Particulate matter is one of the primary atmospheric pollutants with significant effects on human health. Accurately and reliably forecasting air quality for future horizons makes it possible to take the necessary precautions to minimize potential risks. In this study, monthly PM₁₀ concentration forecasts were made for Erzurum in Turkey. The first ten years of monthly data between 2006 and 2018 were used for training of the model, and the last two years were used to test predictions with the model. PM₁₀ data had trends and seasonal effects removed with seasonal adjustment and were decomposed to three levels with MODWT. For each subseries obtained, modelling was performed with appropriate coefficients chosen with ARIMA. Particulate forecasting was performed with wavelet reconstruction for the approximate and detail series. According to the experimental results, the wavelet-transform based hybrid WT-ARIMA model was more successful than the traditional ARIMA model with regard to the RMSE, R², IA, MAE and MAPE. The developed model had values of RMSE 1.50, R² 0.99, IA 99.92%, MAE 1.26 and MAPE 3.02%. The proposed model may be used as reference for early warning in regions with high air pollution observed due to accurate forecasting capability for particulate matter pollution.

颗粒物是主要的大气污染物之一，对人类健康有重大影响。准确可靠地预测未来的空气质量，可以采取必要的预防措施，最大限度地减少潜在风险。在这项研究中，对土耳其埃尔祖鲁姆进行了每月 PM ₁₀浓度预测。2006年至2018年的前十年月度数据用于模型训练，后两年用于测试模型预测。下午_{10 点}数据通过季节性调整去除了趋势和季节性影响，并使用 MODWT 分解为三个级别。对于获得的每个子系列，使用 ARIMA 选择的适当系数进行建模。通过小波重建对近似和细节序列进行粒子预测。根据实验结果，基于小波变换的混合WT-ARIMA模型在RMSE、R ²、IA、MAE和MAPE方面比传统ARIMA模型更成功。开发的模型具有 RMSE 1.50、R ² 0.99、IA 99.92%、MAE 1.26 和 MAPE 3.02% 的值。由于对颗粒物污染的准确预测能力，所提出的模型可作为观测到的高空气污染地区的预警参考。