«A Dissertation Presented to The Academic Faculty by Fernando García Menéndez In Partial Fulfillment of the Requirements for the Degree Doctor of ...»
HIGH-RESOLUTION THREE-DIMENSIONAL PLUME
MODELING WITH EULERIAN ATMOSPHERIC CHEMISTRY
AND TRANSPORT MODELS
The Academic Faculty
Fernando García Menéndez
In Partial Fulfillment
of the Requirements for the Degree
Doctor of Philosophy in the
School of Civil and Environmental Engineering
Georgia Institute of Technology
COPYRIGHT 2013 BY FERNANDO GARCIA MENENDEZ
HIGH-RESOLUTION THREE-DIMENSIONAL PLUME
MODELING WITH EULERIAN ATMOSPHERIC CHEMISTRY
AND TRANSPORT MODELS
Dr. M. Talat Odman, Advisor Dr. Michael H. Bergin School of Civil and Environmental School of Civil and Environmental Engineering Engineering Georgia Institute of Technology Georgia Institute of Technology Dr. Armistead G. Russell, Advisor Dr. Michael Chang School of Civil and Environmental Brook Byers Institute for Sustainable Engineering Systems Georgia Institute of Technology Georgia Institute of Technology Dr. Athanasios Nenes School of Earth & Atmospheric Sciences Georgia Institute of Technology Date Approved: August 7th, 2013 To Adriana,
ACKNOWLEDGEMENTSI would like to express my very great appreciation to Dr. Talat Odman for his support and guidance throughout my doctoral studies. Participating in Dr. Odman’s scientific research has been very rewarding and hopefully we will continue to collaborate in the future. I would also like to thank Dr. Ted Russell for his advising and the opportunity to join an excellent research group. Working alongside the members of the Russell group has been a great experience. I am particularly grateful for the help provided by Dr. Yongtao Hu. Additionally, I appreciate the valuable assistance of Dr. Michael Chang, Dr. Athanasios Nenes and Dr. Michael Bergin as members of the thesis committee. Finally, I would like to thank my family for their encouragement and love during these years.
TABLE OF CONTENTSPage
LIST OF TABLES xLIST OF FIGURES xi SUMMARY xvii
1 Introduction 1
3 Simulating Smoke Transport from Wildland Fires with a Regional-Scale Air Quality Model: Sensitivity to Spatiotemporal Allocation of Fire Emissions 34
4 Simulating Smoke Transport from Wildland Fires with a Regional-Scale Air Quality Model: Sensitivity to Uncertain Wind Fields 66
Figure 2.7: Predicted surface-level PM2.
5 concentrations (µg m-3) over the southeastern U.S. during the Savannah smoke incident of May 17, 2007 at 14:00 LT. The location of Savannah is indicated by the black marker. 24
Figure 2.9: Three-dimensional iso-surfaces defined by PM2.
5 concentration equal to 50 μg m-3: (A) Savannah smoke incident of May 17, 2007 and (B) Atlanta smoke incident of May 16, 2007. Viewer position is from the upper north-west corner of the domain in (A) and from the eastern domain boundary in (B). 26
Figure 3.2: Total fire-related PM2.
5 emissions by vertical layer for Oconee and Piedmont fires. Approximate full layer heights (m AGL) are also included. 39
Figure 3.5: (a) Brute-force first-order sensitivity coefficient for PM2.
5 concentration response to fire emissions, and (b) fire and non-fire contributions to modeled PM2.5 concentrations at South DeKalb on 28 February 2007 (LT). Base-case CMAQ predictions are also included in (b). 46
Figure 3.7: (a) Contributions to modeled PM2.
