Singular value decomposition analysis of spatial relationships between monthly weather and air pollution index in China
| Authors: | Libo Zhang, Yongqiang Liu, Fengjun Zhao |
| Year: | 2018 |
| Type: | Scientific Journal |
| Station: | Southern Research Station |
| DOI: | https://doi.org/10.1007/s00477-017-1491-z |
| Source: | Stochastic Environmental Research and Risk Assessment |
Abstract
Spatial patterns are important features for understanding regional air quality variability. Statistical analysis tools, such as empirical orthogonal function (EOF), have been extensively used to identify and classify spatial patterns. These tools, however, do not directly reveal the related weather conditions. This study used singular value decomposition (SVD) to identify spatial air pollution index (API) patterns related to meteorological conditions in China, one of world’s regions facing catastrophic air pollution. The monthly API and four meteorological variables (precipitation, surface air temperature, humidity, and wind speed) during 2001–2012 in 42 cities in China were used. The two leading SVD spatial patterns display the API anomalies with the same sign across China and opposite signs between northern and southern China, respectively. The meteorological variables have different relationships with these patterns. For the first pattern, wind speed is the most important. The key regions, where the correlations between the API field and the wind speed’s SVD time series are significant at the 99% confidence level, are found nationwide. Precipitation and air temperature are also important in the southern and northern portions of eastern China, respectively. For the second pattern, the key regions occur mainly in northern China for temperature and humidity and southern China for wind speed. Air humidity has the largest contribution to this pattern. The weather-API relationships characterized by these spatial patterns are useful for selecting factors for statistical air quality prediction models and determining the geographic regions with high prediction skills.