Prediction of haze episode potential for mountain-valley urban area using synoptic-climatological modeling approach


The Chiangmai city of Thailand is located in a complex mountainous area of Northern Thailand. In recent years, the city has been experiencing air pollution haze episodes during the dry season. The most severe haze was observed in March 2007 when the entire city was blanketed with smoke/haze for a few weeks. The 24-h average PM10 level in the city reached the peak of 396 μg/m3. Recurring haze episodes have become a critical environmental problem that is linked to sharp increases in hospital admissions. This study investigates the main causes of the Chiangmai haze episodes, considering both meteorological and emission sources, with the aim to produce early warning. In the absence of a detailed emission inventory necessary for more comprehensive 3D air pollution model applications for the haze episode warning a synoptic climatological approach is proposed. An automated meteorological classification scheme was developed based on the spatial synoptic index, to identify the most predictive haze-prone meteorological patterns in the area during the months of February-April from 2001-2009. Among the four synoptic patterns that were recognized for the region, pattern 2 was found to prevail on the days with high PM10 in Chiangmai and occurred with a high frequency in the month of March. This pattern is characterized by presence of vast thermal low pressure cells over Chiangmai and Indian continent with hot, dry and stagnant air over the Northern part of Thailand. All identified as episodic high PM10 (higher than Thailand air quality standard of 120 μg/m3) in March 2007 were observed to belong to pattern 2. In addition, the increased levels of PM10 in this month coincided with the high emission from biomass burning and seasonal industries. Backward trajectories also show the air masses wandering over the regions of dense fire hotspots, which represented biomass burning, before arriving to Chiangmai on the haze episode days. By utilizing stepwise regression model for PM10 prediction on the next day, the results of the model were satisfactory for both the dependent set of data of 2007-2009 (R2=88%) and the independent sets of data for 2004-2010 (R2=79 %). The corrective percentage of ""good"" ([PM10] ≤ 40 μg/m3), ""moderate"" (40 < [PM10] ≤ 120 μg/m3) and ""haze alerts"" (120 μg/m3< [PM10]) air quality category forecasts in Chiangmai are 78%, 81% and 89%, respectively. Based on our analysis, we proposed an episode warning procedure that combines the screening for synoptic pattern 2 during February-April period and prediction for haze alerts day in Chiangmai.


Kim Oanh, N. T. Leelasakultum, K


HARMO - Proc. Int. Conf. Harmon. Atmos. Dispers. Model. Regul. Purp,2010


Biomass burning, Chiangmai, Episodic haze warning, Monthly emission variation, Synoptic classificati