Atmospheric Dispersion Modeling for Odor Buffer by wwz28639


									       Atmospheric Dispersion Modeling
for Odor Buffer Distances from Florida Landfills

                     FINAL REPORT

                       June 22, 2009

                   C. David Cooper, Ph.D.
           Professor of Environmental Engineering

                University of Central Florida

              State University System of Florida
                     HINKLEY CENTER
               4635 NW 53 Avenue, Suite 205
                    Gainesville, FL 32653

                    Report # 16207042
                              Table of Contents
Executive Summary                                                       iii

Introduction                                                           1

Objectives                                                              1

Background                                                              2

A Methodology for Predicting Odor Buffer Distances for Landfills        3

    Overview                                                            3

    Step 1: Obtain Methane Concentrations and Atmospheric Conditions    4

    Step 2: Estimate Methane Emissions                                 5

    Option: Conduct Sensitivity Study                                   7

    Step 3: Derive an Odorous Gas-to-Methane Ratio                     8

    Step 4: Choose Odor Tolerance Limits and set Methane Limits         9

    Step 5: Conduct Air Dispersion Modeling                            10

    Step 6: Generate Methane Concentration Plots                       13

Conclusions and Recommendations                                        22

Executive Summary

Odors from a large landfill can be a significant nuisance to nearby neighborhoods and
businesses. As population in Florida continues to grow and create development pressures,
housing projects creep closer to existing landfills. This research project was undertaken to
develop a detailed modeling methodology for use by counties and other landfill owners to
provide them with an objective and scientifically defensible means to establish buffer zones
around landfills.

This research project started in January of 2007. Two EPA dispersion models, AERMOD and
CALPUFF, were tested and compared, and a new modeling methodology was developed.
These models are complex, and the tasks of learning the models and comparing them took
longer to complete than first anticipated. As the project took shape, the researchers decided
that it would be necessary to obtain real-world data on emissions. Although detailed
knowledge of emissions is an essential pre-requisite for dispersion modeling, it was first
believed that the project could proceed using hypothetical emissions data. However, it later
became evident that realistic modeling depended on realistic emissions.

One of the major accomplishments of this project was the development of a new technique for
estimating emissions from landfills. This new technique is based on measuring hundreds of
ambient methane concentrations near the surface of the landfill, and combining that data with
matrix inversion mathematics to back-solve the dispersion equations. Methane emission rates
are determined at numerous locations throughout the landfill. Knowing the methane
emissions, along with the ratio of concentration of odorous gases to methane, one can then
estimate the odor emission rate, and then model the downwind concentrations of odorous
gases. In addition, the methodology that was developed in this project provides a robust way
to estimate methane emissions with less effort and perhaps more accuracy than other
methods, which is a good contribution to making an inventory of emissions of global
warming gases.

This research project resulted in a graphical means to portray odor buffer distances around a
landfill. The graphical tool shows isopleths of “odor” concentrations at various distances, and
color codes the isopleths into a “green-yellow-red” scheme (analogous to a traffic signal) that
depicts zones where homeowners should be safe from odors, where they may experience
some odors, or where they likely will experience odors. The “likelihood” can be quantified by
selecting the Nth highest concentrations to form the plot. In this study, N was very
conservatively selected as 4; the 4th highest concentration in a year corresponds to a 99.95%
probability of not exceeding that concentration at that distance in any future year.

The graphical tool depends on having an estimate of the rate of emissions of odors (or the rate
of emissions of methane and a ratio of odors-to-methane), and at least one year’s worth of
hourly meteorological data (wind speed, direction, and stability class). The meteorological
data can be obtained with relative ease for most locations in Florida; however, the emission
data must be obtained from on-site measurements for any given landfill. Uncertainties in the
methodology reflect the uncertainties and variability in emissions more so than the statistical
variations in the meteorology.


Odors from a large landfill can be a significant nuisance to nearby neighborhoods and
businesses. As population in Florida continues to grow and create development pressures,
housing projects creep closer to existing landfills. Even though the landfill may have been
there “first,” homeowners in new developments may still complain vigorously about odors.
Counties need an objective and scientifically defensible means to establish buffer zones to
prevent future housing from coming too close. With an accurate modeling tool to help predict
appropriate buffer distances, local governments will have more ammunition to help preserve
these buffer zones. This research project was undertaken to develop a detailed modeling
methodology (and ultimately a modeling/screening tool) for use by counties and other landfill
owners to provide them with an objective and scientifically defensible means to establish
buffer zones around landfills. Such buffer zones provide more time and space for dispersion
and dilution of odorous gases that drift away from a landfill.

The recent and projected population growth in Florida has put more development pressure on
areas surrounding landfills. In the past, there were often large distances between landfills and
the nearest homes, but in recent years, development has been getting closer and closer to
landfill boundaries. Examples of such landfills in Florida include Orange County, GEL,
Holmes, DeLand, Bass Road, Grantham, and Pine Ridge. After moving into a new
development, homeowners may later complain about odor events caused by odorous
compounds emitted from landfills. Odor events tend to increase in frequency and intensity as
one gets closer to the source, and housing is getting closer to existing landfills.

