Our research on atmospheric chemistry problems aims at exploring the fundamental processes that govern the concentrations and distributions of trace gases and aerosols. Our research helps unravel of the complicated relations between the changing environment and human activities. We use statistical and numerical methods to analyze large datasets of tropospheric chemical measurements at surface sites, from balloons and aircraft, and by satellites. We have many colleagues in our school and Civil and Environmental Engineering working on various aspects of environmental issues. Statistical analysis   We use data analysis to extract the underlying scientific principals from complex measurement data. Fairly simple statistical analysis can be powerful tools (e.g., Wang et al. [2000a]). Increasingly, however, the large amounts of measurements data demand more advanced statistical methods. One particular method we have used is the positive matrix factorization (PMF), which explores the covariance structure of the datasets. The technique has been used to examine the effects of intercontinental transport on tropospheric ozone (Wang et al. [2003b]); the sources of biogenic volatile organic compounds (VOCs); the sources of aerosols (PM2.5) in the Southeast (Liu et al. [2005, 2006], Lee et al. [2008], Ke et al. [2008]). It is also used as a new way to evaluate model simulated ozone enhancements by tropospheric photochemical production and transport from the stratosphere (Shim et al. [2008]) and to estimate sources of biogenic emissions (Shim et al. [2007]). Bayesian inversion of trace gas emissions is another advanced statistical method we use in analyzing measurements and model simulations. By quantifying the difference between model simulations and observations, we derive the a posteriori emissions that best represent our current knowledge of the emissions processes and available atmospheric measurements. We have used the method to derive the global emissions of two environmentally important trace gases, isoprene (Shim et al. [2005]) and CH3Cl (Yoshida et al. [2006]). 0-1 D modeling   Point (0-D) model is built with our current understanding of photochemical reactions. It captures the complexity of  photochemical interactions. The models are generally constrained by the observed concentrations of some (long-lived) species. The model predictions of the other (generally shorter-lived) species can be compared to the observations. Disagreement between simulated and observed concentrations therefore indicates the problems in our current understanding of photochemistry and points to the need for laboratory studies. We have applied a point model to examine the photochemistry at northern mid and high latitudes (Wang et al. [2003a], Fried et al. [2003], Kondo et al. [2004]). The 1-D model captures the effects of vertical transport in addition to photochemical reactions in the 0-D model. Over the tropical Pacific, convective transport is a critical process that defines the characteristics of the photochemical environment (Wang et al. [2000b; 2001]). When analyzing the measurements over the South Pole, we apply the 1-D model to derive NOx emissions from snow based on surface measurements. Parameterization of snow NOx emissions allowed us to conduct the first 3-D chemical transport model study of photochemistry over the Antarctic Plateau (Wang et al. [2007]). Regional and global 3-D modeling   The 3-D models attempt to simulate the interaction between photochemistry and dynamic transport. The processes include emissions, transport, photochemistry, and deposition. We have developed or applied several regional or global chemical transport models. We have developed a regional 3-D chemical transport model. Data assimilation of the meteorological fields is conducted using the NCAR/Penn State MM5 model or the Weather Research and Forecasting (WRF) model. For high-latitude simulations, the polar version of MM5 is used. We applied the modeling system to examine the halogen-driven oxidation of ozone and hydrocarbons in the Arctic spring (Zeng et al. [2003, 2006], Ridley et al. [2007]); photochemistry driven by snow emissions and aerosol transprot over the Antarctica (Wang et al. [2007], Arimoto et al. [2008], Davis et al. [2008]); the signals of lightning NO production and convective transport of CO over North America and the western North Atlantic in satellite measurements (Choi et al. [2005]); the late-spring increase of trans-Pacific pollution transport in the upper troposphere (Wang et al. [2006]); the seasonal transition of ozone photochemistry and outflow from spring to summer over North America (Jing et al. [2006], Choi et al. [2008a, 2008b]). Using this regional 3-D modeling system, we have provided operational (daily) 48-hour forecasts of O3 and its precursors over the U.S (Guillas et al. [2008]) since summer 2003. In addition to this modeling system, we use the EPA Models-3/CMAQ system to investigate the interactions of biosphere, forest fires, air quality, and climate over the Southeast (Tian et al. [2008]; Zeng et al. [2008]). Previously we have applied a global 3-D chemical transport model to examine the origin of tropospheric ozone, global ozone and OH changes due to human activities since preindustrial tims, and the climatic impacts (Wang et al. [1998a, b, c], Wang and Jacob [1998], Hansen et al. [2002]). We currently use the GEOS-CHEM model for global simulations. The assimilated meteorological fields are provided by the Goddard Earth Observing System (GEOS) of the NASA Global Modeling Assimilation Office. The model has been used to investigate the distributions of global CH3Cl (Yoshida et al. [2004]); Bayesian inversion of global isoprene and CH3Cl sources (Shim et al. [2005]; Yoshida et al. [2006]); the tracer (ethane and propane) correlation in the troposphere and its usage in evaluating model transport (Wang and Zeng [2004]); ozone enhancements at northern mid latitudes (Shim et al. [2008]).