Estimation of regional trends in sulfur dioxide over the eastern United States

Abstract

Emission reductions were mandated in the Clean Air Act Amendments of 1990 with the expectation of concomitant reductions in ambient concentrations of atmospherically-transported pollutants. To evaluate the effectiveness of the legislated emission reductions using monitoring data, this paper proposes a two-stage approach for the estimation of regional trends and their standard errors. In the first stage, a generalized additive model (GAM) is fitted to airborne sulfur dioxide (SO 2 ) data at each of 35 sites in the eastern United States to estimate the form and magnitude of the site-specific trend (defined as percent total change) from 1989 to 1995. This analysis is designed to adjust the SO 2 data for the influences of meteorology and season. In the second stage, the estimated trends are treated as samples with site-dependent measurement error from a Gaussian random field with a stationary covariance function. Kriging methodology is adapted to construct spatially-smoothed estimates of the true trend for three large regions in the eastern U.S. Finally, a Bayesian analysis with Markov Chain Monte Carlo (MCMC) methods is used to obtain regional trend estimates and their standard errors, which take account of the estimation of the unknown covariance parameters as well as the stochastic variation of the random fields. Both spatial estimation techniques produced similar results in terms of regional trend and standard error.

Investigating the association between late spring Gulf of Mexico sea surface temperatures and U.S. Gulf Coast precipitation extremes with focus on Hurricane Harvey

“…In this work, we exclusively utilize the exponential covariance function as we believe that use of alternative covariance functions would make little difference in terms of estimating θ ( s ) at , which is ultimately the purpose of the analysis. This issue has been addressed in prior literature, and results show that the choice of the spatial covariance function makes little difference in the predictions of interest (Holland, Caragea, & Smith, ; Holland et al, ).…”

An autoregressive point source model for spatial processes

“…This is commonly done using a linear or non-linear function where the functional form of f (·,·) is completely known and only the vector β is unknown. Generalized additive models GAMs, (Hastie and Tibshirani, 1990) may also be used as a first-stage approximation of the trend assuming errors are independent and identically distributed; see Holland et al (2000) for an application of GAMs to environmental data.…”

Interpolation performance of a spatio-temporal model with spatially varying coefficients: application to PM10 concentrations in Rio de Janeiro

“…Also, a preliminary estimate for r 2 e can be obtained as the sample variance of the set of all residuals fY ðs i ; t j Þ Àlðs i ; t j Þg, wherê lðs i ; t j Þ ¼b 0 ðs i Þ þb 1 ðs i ÞX 1 ðs i ; t j Þ þ Á Á Á þb p ðs i ÞX p ðs i ; t j Þ, i=1,...,N, j=1,...,T. These preliminary estimates serve as the location for the respective priors in (5), while the spread of the priors are determined subjectively. Similar methods were used, in a non-Bayesian context, by Eynon and Switzer (1983), Oehlert (1993) and Holland et al (2000) to obtain parameter estimates in some space-time models. For the precision parameter r À2 / we use the non-informative prior (NIP) r À2 / $ Gð10 À6 ; 10 À6 Þ.…”