For the past twenty five years or so, the workhorse of dispersion modeling has been the EPA
program ISCST. In November, 2006, ISCST was replaced officially by EPA with AERMOD
(American Meteorological Society/Environmental Protection Agency Regulatory Model) as
the EPA-preferred regulatory model. Both ISCST and AERMOD are steady-state Gaussian
models, which predict better over smooth simple terrain, over shorter distances, and for
longer averaging times. But AERMOD has significantly better scientific algorithms to
describe dispersion than ISCST. Because landfills in Florida are often built up significantly
higher than surrounding terrain, and because odors are very short-term phenomena, the
simpler and easier-to-use steady-state Gaussian models may not be the best to portray
potential odor events. A relatively newer model, CALPUFF, which has been endorsed by the
US EPA as a “preferred” model in mountainous terrain, and for longer distances, was
investigated as part of this research. Also, CALPUFF is significantly harder to learn and use
than AERMOD.


The main objectives of this research were to develop a dispersion modeling approach to
predicting odor transport from landfills, and to develop a graphical screening tool for
predicting odor buffer distances around landfills. The purpose of the technique (and tool) was
to provide solid waste managers an impartial and technically-sound methodology to predict
and justify appropriate buffer distances around landfills. This was a multi-year research
project; the objectives were as follows:
     Develop a modeling methodology for estimating odor emissions from landfills, using
      ambient methane concentration measurements,
     Comparing AERMOD, CALPUFF, and ISCST to predict odors and appropriate odor
      buffer distances around landfills, choose the most appropriate model,
     Develop and demonstrate a simplified modeling methodology for one selected landfill
      in Florida,
     Publish papers and transfer the technology.


Human reactions to odor depend on the type and concentration of odorous compounds in the
air, the duration and frequency of encountering high concentrations of odorous compounds,
place of exposure, time of day of exposure, and personal odor sensitivity. If odors are
encountered at home in the evening or on weekends, the annoyance factor may be higher than
if the same odors were encountered elsewhere or at other times. The main determining factors
that influence the concentrations of odors encountered at any particular place or time are the
odor emission rate at the source, the distance from the odor source, the local meteorology
(including wind speed, wind direction, temperature, and atmospheric stability class), the time
variability of meteorological parameters, and the local topography.

Modeling involves many assumptions, and given those assumptions and the inherent
limitations of mathematical models, one might ask questions like: “Why bother conducting
computer modeling?” or “Why not simply measure what we want to know?” In answer to
these questions, one must keep in mind the following three truths. First, measuring odor
impacts is notoriously difficult and expensive, and often unreliable. Second, computer
modeling to predict odor concentrations in neighborhoods and other places near existing
landfills can be an impartial and technically-sound tool for solid waste managers. Third,
modeling is the only way to predict impacts before they happen. That is, with modeling we
can test the potential impacts of proposed new landfills or new expansions of the landfill. Or
we can use modeling to help justify the purchase of surrounding land to be used as a buffer
distance before that land is developed for residential housing.

Thus, a number of good reasons for computer modeling can be enumerated, including:
    It provides impartial, reproducible, and quantitative estimates of odor concentrations at
      many points throughout the domain of interest
    It allows for evaluation of numerous factors individually and in concert
    It allows for numerical experiments, which would be cost-prohibitive or impossible in
      the real world
    It allows for “worst-case” scenario planning
    It allows for evaluation of potential source reduction steps
    It is substantially lower in cost than ambient air quality monitoring studies, and
    It is the ONLY way to evaluate the impact from systems that are not yet built.

The traditional air dispersion modeling approach endorsed for many years by the US EPA has
relied on traditional steady-state Gaussian plume models to predict the hourly concentrations
of conservative pollutants at various distances from the source. For more than 25 years, the
workhorse model in the United States has been ISCST and its refinements. Recently, EPA has
introduced a next-generation model, AERMOD which has incorporated better boundary-layer
physics to describe the atmospheric dispersion processes. However, both are still steady-state

models that predict a 1-hour (or longer) average concentration of the target pollutant. Even
with these models, the results are limited because people can detect and be annoyed by odors
in a very short (two to five minutes) period of time. CALPUFF is a recent model that has an
apparent advantage: it is a non-steady-state model. It has been endorsed by the U.S EPA for
longer-range dispersion and for use in complex terrain.

A Methodology for Predicting Odor Buffer Distances for Landfills


A major effort was undertaken during this project that had not been anticipated at the start.
This new work came about when it proved necessary to develop a means for estimating gas
emission rates from a landfill before we could apply the dispersion model. This is a very
difficult problem, and the researchers spent considerable time addressing this. The authors
developed an exciting new methodology to solve this problem. The new methodology uses
hundreds of ambient air methane measurements that can be obtained by one person in a few
hours doing a walking survey. In our view, this new method is exciting due to its ease of
implementation in field measurements. The authors believe that it can replace the more
laborious and costly methods using flux chambers, and will prove to be more accurate and
more robust. A peer-reviewed presentation was given on this topic at the annual AWMA
conference held in June in Portland, Oregon, in June 2008. Furthermore, two papers were
written and submitted to two different peer-reviewed journals. Both have been accepted to
appear in print in 2009 (see Appendix 1).

