Spatio-Temporal Analysis of Total Nitrate Concentrations Using Dynamic Statistical Models

Abstract

Atmospheric concentrations of total nitrate (TNO 3 ), defined here as gas-phase nitric acid plus particle-phase nitrate, are difficult to simulate in numerical air quality models due to the presence of a variety of formation pathways and loss mechanisms, some of which are highly uncertain. The goal of this study is to estimate the relative importance of these different pathways across the Eastern United States by identifying empirical relationships that exist between TNO 3 concentrations and a set of covariates (ammonium, sulfate, ozone, wind speed, relative humidity, and precipitation) measured from January 1997 to July 2004. We develop two dynamic statistical models to quantify these relationships. A major advantage of these models over typical linear regression models is that their regression coefficients can vary temporally. Results show that TNO 3 is sensitive to ozone throughout the year, indicating an importance of daytime photochemical production of TNO 3 , especially in the Southeast. Sensitivity of TNO 3 to residual ammonium (NH 4 + –2SO 4 2− ) is most pronounced during winter, indicating a seasonal importance of gas/particle partitioning that is accentuated in the Midwest. Using a number of physical and chemical explanations, confidence is established in the spatial and temporal patterns of several such empirical relationships. In the future, these relationships may be used quantitatively to improve our mechanistic understanding of TNO 3 formation pathways and loss mechanisms in the atmosphere.