Assessment of the relationship between satellite AOD and ground PM10 measurement data considering synoptic meteorological patterns and Lidar data

•Automated scheme for classification of synoptic patterns governing central Thailand•Distinct differences in vertical profiles of Lidar backscatter coefficients between patterns•Improvement of AOD-PM10 relationship in patterns with shallow mixing layers

Correlation between satellite aerosol optical depth (AOD) and ground monitoring particulate matter (PM) depends on the meteorology that determines PM optical properties, its dispersion, accumulation and vertical distribution. This study presents a novel approach to analyze PM-AOD relationship considering the totality of meteorological factors expressed as synoptic patterns. Meteorological observations at 07:00 Bangkok time from 9 regional meteorological stations, in dry seasons (November-April) of 11 years (2000-2010), were used to categorize governing meteorology over Central Thailand into four categories representing the typical observed synoptic patterns. The MANOVA analysis showed that these patterns were statistically different. PM10 recorded at 22 air quality stations in Bangkok Metropolitan Region were examined which showed the highest levels for the days belonging to pattern 1, followed by pattern 4, both with presence of a high pressure ridge, while the minimum for pattern 2 when thermal lows dominated. Lidar aerosol backscatter profiles recorded at Pimai station were used as indicator of PM vertical distribution that showed similarity within each pattern. R2 between MODIS and Sun photometer AODs at Pimai was above 0.8. Correlation coefficients (R) between MODIS AOD and corresponding 1h PM10 for clear sky days (cloudiness ≤ 3/10) were examined for each pattern in comparison with lump case. Significant improvements were observed for pattern 1, average R across 22 stations was 0.46 for Terra and 0.38 for Aqua AOD compared to lump case with R of 0.34 and 0.31, respectively. Comparable improvement was also observed for pattern 4. For pattern 2, R values were significantly reduced which may be caused by the deeper mixing layers and varying vertical profiles with overall low values of Lidar backscatter coefficients. Improved R values in pattern 1 and 4, which had highest PM10 in BMR, suggested a better potential of using MODIS AOD for PM10 monitoring with synoptic pattern classification.