A methodology for modeling odors from any particular landfill in Florida, and then
determining odor buffer zones around that landfill has been developed. The methodology
relies on the measurement of ambient methane concentrations at many locations near the
surface of the landfill and using those measurements to estimate methane emissions. Next,
through on-site measurements or other estimation technique, the ratio of the principal odorous
gas to methane is determined for the landfill. From the literature, odor limits are determined
for that gas, corresponding to “red, yellow, or green” limits, where red means a concentration
high enough to definitely be a nuisance, green means a concentration small enough so that
there is little or no chance of smelling it, and yellow is a range in between. Once these limits
are established, dispersion modeling is conducted and distances from the landfill are
determined where these limits occur. The dispersion modeling relies on at least one year’s
worth of meteorological data from a site (such as an airport) near the landfill being modeled.
Finally, plots are generated that display the odor buffer distances. The methodology is
summarized in Table 1, and each step is discussed in more detail following the table.

Table 1. Methodology for Establishing Odor Buffer Distances for a Florida Landfill

            Obtain Methane           Either by sampling or from existing reports of sampling,
 Step       Concentrations           obtain methane concentration data and estimate the
  1         and Atmospheric          atmospheric conditions (wind speed, wind direction, and
            Conditions               stability class) that were present at the time of sampling.

                                     Estimate methane emissions using the inverse modeling
 Step       Estimate Methane
                                     method developed in this study. Optional – conduct
  2         Emissions
                                     sensitivity study.

            Derive Odorous
            Gas-to-Methane           Sample landfill gas to estimate odorous gas content and
            Ratio                    derive odorous gas/methane ratio.
                                     Use the odor/methane ratio to calculate projected methane
                                     concentrations corresponding to red, yellow, and green
 Step       Set Limits               “odor threat levels”. (Note suggested limits for hydrogen
  4                                  sulfide are 1-hour H2S concentrations of >30 ppb for red, 15-
                                     30 ppb for yellow, and any concentrations under 15 ppb for

                                     Gather meteorological data and run AERMOD to determine
            Conduct Air
 Step                                the distances where these maximum concentrations will
  5                                  occur. (Note-4th highest concentrations represent the 99.95th

                                     Using limits set in step 4 and results from air dispersion
 Step       Generate Plots           modeling generate colored plots establishing odor buffer

A more detailed discussion of each step introduced below is provided in the following pages,
along with a test study done on the Seminole County Landfill (SCL).

Step 1: Obtain Methane Concentrations and Atmospheric Conditions

Since many landfills are required to conduct quarterly surface VOC monitoring, the
researchers believed that these existing reports could be used to estimate methane emissions.
VOC monitoring produces numerous ambient air VOC measurements (between 350 – 450
individual concentrations above the surface of the landfill), and essentially 99% of the VOC is
methane. Using these reports not only provides a large number of methane concentrations for
use in calculating methane emissions, but also helps ensure that no unusually large methane
concentrations (hot spots) are missed. If there is a large methane concentration recorded, this
could indicate a leak in the biogas collection system or a part of the landfill that has
developed a crack in the covering. This would not only reveal where a problem is occurring,
but also could be a major source of where odors may be coming from.

Each quarterly surface VOC monitoring report should come with a NSPS Surface Emissions
Monitoring Calibration and Pertinent Data Form. This form is used to record local
meteorological conditions that were observed at the time of monitoring. These observations
are important for dispersion modeling. The form has weather observations such as wind
speed, wind direction, barometric pressure, air temperature, and other general weather
conditions. All these observations are taken by the field service surveyor at the time of
monitoring at the landfill. If the NSPS Surface Emissions Monitoring Calibration and
Pertinent Data Form is not available, the National Weather Service (NWS) archives of the
hourly measurements of atmospheric conditions that existed on the day of monitoring (as
recorded at a nearby airport) can be used.

Three separate cases of quarterly surface emissions monitoring reports for the Seminole
County landfill were selected for use in this study, namely 4th Quarter 2006, 2nd Quarter 2007,
and 2nd Quarter 2008. In each case, the data consisted of several hundred VOC
concentrations, a NSPS data form, and a route map of the walking survey from which the
research team estimated the coordinates of each measured concentration. In addition,
atmospheric stability class was estimated from NWS records for that same day from the
Sanford-Orlando airport.

Step 2: Estimate Methane Emissions

Using hundreds of ambient air VOC concentrations and their locations within a MSW landfill
(recorded as described in Step 1), methane emissions from the landfill can be estimated. The
measurement locations are set as receptors, and numerous other locations are chosen as point
source locations. These locations can be chosen by a person (expert selection), taking into
account the wind direction and the receptor locations, as described by Figueroa t. al. (2009),
or can be selected using Voronoi diagrams, as discussed in detail by Mackie and Cooper
(2009). (Note- both papers have been accepted for publication, and one just recently has
appeared in print - see Appendix 1). Standard Gaussian dispersion equations are written for
each source-receptor pair, then the equations are inverted and solved by matrix methods, to
determine the methane emission rates at numerous points throughout the landfill as described
in Figueroa et. al. (2009). The total emissions are estimated by summing those from all the
points. But additional information is available: the emissions are quantified with respect to
location within the landfill.