5 concentrations at South DeKalb from the Oconee and Piedmont fires on 28 February 2007 (LT). Non-fire contribution and base-case CMAQ predictions are also included. (b) First-order sensitivity coefficients for PM2.5 concentration response to fire emissions from each fire at South DeKalb (left axis) and site observations (right axis). 48 Figure 3.8: Change in (a) Oconee fire PM2.5 contribution and (b) maximum simulated PM2.5 concentration relative to base-case at Jefferson St. after relocating Oconee fire emissions into grid cells adjacent to original injection cell. 50 Figure 3.9: Change in (a) Piedmont fire PM2.5 contribution and (b) maximum simulated PM2.5 concentration relative to base-case at Jefferson St. after relocating Piedmont fire emissions into grid cells adjacent to original injection cell. 51 Figure 3.10: Change in (a) Oconee fire PM2.5 contribution and (b) Piedmont fire PM2.5 contribution relative to base-case at Jefferson St. after relocating each fire’s emissions into grid cells downwind (northwest) of fire location. 52
Figure 3.12: Brute-force first-order sensitivity coefficients for PM2.
5 concentration response to fire emissions by vertical CMAQ layer at South DeKalb (left axis) and site observations (right axis) on 28 February 2007 (LT). 55
Figure 3.15: Fire contributions to modeled PM2.
5 concentration at South DeKalb by hour of emissions on 28 February 2007 (LT). Emissions are labeled at the start of the hour. Non-fire contribution and base-case CMAQ results are also included. 58 Figure 3.16: Brute-force first-order sensitivity coefficients for PM2.5 concentration response to fire emissions by hour of emissions at South DeKalb for uniform distribution into 10 lower layers (left axis), and site observations (right axis) on 28 February 2007 (LT). Emissions are labeled at the start of the hour. 60
Figure 4.5: CMAQ-predicted PM2.
5 concentrations (μg m-3) over northern Georgia at 1900 LT on 28 February 2007 using -5° and +5° perturbations to wind direction. Black shaded circles indicate monitoring station locations. The Oconee and Piedmont fire sites are denoted by white shaded markers. 78
Figure 4.7: Maximum CMAQ-predicted PM2.
5 concentrations at the Confederate Ave., Jefferson St., South DeKalb, and McDonough monitoring sites with wind speed perturbations ranging from -50% to +50%. 80
Figure 4.11: Standard deviation (σ) of CMAQ-predicted PM2.
5 concentrations on 28 February 2007 (LT) at Confederate Ave. for all simulations carried out under different PBL heights (±10%, ±20%, and ±30%, and base case) and base case PBL height prediction at Confederate Ave. (right vertical axis). The estimated fire-related contribution to PM2.5 concentration (ΔPM2.5) is also included. 87
Figure 6.3: Three-dimensional visualization of smoke plumes and PM2.
5 concentrations (μg m-3) on March 1, 2007 at 0:30 UT using A) static grid CMAQ and B) AGCMAQ, and at 2:15 UT using C) static grid CMAQ and D) AG-CMAQ. 157
Figure 6.5: Simulated PM2.
5 concentrations (μg m-3) on February 28, 2007 at 22:30 UT using A) static grid CMAQ and B) AG-CMAQ, and on March 1, 2007 at 02:00 UT using C) static grid CMAQ and D) AG-CMAQ. The locations of the McDonough (green), South DeKalb (pink), Confederate Avenue (black), Fort McPherson (blue), Jefferson Street (white), and Fire Station 8 (yellow) air quality monitoring sites are indicated by the colored circles. 161
Figure 7.3: Side view of grid response to a normalized weight field using (a) unconstrained adaptation and (b) vertically constrained adaptation.
Figure 7.9: Maximum EC concentration predicted by a fixed grid simulation, adaptive grid simulations with 5 and 10 adaptation iterations, and a vertically constrained adaptive grid simulation with 5 iterations.
Eulerian chemical transport models are extensively used to steer environmental policy, forecast air quality and study atmospheric processes. However, the ability of these models to simulate concentrated atmospheric plumes, including fire-related smoke, may be limited. Wildland fires are important sources of air pollutants and can significantly affect air quality. Emissions released in wildfires and prescribed burns have been known to substantially increase the air pollution burden at urban locations across large regions.
