A detrended cross-correlation analysis of meteorological and API data in Nanjing, China

• Scaling behaviors of cross-correlation of temperature and atmospheric pressure against the API are related to time scales.
• There are crossover time scales in cross-correlation of temperature and atmospheric pressure against the API.
• MF-CCA method is more subtle and suitable to reflect cross-correlation than MF-DCCA.

The cross correlation between daily meteorological data and air pollution index (API) records in Nanjing during the past 12 years is studied by means of a detrended cross-correlation analysis (DCCA). In this study, we use statistical significance tests and power-law statistical tests to verify cross correlation between meteorological data and the API. Through calculating the DCCA cross correlation coefficient ρDCCA, we intend to obtain a range of cross correlation levels between the meteorological data and the API at different time scales. Utilizing the multifractal detrended cross correlation analysis (MF-DCCA) and algorithm-multifractal cross correlation analysis (MF-CCA) proposed by Oświecimka, we observe multifractal cross-correlation behavior between meteorological factors and the API. Our results show a cross correlation between meteorological factors and the API in Nanjing. The cross-correlation between diurnal temperature ranges and the API is persistent at studied time scales, while the cross correlations of wind speed, relative humidity, and precipitation with the API are anti-persistent at studied time scales. Next, a cross correlation of temperature with the API finds persistent cross correlation at smaller time scales, and anti-persistent cross-correlation at larger time scales; the cross correlation of atmospheric pressure with the API, however, results in anti-persistent cross correlation at smaller time scales, and persistent cross correlation at larger time scales. The MF-DCCA demonstrates that all underlying fluctuations have a weak multifractal nature where one scaling exponent is obtained. However, the MF-CCA suggests that some crossovers exist in the cross-correlation fluctuation function in terms of time scales of temperature and atmospheric pressure versus the API. The MF-CCA method is more subtle and suitable for reflecting the cross correlation of the two given time series. Compared with a traditional correlation analysis, the DCCA can uncover more cross-correlation information between API and meteorological factors. Therefore, the DCCA is also recommended as a comparatively reliable method for detecting the correlations between the API and meteorological data, and can also be useful for our understanding of the cross correlation between air quality and meteorological elements.