After determining methane emission rates from the landfill, forward-modeling was done
using ISCST to see how well our calculated emission rates could predict the measured
methane concentrations for a particular monitoring event. Figure 1 is a scatterplot of the
modeled methane concentrations vs the actual measured concentrations for one monitoring
event using the estimated methane emissions obtained by the method developed in this
research. The fit is quite good, indicating that this method gives decent spatial resolution of
the emission estimates.

                                                                                                   y = 1.355x
  Predicted Methane Concentration (PPM as methane)
                                                                                                   R2 = 0.886
                                                                                     y = 2x

                                                      400                                                       y=x

                                                                                                                            y = 0.5x



                                                            0           100            200           300              400              500
                                                                          Measured Ambient VOC Concentration (PPM as Methane)

Figure 1. Scatterplot of Predicted vs. Measured Methane Concentrations, 2nd Quarter

However, it was discovered that the estimate of total emissions as well as the locations of the
maximum emissions can vary from one data set to the next for the same landfill. Although
landfilling operations can certainly change locations over time, it is also recognized that the
estimate of total emission is very dependent on accurate measurements of the meteorological
parameters. The estimates of the total landfill emissions based on the three separate data sets
from the Seminole County Landfill are presented in Table 2.

Table 2. Summary of Meteorological Parameters and Estimated Total Emissions for
three Quarterly Monitoring Events

                                                                              2006 4th Quarter       2007 2nd Quarter           2008 2nd Quarter
                                                                              Total Emissions        Total Emissions            Total Emissions
                                                       Wind Angle                    130                    40                         150
                                                       Wind Speed
                                                           (m/s)                    0.89                     1.34                        2.24
                                                      Stability Class                B                        B                           B
                                                      Concentration               3.1 ppm                  2.45 ppm                    1.82 ppm
                                                     Total Emissions
                                                           (g/s)                   608.7                    707.5                       1233.4

Note that while there is good agreement between the first two cases, the third case has a
significantly higher estimate. One possibility is that our estimate of stability class was
incorrect. Had the stability class been selected as Class C instead of B, the emissions estimate
would have been 766 g/s, about one half of the estimate reported in Table 2, and much closer
to the estimates derived from the other two quarters. This shows the sensitivity of the method
to the meteorological parameters, which topic is discussed further in the next paragraph.
Another possible explanation is the fact that the gas well field was being worked on and
expanded during the second quarter of 2008, and several leaks and/or blockages developed
that resulted in much more methane escaping during that period of time.

Option: Conduct Sensitivity Study

The Gaussian equation, the basis for solving for the methane emission rates, relies heavily on
meteorological parameters, such as wind speed, and the horizontal and vertical dispersion
(spread) functions. The values used are based on the meteorology (especially stability class)
at the time of the monitoring event. But since the monitoring event can last several hours, and
since there are no hourly on-site meteorological data provided at the landfill, the results can
be greatly affected if the conditions are not measured accurately or if they change during the
time of monitoring.

A sensitivity study on these meteorological parameters was performed to quantify the
potential effects on the total emissions predictions. For this study, the 2nd quarter 2007
monitoring report was used and the conditions that were estimated at the time of this survey
were used as the initial conditions (Table 3). These values were changed one at a time to
determine the sensitivity of each variable in the Gaussian equation.

Table 3. 2007 2nd Quarter Monitoring Event Base Case Conditions

                            Wind Angle, deg from north          40
                            Wind Speed, m/s                    1.34
                            Stability Class                     B
                            Background Concentration,
                            ppm                                2.45
                            Total Emissions (g/s)             707.5

The sensitivity study was conducted and the results can be seen in Table 4. This table shows
how much the total emissions of the landfill are affected by each of the parameters, especially
stability class. The wind speed correlates well when comparing the percent change in the
input parameter and percent change in predicted emissions; however, as wind angle and
stability class change, the predicted emissions vary considerably. This study demonstrated the
need to get accurate on-site measurements of meteorology as well as good ambient air
methane measurements.

Table 4. Sensitivity Study of Meteorological Conditions, using 2007 2nd Quarter data

                                                         Total       % Change in
                                         % change      Emissions      Predicted
               Wind Speed (u) (m/s):     from Base       (g/s)        Emissions
                          1                  -25.4%        528.0          -25.4%
                         1.25                 -6.7%        660.0           -6.7%
                     1.34 (base)               0.0%        707.5            0.0%
                         1.5                  11.9%        791.9           11.9%
                          2                   49.3%       1055.9           49.3%
                         2.5                  86.6%       1319.9           86.6%
                  Stability Class:
                          D                   -33.3%        281.3          -60.2%
                          C                   -16.7%        415.1          -41.3%
                      B (base)                  0.0%        707.5            0.0%
                          A                    16.7%       1302.7           84.1%
             Theta (θ) (Direction wind
                          20                 -11.11%        820.9           16.0%
                      40 (base)                0.00%        707.5            0.0%
                          60                  11.11%        794.8           12.3%
                         320                  44.44%        978.2           38.3%

Step 3: Derive an Odorous Gas-to-Methane Ratio

In this study, the odorous gas was assumed to be hydrogen sulfide (H2S) – a typical odorous
gas that is emitted from landfills. It is recognized that landfills emit many odorous gases, and
that each landfill must be studied to select the most representative gas or combination thereof.