Air quality forecasts generated with numerical models can provide valuable information to environmental regulators and land managers about the potential impacts of fires. Eulerian models present an attractive framework to simulate the transport and transformation of fire emissions. Still, the limitations inherent to chemical transport models when applied to replicate smoke plumes must be identified and well understood to adequately interpret results and further improve the models' predictive skills. Through this work, the capability of current chemical transport models to replicate fire-related air quality impacts was evaluated, key research needs to achieve effective simulations were identified, and numerical tools designed to improve model performance were developed.
A modeling framework centered on the Community Multiscale Air Quality modeling system (CMAQ) was used to simulate several fire episodes that occurred in the Southeastern U.S. and investigate the sensitivity of fine particulate matter concentration (PM2.5) predictions to various model inputs and parameters. Significant uncertainties associated with fire emissions estimates and their distribution on gridded modeling domains were identified. PM2.5 concentrations predicted by a regional-scale air quality
plume rise and responsive to the horizontal and temporal distribution of fire emissions. In addition to realistic estimates of emitted mass, effectively modeling smoke transport with chemical transport models depends on an accurate spatiotemporal allocation of emissions. The predictions from a regional-scale air quality model also proved to be extremely sensitive to meteorological fields. Normalized errors in model predictions attempting to forecast the regional impacts of fires on PM2.5 levels could be as high as 100% due to inaccuracies in wind data, suggesting that fire-related regional-scale air quality simulations are limited by the performance of existing numerical weather models.
To investigate the influence of grid resolution on model predictions, adaptive grid modeling is explored as a strategy to simulate fire-related plumes. An adaptive version of CMAQ, capable of dynamically restructuring the grid on which solution fields are estimated and providing refinement at the regions where accuracy is most dependent on resolution was developed. In an evaluation simulation aiming to reproduce smoke transport from two prescribed fires, the adaptive grid algorithm reduced artificial diffusion, produced better defined plumes and led to more accurate PM2 5 concentration predictions. Additionally, a three-dimensional adaptive grid algorithm capable of simultaneously refining horizontal and vertical grid resolution is presented. Extremely high levels of grid resolution can be achieved using this grid refinement method. The fully adaptive three-dimensional modeling technique can be applied to gain insight into plume dynamics unattainable with static grid models.
Atmospheric pollution is a major global concern. The most recent Global Burden of Disease Study finds that in 2010 over 3 million deaths and nearly 80 million disabilityadjusted life-years were attributable to ambient particulate matter and ozone pollution (Lim et al., 2012). Accordingly, ambient particulate matter pollution ranked ninth among risk factors by attributable burden of disease in 2010. Multiple studies have identified associations between air pollution and increased mortality (Pope and Dockery, 2006).
For instance, one study finds that a 10 µg m-3 decrease in fine particulate matter concentration is associated with an increase in mean life expectancy of approximately 0.6 years (Pope et al., 2009). Furthermore, no consensus has been reached regarding the existence of a threshold for major pollutants under which concentrations would cease to have health effects (Brunekreef and Holgate, 2002).
Anthropogenic emissions have significantly increased the air pollution burden across vast regions, impacting public health and leading to a diverse set of problems which include damage to property, disruption of ecosystems and climate change. In response, legislation has been enacted throughout the world setting standards and regulations designed to limit the emissions of atmospheric pollutants and pollutant precursors and maintain pollution below concentrations selected to protect public health and welfare. In the United States, for example, the Clean Air Act was promulgated to protect air quality. The benefits and costs associated with implementing the legislation are significant. For the 1990 Clean Air Act Amendments alone, the annual costs and benefits of implementation, relative to a baseline maintained at the control levels defined by the 1970 and 1977 Clean Air Act Amendments, are estimated to reach approximately $65 billion and $2 trillion respectively in 2020 (2006 dollars) (U.S. Environmental
better understanding the sources, transport, and transformations of pollutants in the atmosphere.