Grantee Research Project Results
2001 Progress Report: Improved Simulation of Advection and Dispersion of Urban Air Toxics
EPA Grant Number: R827929Title: Improved Simulation of Advection and Dispersion of Urban Air Toxics
Investigators: Walcek, Chris
Institution: The State University of New York
EPA Project Officer: Chung, Serena
Project Period: December 1, 1999 through December 1, 2002 (Extended to August 2, 2004)
Project Period Covered by this Report: December 1, 2000 through December 1, 2001
Project Amount: $347,991
RFA: Urban Air Toxics (1999) RFA Text | Recipients Lists
Research Category: Air Quality and Air Toxics , Air
Objective:
The objective of this research project is to further develop and refine a highly accurate and computationally efficient algorithm for simulating the advection of poorly resolved point sources of toxic pollution in urban environments. Current Lagrangian and Eulerian pollution models use advection algorithms that can under many conditions inaccurately simulate calculated impact areas, and attempts to "source apportion" measurements of toxic pollutants will contain large errors if advection and diffusion are inaccurately simulated. The approach is to first refine a recently developed, highly accurate, and computationally efficient algorithm for calculating the advection of pollutants in atmospheric models. This scheme is absolutely monotonic, mass conserving, and is capable of advecting poorly resolved features with errors that are appreciably smaller than the best algorithms used today. The innovative feature of this algorithm that enhances its accuracy is the use of a flux adjustment near local extremes of a tracer distribution to reduce numerical diffusion. Unfortunately, most algorithms including this scheme fail to accurately advect tracer distributions containing steep gradients.This algorithm can advect some sharp gradients with high accuracy, and it is hypothesized that it is possible to "adjust fluxes" near large gradients in a manner similar to the "peak" adjustment to limit numerical diffusion around steep gradients. An appropriate algorithm for identifying and correcting fluxes around sharp gradients must be derived. This highly accurate numerical advection algorithm then will be used within an urban-scale three-dimensional model of the planetary boundary layer (PBL) to simulate the transport and shear-induced diffusion of point sources of pollutants in urban areas. It is hypothesized that small amounts of vertical shear of the horizontal wind direction coupled with small amounts of isotropic turbulence will induce substantial horizontal dispersion that currently is poorly understood and simulated by urban-scale dispersion models, which assume "uniform" horizontal dispersion coefficients without recognizing that there is a preferential direction of horizontal dispersion aligned with the vertical wind shear vector. The results of this research effort will be a highly accurate numerical advection algorithm for use in many applications, as well as a more thorough understanding of dispersion within the PBL. Methods and algorithms developed by this project could be used by other models to provide more accurate exposure and risk assessments of toxic pollutants, and also could be used to improve the accuracy of source-apportionment investigations.
Progress Summary:
During the second year of this research project, efforts focused on an exhaustive numerical search and investigation to alleviate one area still deficient in accuracy in all advection algorithms, including the scheme that was successfully refined and published under funding from this research project. This problem relates to the ability of advection algorithms to advent steep gradients that are embedded in tracer distributions not near local extremes.During the first year of this research effort, it was discovered that minor adjustments of slopes of lines used in a piecewise-linear advection algorithm near local extremes vastly improves overall advection performance. Extensive efforts were devoted to determining robust Courant-number dependent slope adjustment factors that optimally improved advection performance when applied only around local extremes (maximums and minimums) of a tracer distribution. Generally applicable "slope adjustment" factors were defined empirically near local extremes. Using these relatively minor adjustments, the accuracy of a low-order, piecewise linear advection scheme was found to exceed the accuracy of the highest order schemes used at a fraction of the computational cost for virtually every test problem suggested for advection algorithm performance.
The logical next step was to use a similar approach to find slope adjustment factors that could be applied only in areas where there are abrupt gradients in tracer distribution, in addition to the already discovered adjustments around extremes.
Basically, a search was conducted for slope adjustment factors of the piecewise-linear approximations near regions of steep gradients. Unfortunately, despite an exhaustive search, robust slope adjustment factors could not be found that universally improved overall advection performance for a wide range of test problems. Part of the problem involves the method of defining "steep gradients" within in a tracer distribution. Defining local peaks is relatively simple, and during advection a local extreme remains a local extreme despite any peak reduction or numerical diffusion that occurs during advection. In contrast, defining regions of "steep gradients" is much more difficult, especially for Aileron grid models where gradients are "smoothed" out during the advection process in a manner that is appreciably different from the way peaks are "spread out" during advection. Because local extremes are preserved during advection, it is possible to "sharpen" mass around local extremes, which was the technique pursued in the first version of this algorithm. In contrast, regions of strong gradients are smoothed out or diffused in a manner that cannot be easily reversed without introducing artificial gradients in smooth areas of a tracer distribution.