The best way to derive the H2S-to-methane ratio would be to actually sample the ambient air
H2S concentration at various points during the walking survey of the monitoring report, using
a portable H2S analyzer that could measure H2S to levels as close as 0.1 ppb. If both the H2S
and methane concentrations were sampled at various points during the walking survey, a set
of those data could be used to provide an accurate H2S-to-methane ratio for that particular
landfill. The same theory could be applied to any odorous gas at a landfill to derive a useable

The above method is costly and time consuming (if it can be done at all), and was not done
for this study. Therefore, further research of the literature was undertaken to determine a
reasonable range of H2S-to-methane ratios (see Table 5). The ratios reported were all for
whole landfill gas (samples extracted from below the surface), and may not be representative
of the ratio that might exist in ambient air (due to absorption or oxidation of H2S by the cover
soil as the gas migrates upwards).

Table 5. H2S-to-CH4 Ratios from Literature Review

                                                                        H2S /
                                 Hydrogen                               CH4
                                  Sulfide          Methane              Ratio
                Case 122          63.3 ppm       540000 ppm            0.00012
                Case 2
                                  70.0 ppm       500000 ppm            0.00014
                Case 324
                                  900 ppm        730000 ppm            0.00123
                Case 425
                                  20 ppm         532830 ppm            0.00004
                Case 58
                                 10000 ppm       450000 ppm            0.02222
                Case 627
                                 247.8 ppm        286000 ppm           0.00087
                Case 727
                                 115.3 ppm        585000 ppm           0.00020
                Case 827
                                 2344 ppm        316000 ppm            0.00742

Table 5 has eight different cases, and displays a wide range of values for the H2S-to-CH4
ratio, each determined from a sample of whole landfill gas (from the subsurface). Each case
assumes that the whole gas comprises 1 million ppm (considering all the components of
landfill gas as methane, carbon dioxide, nitrogen, oxygen, H2S, and other trace chemicals).
Cases 1, 2, 4, 6, and 7 are representative of “average” MSW landfills. Cases 3, 5, and 8 are
worst-case scenarios that show very high ratios. In fact, case 5 was not based on
measurements, but rather was based on an assumption that H2S accounted for all the trace
chemical species, which is a greatly exaggerated, highly conservative approach based on what
is known about MSW landfill gas and what was found in the literature review. It was decided
for this study that an H2S-to-CH4 ratio of 0.00013 (the average of cases 1, 2, 4, and 7) would
be used as an example for this work. It is emphasized that actual H2S measurements were not
made at the Seminole County Landfill.

Step 4: Choose Odor Tolerance Limits and set Methane Limits

The ratio calculated from Step 3 was used to determine the projected methane concentration
limits to correspond to desired odor tolerance limits. Equation (1) below can be used to do
this once a ratio is determined, or equation (2) can be used if you have the exact percentage of
methane and H2S in the landfill.

               Methane Limit ( ppm) 
                                         H 2S                 1000                         (1)
                                         CH   Ratio                        
                                            4         ( H 2 S Limit ( ppb)) 

                                         %CH 4 ( H 2 S Limit ( ppb))
                Methane Limit ( ppm)                                                       (2)
                                         %H 2 S        1000

To use equation (1), one must know the applicable detection and annoyance levels for the
odorous gas (in this case, H2S). The limits chosen for this work are H2S concentrations of
above 30 ppb, 15-30 ppb, and under 15 ppb. The 30 ppb level is a 1-hour maximum that has
been used in California and Massachusetts regulations. The 15 ppb level has been used in
Massachusetts for an 8-hour averaging time. These can be used to determine the methane
concentration limits of any landfill, if H2S is the offensive gas, and if the H2S-to-methane
ratio is known. These limits were used in this study to generate the plots for odor buffer
distances in Step 6. Using Equation 1, and the H2S limits of above 30 ppb, 15-30 ppb, and
under 15 ppb for three levels of odor exposure, and assuming the aforementioned H2S-to-CH4
ratio, corresponding methane concentration limits were developed as listed in Table 6.

Table 6. Methane Concentration Limits (assumes an H2S-to-methane ratio of 0.00013)

                        H2S Limit            Methane Concentration
                         <15 ppb                  <115 ppm
                        15-30 ppb                115-231 ppm
                         >30 ppb                  >231 ppm

Tables similar to this can be created for other odors if it is determined H2S is not the main
contributor to odors at a certain landfill. Converting the particular odor limits into
corresponding methane concentrations also helps when generating plots because instead of
making multiple plots for different odors, one can simply look up the corresponding methane
concentration limits for that odor and utilize just one plot of methane concentrations for each

Step 5: Conduct Air Dispersion Modeling

Based on a comparison of AERMOD, ISCST, and CALPUFF, the research team selected
AERMOD as the best model for this kind of air dispersion modeling of odors. One of the
comparisons is shown in Table 7.

