The analysis and application of a new hybrid pollutants forecasting model using modified Kolmogorov-Zurbenko filter

- Two modified KZ filter techniques (MKZ and MKZA) are developed to overcome the drawback by traditional KZ filter.
- The statistical description of the data are made to discover the characteristics of the pollutants.
- A decomposition-ensemble technique is employed in the developed hybrid model to make forecasting.
- A comprehensive experiment is conducted to validate the performance of the model.

Cities in China suffer from severe smog and haze, and a forecasting system with high accuracy is of great importance to foresee the concentrations of the airborne particles. Compared with chemical transport models, the growing artificial intelligence models can simulate nonlinearities and interactive relationships and getting more accurate results. In this paper, the Kolmogorov-Zurbenko (KZ) filter is modified and firstly applied to construct the model using an artificial intelligence method. The concentration of inhalable particles and fine particulate matter in Dalian are used to analyze the filtered components and test the forecasting accuracy. Besides, an extended experiment is made by implementing a comprehensive comparison and a stability test using data in three other cities in China. Results testify the excellent performance of the developed hybrid models, which can be utilized to better understand the temporal features of pollutants and to perform a better air pollution control and management.

275