Mutation test and multiple-wavelet coherence of PM2.5 concentration in Guiyang, China

The negative effects of PM2.5 concentration in urban development are becoming more and more prominent. Bernaola-Galvan Segmentation Algorithm (BGSA) and wavelet analysis are powerful tools for processing non-linear and non-stationary signals. First, we use BGSA that reveals there are 41 mutation points in the PM2.5 concentration in Guiyang. Then, we reveal the multi-scale evolution of PM2.5 concentration in Guiyang by wavelet analysis. In the first part, we performed one-dimensional continuous wavelet transform (CWT) on the eight monitoring points in the study area, and the results showed that they have obviously similar multi-scale evolution characteristics, with a high-energy and significant oscillation period of 190–512 days. Next, the wavelet transform coherence (WTC) reveals the mutual relationship between the PM2.5 concentration and the atmospheric pollutants and meteorological factors. PM2.5 concentration variation is closely linked to that of PM10 concentration. But, it is not to be ignored that the increase in the SO2 and NO2 concentrations will cause the PM2.5 concentration to rise on different scales. Lastly, the variation of the PM2.5 concentration can be better explained by the combination of multiple factors (2-4) using the multiple-wavelet coherence (MWC). Under the combination of the two factors, the average temperature (Avgtem) and relative humidity (ReH) have the highest AWC and PASC. In the case of the combination of four factors, CO–Avgtem–Wind–ReH plays the largest role in determining PM2.5 concentration.