Table 7. Comparison of CALPUFF and AERMOD Results at the 5 ppb criterion

                                       CALPUFF                        AERMOD
  Ring            Total       Receptors in   Percent of      Receptors      Percent of
  Distance, m     Receptor    Exceedance     Exceedances     in             Exceedances
                  Hours                                      Exceedance
    800           315360              1264          0.40%             613          0.19%
    900           315360               813          0.26%             313          0.10%
   1000           315360               543          0.17%             145          0.05%
   1100           315360               390          0.12%              67          0.02%
   1200           315360               315          0.10%              26          0.01%
   1400           315360               162          0.05%               6          0.00%

It is obvious that CALPUFF gave higher concentrations than AERMOD. However, without
real data, it is impossible to tell which model is more accurate for Florida landfills. Based on
our literature review, AERMOD was judged to be better than CALPUFF for shorter and
intermediate distances. Furthermore, it was deemed to be easier to use, and for this and other
reasons, was chosen for this work in modeling odor dispersion from Florida landfills.

For the dispersion modeling, one entire year’s worth of meteorological data was used. The
output concentrations from AERMOD were recorded at 300 receptors, and then the results
were used to generate plots of the output concentrations as described below in Step 6. The 4th
highest ranked concentrations were used to generate the plots. The 4th highest has been used
by EPA in setting ambient ozone standards. It is more statistically stable than the 1st, 2nd, or
3rd highest, but is still extremely conservative. The 4th highest is out of 8760 hourly data
points for one year, at each receptor, and thus represents an extremely low probability (4 out
of 8760) of being exceeded. In other words, 99.95% of all the modeled concentrations were
less than this number. The 4th highest was used as a way of being extra conservative in
predicting odor buffer distances. If the 10th highest were to be used, the concentrations would
be lower, and the plots would show buffer distances that would be closer to the landfill.
However, the 10th highest (10 out of 8760) would still be very conservative, representing the
99.88th percentile.

The meteorological data gathered were from January 1, 1999 to December 31, 2003. The
hourly surface data are from the Orlando International Airport and the upper air data are from
Ruskin, FL. The meteorological data for the year of 2001 were used for this research, but any
of the four years could have been used. A 7.5-minute horizontal datum DEM file for
Seminole County was used creating two output files, one for the sources and one for the
receptors. The receptor file was used when running AERMOD since the terrain outside the
landfill is relatively the same as when the DEM file was created, but the source file was not
used since the landfill terrain is constantly changing (as more and more waste is deposited).
Instead, the topographic maps provided with the Surface Emission Monitoring reports were
used to get reasonably accurate estimates of the terrain inside the landfill boundaries.

The input file was set up with an averaging time of 1-hour and the pollutant modeled in this
case was methane. The source pathway specified the number of sources, the emission rate for
each source, and the base elevation of each source. AERMOD was then run (with 2001
meteorological data) and the 4th highest concentrations of methane in (µg/m3) were recorded
at all 300 receptors, input into Excel, and then converted to parts per million (ppm). Receptors
(numbers 1-252) were placed on rings (36 receptors on each ring – 1 every 10 degrees) at
distances of 800m, 900m, 1000m 1100m 1200m, 1300m, and 1400m from the center of the
landfill. Receptors 252-300 were chosen to represent the fence line of the landfill (see Figures
2 and 3).

Figure 2. The landfill showing ring distances and wind angles modeled.








              490000 490500 491000 491500 492000 492500 493000 493500

Figure 3: Schematic Diagram showing Receptors Surrounding the Landfill

Step 6: Generate Methane Concentration Plots

The limits set in Step 4 and results from air dispersion modeling were used to establish odor
buffer distances. The odor buffer distances were separated into three different zones,
identified by colors of red, yellow, and green. The colors represent suggested areas around the
landfill where odors may occur at different levels. The red zone is a “do not build zone” and
corresponds to H2S concentrations above 30 ppb (or, in this case, methane > 231 ppm). The
yellow zone is a “proceed with caution” zone and corresponds to H2S levels from 15-30 ppb
(or methane from 115-231 ppm). The green zone is a “safe to build zone” and corresponds to
H2S concentrations under15 ppb (or methane < 115 ppm). Shadings within each color indicate
the gradation of concentrations within each range.

H2S was assumed to be the odorous substance in this study, and the corresponding methane
concentration limits from Table 6 (based on the assumed H2S-to-CH4 ratio of 0.00013) were
used to create the plots for each quarter. Of course, H2S is not the only odorous compound
found in landfill gas, and the ratio of the prevalent odorous compound to methane may well
change from one landfill to the next. Thus, Table 6 is only valid for this study of H2S, and
only for the assumed ratio of 0.00013. Three-color plots depicting odor buffer distances are
shown in the following pages, based on the emissions estimates for the three data sets: 4th
Quarter 2006, 2nd Quarter 2007, and 2nd Quarter 2008. These color plots were created using
Surfer, a software that is used for contouring and 3D surface mapping. In this case, the
software took the X and Y coordinates of the receptors along with the concentrations at those
receptors (from AERMOD) to form contours around the landfill, which were then color coded
for the three different ranges previously discussed. Note that the dispersion modeling results
differ for each quarter, even though the meteorological data used for modeling the whole year
were identical. The differences arise due to both magnitude and locations of the estimated
emissions. Even if the magnitudes for different data sets are nearly constant, the locations
may well change due to changes in where the operations personnel are emplacing waste at
any given time, or perhaps due to subtleties contained in the inverse modeling technique.

