Receptor modeling for multiple time resolved species: The Baltimore supersite

A number of advances have been made toward solving receptor modeling problems using advanced factor analysis methods. Most recently, a factor analysis method has been developed for source apportionment utilizing aerosol compositional data with varying temporal resolution. The data used in that study had time resolutions ranged from 10 min to 1 h. In this work, this expanded model is tested using a data set from the Ponca Street site of the Baltimore supersite with time resolutions ranging from 30 min to 24 h. The nature of this data set implies that traditional eigenvalue-based methods cannot adequately resolve source factors for the atmospheric situation under consideration. Also, valuable temporal information is lost if one averaged or interpolated data in an attempt to produce a data set of the identical time resolution. Each data point has been used in its original time schedule and the source contributions were averaged to correspond to the specific sampling time interval. A weighting coefficient, w24, was incorporated in the modeling equations in order to improve data fitting for the 24-h data in the model. A total of nine sources were resolved: oil-fired power plant (2%), diesel emissions (1%), secondary sulfate (23%), coal-fired power plant (3%), incinerator (9%), steel plant (12%), aged sea salt (1%), secondary nitrate (23%), and spark-ignition emissions (26%). The results showed the very strong influence of the adjacent interstate highways I-95 and I-895 as well as the tunnel toll booths located to the south of the sampling site. Most of the sulfate observed was found to be associated with distant coal-fired power plants situated in the heavily industrialized midwestern parts of the United States. The contribution of the steel plant (<10 miles, 141°SE) to the observed PM concentrations (12%) was also significant.