An investigation into the spatio-temporal patterns of modelling SO2, NOx and surface O3 across the Highveld priority area, South Africa

Abstract

The Highveld is identified as an air pollution ‘hotspot’ area where pollutant concentrations are elevated due to the high density of industrial and non-industrial air pollution sources. To enhance air quality across the Highveld, it was declared a priority area to manage and monitor pollutants to reduce their negative impact on the environment and society. Hence, the aim of this study was to investigate ambient air pollution across the Highveld Priority Area (HPA), using ground-level SO2, NOx and surface O3 concentrations, meteorological parameters and Moderate resolution imaging spectroradiometer (MODIS) atmosphere products, for January to December 2011, to develop new modelling techniques to aid in the management of air pollution. Results show the annual mean trace gas concentrations of SO2, NOx and surface O3 were 12.14, 14.75 and 28.77 ppb, respectively. SO2 and NOx concentrations were highest during winter at an average of 17.56 and 20.96 ppb, where surface O3 concentrations were highest during spring at an average of 32.82 ppb. Diurnal patterns of SO2 and surface O3 were similar, where a midday peak occurred. NOx concentrations instead showed peaks during traffic hours. Ambient air temperature, solar radiation, relative humidity, wind speed and rainfall levels peaked during summer. Atmospheric pressure was relatively stable throughout the year. Winds typically ranged from N to E up to April and from S to NW from May. Very little variation in SO2 and NOx concentrations was explainable by meteorology, 4 to 29 % and 5 to 23 %, while the influence of meteorology on surface O3 concentrations was more significant, 23 to 53 %. Spatial multiple regression statistical models using a cross validation approach for model validation were made over a number of temporal scales. The model fitting and validation processes indicated that the models were not a good fit as only up to 69, 74 and 58 % of SO2, NOx and surface O3 concentrations with high root means square error (RMSE) values of up to 22.10, 15.56 and 18.59 ppb, respectively, could be explained by the models. This process revealed the potential to model pollutants across the HPA, and as a pilot study future work can be based on this study. It is clear that spatial modelling for pollution estimation and management is necessary as seen by the frequent exceedances of the national and international ambient air quality standards.