For each quarter’s data, the isopleths of the estimated methane emissions in grams/sec are
shown in figures below, each directly followed by the color plot of the buffer distances. These
isopleths of methane emissions give some clues to where the greatest concentrations of odors
may occur even before air dispersion modeling is done (most likely, near the edge of the
landfill that is closest to the areas of high emissions). As an example, based on Figures 4 and
5, it can be seen that the highest concentrations of odors would be expected just outside the
northern fence line, nearest the highest emissions within the landfill. Figures 6 and 7 (based
on the 2nd Quarter 2007 monitoring data) also depict the relationship between locations of
emissions and locations of high methane concentrations outside the landfill. But note that in
Figure 6, the “hotspot” of emissions has shifted more to the east, and the corresponding
modeled high methane concentration is now east of the landfill as shown in Figure 7.










            491200   491300   491400   491500   491600   491700   491800   491900   492000

Figure 4. Methane Emission Isopleths, based on 4th Quarter 2006 data.


                                                                               250 ppm
                                                                               240 ppm
                                                                               231 ppm
                                                                               220 ppm
                                                                               210 ppm
                                                                               200 ppm
3185500                                                                        190 ppm
                                                                               180 ppm
                                                                               170 ppm
                                                                               160 ppm
                                                                               150 ppm
                                                                               140 ppm
                                                                               130 ppm
3185000                                                                        120 ppm
                                                                               115 ppm
                                                                               110 ppm
                                                                               100 ppm
                                                                               90 ppm
                                                                               80 ppm
3184500                                                                        70 ppm
                                                                               60 ppm
                                                                               50 ppm
                                                                               40 ppm
                                                                               30 ppm
                                                                               20 ppm


            490500      491000      491500       492000      492500   493000

Figure 5. Odor Buffer Zone Plot, based on 4th Quarter 2006 data









          491200   491300   491400   491500   491600   491700   491800   491900   492000   492100

Figure 6. Methane Emission Isopleths, based on 2nd Quarter 2007 data



                                                                                            600 ppm
                                                                                            560 ppm
                                                                                            520 ppm
                                                                                            480 ppm
3185500                                                                                     440 ppm
                                                                                            400 ppm
                                                                                            360 ppm
                                                                                            320 ppm

3185000                                                                                     280 ppm
                                                                                            240 ppm
                                                                                            220 ppm
                                                                                            180 ppm
                                                                                            140 ppm
                                                                                            100 ppm
                                                                                            60 ppm
                                                                                            20 ppm


             490500        491000        491500        492000        492500        493000

Figure 7. Odor Buffer Distance Plot, based on 2nd Quarter 2007 data.

The previous two quarters’ data present a good example of how estimated emissions within
the landfill can change and can affect the odor buffer distances and locations. Comparing
Figures 4 and 6, and Figures 5 and 7, one can see that the area of highest emissions, has
shifted to the east, and resulted in the odor buffer distances shifting in the same direction. In
addition, the intensity of emissions has increased, resulting in higher odor concentrations
further away from the edge of the landfill. Modeling numerous quarters of data may give a
greater level of confidence for determining where odors are likely to occur in the future.
Finally, plots based on the third quarter of data (2nd quarter of 2008) are presented in Figures
8 and 9. It is noted that for this quarter the estimated emissions are almost twice as high as
with previous quarters’ data. The emissions were modeled as being more spread out, so the
the “red zone” in Figure 9 covers more area. Also, due to the larger total emissions, the size of
the red zone extends further from the landfill boundaries.










           491200   491300   491400   491500   491600   491700   491800   491900   492000   492100

Figure 8. Methane Emission Isopleths, based on 2nd Quarter 2008 data.


                                                                                             600 ppm
                                                                                             550 ppm
                                                                                             500 ppm
                                                                                             450 ppm
                                                                                             400 ppm
                                                                                             350 ppm
                                                                                             300 ppm
                                                                                             250 ppm
                                                                                             231 ppm
                                                                                             200 ppm
                                                                                             150 ppm
                                                                                             115 ppm
                                                                                             100 ppm
                                                                                             50 ppm
                                                                                             0 ppm


             490500        491000         491500        492000        492500        493000

Figure 9. Odor Buffer Distance Plot, based on 2nd Quarter 2008 data.

It is noted that for a future landfill, one might be able to estimate total emissions from theory,
but it would be highly unlikely that one could predict where within the landfill the emission
hotspots will be. Figure 10 displays the odor buffer plot based on the same total emissions as
the second quarter of 2007. However, this time the emissions are spread uniformly over the
landfill surface. Note that there are essentially no red zones around this uniformly-emitting
landfill. Compare Figure 10 to Figure 7 to see the significant difference that the assumption of
uniform emissions makes in predicting odor buffer distances.



