997 haze episode in Peninsular Malaysia

Mesoscale features that would not be highlighted by global numerical prediction models, such as the daily land and sea breeze conditions at the selected air pollution stations located near the west coast of Peninsular Malaysia, were predicted with an Index of Agreement of 0.3, which implied a moderate conformity between the predicted and observed values. Tracer analysis of air particles at a selected location in the Straits of Malacca verified the existence of the landward and shoreward movement of the air during the simulation of the low-level wind field. Air pollutants from the burning areas of neighbouring Sumatra just across the straits were transported towards the Klang valley during the daytime and seawards during the nighttime, highlighting the recirculation features of aged and newer air particles over the seven days (13-18 September) of the model simulation. The presence of the central Main Range east of the Klang valley to a certain extent limited further eastward movement of the air particles. Near calm conditions at low levels were simulated from midnight to midday on 14 September, where the movement of the polluted air particles from the uncontrolled burning in Sumatra was confined within the Straits of Malacca. Turbulence within the planetary boundary layer in terms of the total kinetic energy was found to be weak from 14 to 15 September, congruent with the weak strength of low level winds that reduced the ability of the air to transport the pollutants during the period of severe smoke haze. Statistical evaluation showed that parameters such as the systematic RMSE (root mean square error) and unsystematic RMSE for the zonal wind component were slightly higher than for the meridional one, indicating higher errors between the observed and simulated zonal values. Otherwise, the equatorial meteorological parameters such as wind speed and temperature were successfully simulated by the model with comparatively high correlation coefficients, lower RMSEs and moderately high indices of agreement with observed values.