A number of studies have revealed the effects of several meteorological factors on the concentration of ambient particulate matter (PM). However, the corresponding non-linear and lagged effects of those variables on PM concentration remain unclear. This study aimed to explore the effect of meteorological variables and fire hotspot counts on PM concentration in the northern part of Thailand using the two-stage approach. In the first stage, combination of log-linear regression model and distributed lag non-linear model was used to estimate province-specific effect of meteorological factors and fire hotspots on PM_2.5 and PM₁₀ adjusting by confounders. In the second stage, the multivariate meta-analysis was applied to pool province-specific estimate, and the results showed that the negative associations of temperature, relative humidity, wind speed, and rainfall with PM_2.5 and PM₁₀ were observed, while air pressure and fire hotspot was positively associated with PM. Moreover, the effects of temperature, relative humidity, wind speed, rainfall, and fire hotspot on PM_2.5 and PM₁₀ were lasted for 5–8 days, whereas those of air pressure were peaked at lag 7 days and become attenuated toward the null afterward. In conclusion, results from this study could contribute a better understanding on the effects of meteorological variables and fire hotspots on PM concentration and induce the development of more tailored air pollution mitigation strategies depending on weather conditions.

许多研究揭示了几种气象因素对环境颗粒物 (PM) 浓度的影响。然而，这些变量对 PM 浓度的相应非线性和滞后影响仍不清楚。本研究旨在使用两阶段方法探讨气象变量和火灾热点计数对泰国北部 PM 浓度的影响。第一阶段，结合对数线性回归模型和分布式滞后非线性模型，估计气象因素和火灾热点对 PM _2.5和 PM _{10 的}分省效应通过混杂因素调整。第二阶段，多变量荟萃分析应用于池省特定估计，结果显示温度、相对湿度、风速和降雨量与 PM _2.5和 PM ₁₀呈负相关，而气压火热点与 PM 呈正相关。此外，温度、相对湿度、风速、降雨和火灾热点对 PM _2.5和 PM _{10 的影响}持续了 5-8 天，而气压在滞后 7 天达到峰值，之后逐渐减弱。总之，这项研究的结果有助于更好地了解气象变量和火灾热点对 PM 浓度的影响，并根据天气条件制定更量身定制的空气污染缓解策略。