Despite this unsuccessful search, it is still possible to refine the performance of the piecewise linear scheme so that it can accurately advent both local extreme gradients and local extremes. The approach is to use higher order mapping of the tracer distribution. Rather than piecewise linear, piecewise parabolic, or higher order schemes, will be used; then, slope adjust factors near local extremes for these higher order approximations will be defined. Although slightly more computationally expensive, considerable accuracy increases are anticipated.
In summary, progress on the major tasks of this research effort during this year include:
(1) Search for a method to identify and more accurately simulate embedded gradients. Despite a comprehensive and arduous search, a generally applicable method could not be defined for identifying regions of steep gradients, then adjusting slopes in the vicinity of steep gradients to improve advection performance. However, by using higher order approximations other than linear (parabolic, cubic ?), then adjusting mass round local peaks, this problem will be alleviated. Defining "slope" adjustment factors around higher order polynomial approaches following the protocol that worked for the linear order version of this algorithm will be pursued.
(2) Refinement and simplification of advection algorithm. The accuracy of this advection algorithm for advection numerous "test shapes" is usually superior to the highest order advection algorithms available today (Rather, Boat L=8), but the computational requirements of this scheme are appreciably smaller than these algorithms.
(3) Application of advection algorithm to simulate diffusion in sheared environments. A study using the improved advection algorithm was published where "typical" observed boundary layer winds are used to advent puffs of pollution within a typical PBL. Dispersion of pollution puffs is inherently "nondiffusive" and heavily controlled by small amounts of ambient shear that usually is present in the winds of the lower troposphere. In the future, longer term winds will be used to define concentrations downwind of individual point sources.
Future Activities:
The original innovation of the advection algorithm under development and testing was to identify local extremes of tracer distributions and perform minor adjustments during flux calculations to aggregate mass near the extremes to counter numerical diffusion that normally limits the accuracy of Aileron methods for assessing pollutant transport. Similar problems also occur when local steep gradients are advected. During the remaining portion of this research effort, this limitation will be addressed.(1) Try to develop a method of preserving local steep gradients. All existing advection algorithms (including the algorithm developed using EPA support) have trouble advection tracer distributions containing steep gradients. After failing to discover a technique to define and adjust mass around gradients, the already proven successful approach of peak-enhancements to higher order polynomial approaches is being used. "Sharpened" parabolic, cubic, and quartic polynomials, rather than the simple piecewise linear approach already developed will be used. Such algorithms already advent gradients considerably better than linear algorithms, and with peak-enhancements that have already been applied to the linear approach, success in creating a "sharpened" higher order approach that is capable of preserving gradients and local extremes is anticipated.
(2) Apply the advection algorithm to assess urban-scale pollutant transport and dispersion. One limitation of existing plume models used to assess pollutant impacts is that plume models can only accurately assess impacts out to 10-20 km downwind of individual point sources. As more sophisticated longer range Aileron or Lagrangian models are used to assess longer range impacts, numerous shortcomings of these models seriously limit their accuracy. Because the advection algorithm developed is computationally cheaper and significantly more accurate than existing advection algorithms, fairly long-range transport of atmospheric mercury using the Aileron approach will be explicitly simulated with the extremely accurate advection algorithm. Specifically, a series of carefully designed studies to investigate and quantify the effects of shear on pollutant dispersion will be undertaken.
(3) Begin to study long-range pollution impact assessments using refined advection algorithms. More than 150 days of extremely detailed, 80-km resolution, three-dimensional wind fields covering the domain of Eastern North America have been gathered. These model-interpolated wind fields will be used to construct long-range and longer term impact assessments of individual point sources where synoptic-scale meteorological factors become extremely important in determining maximum impacts. Annual aggregation techniques used in previous studies of acid precipitation assessments will be used to assess annual impacts.
Journal Articles on this Report : 1 Displayed | Download in RIS Format
Other project views: | All 16 publications | 3 publications in selected types | All 3 journal articles |
---|
Type | Citation | ||
---|---|---|---|
|
Walcek CJ. Effects of wind shear on pollution dispersion. Atmospheric Environment 2002;36(3):511-517. |
R827929 (2001) R827929 (2002) R827929 (Final) |
Exit Exit Exit |
Supplemental Keywords:
advection, transport, long-range dispersion, air ambient air, atmosphere, troposphere, exposure, risk, physics, engineering, environmental chemistry., RFA, Scientific Discipline, Air, air toxics, Environmental Chemistry, climate change, Chemistry, tropospheric ozone, fate and transport, urban air toxics, Lagrangian approach, urban air, stratospheric ozone, air pollutants, plumes, air quality models, ozone, climate variations, VOCs, urban air pollutants, air pollution models, circulation model, atmospheric pollutant loads, Volatile Organic Compounds (VOCs), air quality, atmospheric models, climate variabilityProgress and Final Reports:
Original AbstractThe perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.