                                                                                           150 ppm
                                                                                           145 ppm
                                                                                           140 ppm
                                                                                           135 ppm
                                                                                           130 ppm
                                                                                           125 ppm
3185500                                                                                    120 ppm
                                                                                           115 ppm
                                                                                           110 ppm
                                                                                           105 ppm
                                                                                           100 ppm
                                                                                           95 ppm
3185000                                                                                    90 ppm
                                                                                           85 ppm
                                                                                           80 ppm
                                                                                           75 ppm
                                                                                           70 ppm
                                                                                           65 ppm
                                                                                           60 ppm
3184500                                                                                    55 ppm
                                                                                           50 ppm
                                                                                           45 ppm
                                                                                           40 ppm
                                                                                           35 ppm


             490500        491000       491500        492000        492500        493000

Figure 10. Odor Buffer Plot, based on 2nd Quarter 2007 - Uniform Emissions

Determining Final Odor Buffer Distances

In determining the final odor buffer distances around a landfill, some judgment will be
required. The color-coded plots presented earlier should be used as a starting point. There are
also other tools that can help determine appropriate levels of risk incurred with the distance
that will be set in creating an odor buffer zone. One of these tools is the probability of a
methane concentration (which can be linked to the odor concentration once a ratio is
determined) being less than a certain value. For any given receptor, the concentrations from
AERMOD (one for each hour of the modeled year) can be plotted on a cumulative probability
scale. This is extremely tedious and should only be done for a few receptors. Figure 11 shows
an example of this for the Seminole County Landfill. Figure 12 replots the same data, but
restricts the plot to the top 100 points (instead of all 8760 points). Once the risk-distance
relationship is understood, decision makers can then decide on an appropriate probability of
exceeding the limiting concentration to be used in determining the buffer distances.

Figure 11. Single Receptor Concentration-Probability Plot, based on 2nd Quarter 2007.

                                                              Probability of Being Less than a Certain Value

                                1 50

                                1 40

                                1 30

                                1 20
 Methane Conce ntration (ppm)

                                1 10

                                1 00





                                 9 8.00%   98.20%   98.40 %    98.60%    98.80%    99 .00%      99.20%   99.40 %   99.60%   99.8 0%   100.00%

Figure 12. Single Receptor Concentration-Probability, Top 100 Points (2nd 2007 data)

Using all the tools discussed above, suggested buffer distances were derived. Using the SCL
example as discussed in this report, with all the assumptions and limitations described earlier,
the suggested buffer distances for this landfill are presented in Table 8. Choosing the 2nd
quarter, 2007, as the most representative data set, a buffer distance of 250 m (about 800 ft) is
recommended. This very conservative approach ensures that almost all (99.95%) of the time,
odors would be below the annoyance level at distances outside the 250-m zone around the
landfill. Also, odors would be below detectable limits at distances greater than 700 m).

Table 8. Odor Buffer Distances for the Landfill

                            Based on      Based on      Based on              Final
                          4th Qtr, 2006 2nd Qtr, 2007 2nd Qtr, 2008         Proposed
         H2S >30 ppb            0m            250 m           700 m           250 m

        H2S 15-30 ppb          200m         250-700 m        >1000 m        250-700 m

         H2S <15 ppb          >200 m          >700 m            n/a           >700m

The distances in the table represent distance from the edges of the landfill. If one could accept
a little greater risk, these odor buffer distances could be moved in even closer to the landfill.
For example, if we used the 10th highest concentrations instead of the 4th highest
concentrations, then the buffer distances would be closer to the landfill, but still 99.88% of
the time odors would not exceed the critical concentration. This is still a very high probability
and may be acceptable to decision makers. Once a final probability is chose, the buffer
distances can be further refined. If land is particularly expensive around the landfill being
studied, decision makers may wish to accept a higher risk of public complaints about odors.

Conclusions and Recommendations

A methodology has been developed to predict odor buffer distances around landfills. The
odor buffer distances are displayed as easy-to-use “red-yellow-green” color areas. The red
zone was defined as any one-hour concentration of H2S greater than 30 ppb, and was
determined from methane modeling using an assumed H2-to-methane ratio of 0.00013. The
buffer distances can be calculated based on different risks of occurrence. Such prediction
requires information specific to each landfill, namely odorous gas emission rates (or ratios to
methane) and local meteorology.
It is recommended that any landfill in the state that wishes to prepare such buffer distance
charts undertake a site-specific study to estimate odor emission rates. The appropriate
meteorological data likely are available for a location near any landfill. Also, it is
recommended that decision makers choose the appropriate probability or risk prior to
undertaking the study, so that the resulting buffer zones can be quantified more precisely.

                               Appendix 1.
   Titles of Papers accepted for Publication in Peer-Reviewed Journals
             (note – pdf copies of papers can be obtained from the authors)

1. Figueroa, V. K., Mackie, K. R., Guarriello, N., and Cooper, C. D. “A Robust Method for
   Estimating Landfill Methane Emissions,” accepted for publication in the Journal of the
   Air & Waste Management Association, to appear in August 2009.

2. Mackie, K.R. and Cooper, C. D. “Landfill gas emission prediction using Voronoi
   diagrams and importance sampling,” Environmental Modelling and Software, 24, 1223-
   1232) June 2009.


To top