AERMOD Sensitivity to AERSURFACE Moisture Conditions and Temporal Resolution
1 like181 views
This paper analyzes the sensitivity of AERMOD-predicted concentrations to varying surface moisture conditions and temporal resolutions across several locations in the U.S. The study finds minimal impacts on short-term concentrations from low-level releases, with more significant effects observed for elevated releases, particularly during varying seasonal or monthly conditions. The research underscores the importance of carefully selecting moisture conditions and land use parameters when using AERMOD for environmental dispersion modeling.
AERMOD Sensitivity to AERSURFACE Moisture Conditions and Temporal Resolution
- 1. Environmental solutions delivered uncommonly well AERMOD Sensitivity to AERSURFACE Moisture Conditions and Temporal Resolution Paper No. 33252 Prepared By: Anthony J Schroeder, CCM – Managing Consultant TRINITY CONSULTANTS 7330 Woodland Drive Suite 225 Indianapolis, IN 46278 (317) 451‐8100 trinityconsultants.com June 2014
- 2. 1 AERMOD Sensitivity to AERSURFACE Moisture Conditions and Temporal Resolution Paper #33252 Anthony J. Schroeder, CCM Trinity Consultants, 7330 Woodland Drive, Suite 225, Indianapolis, IN 46278 ABSTRACT This study reviews AERMOD-predicted concentrations for several hypothetical sources at locations throughout the United States. An analysis of the sensitivity of the AERSURFACE outputs and model-predicted concentrations to the moisture condition and surface land use parameter temporal period used for each location is conducted. This analysis includes a review of the effect of the use of the uniform average moisture condition versus the use of temporally varying moisture conditions (annual, seasonal, and monthly) on concentrations for each location and source type considered in the analysis. A second review focuses on the sensitivity of modeled concentrations to the choice of temporal resolution (annual, seasonal, and monthly) for average moisture conditions. For short term average concentrations from low level releases, the use of different moisture conditions on maximum modeled concentrations is minimal. The impacts of varying moisture conditions is more pronounced for elevated releases. Additionally, AERMOD users must be careful in selecting the temporal period for defining land use parameters in cases where land use types are present that have widely varying parameters based on season. INTRODUCTION The United States Environmental Protection Agency (U.S. EPA) initially released the AERSURFACE tool in January 2008 with an update released in January 2013.1 AERSURFACE is used to develop estimates of surface characteristics (albedo, Bowen ratio, and surface roughness) for use as input into AERMET to produce meteorological data sets that can be used in regulatory dispersion modeling analyses using AERMOD. AERSURFACE uses the land use characteristics of the area surrounding the meteorological data collection site to determine the appropriate surface characteristics for input into AERMET. When executing AERSURFACE, the user must choose between three categories for surface moisture (average, dry, and wet). As discussed in the AERSURFACE User’s Guide, the answer to this question dictates the value of the Bowen ratio output by AERSURFACE. U.S. EPA recommends determining the surface moisture condition by comparing precipitation for the period of data to the 30-year climatological record. If the precipitation for the period is in the upper 30th percentile, wet conditions are selected, if the precipitation is in the lower 30th percentile, dry conditions are selected, else average conditions are selected. The surface moisture condition can be varied at the same frequency that varying land use conditions can be input into AERMET (i.e., annually, seasonally, or monthly). Only one surface moisture condition can be output by AERSURFACE in a single run; therefore, if moisture condition is to be varied seasonally or monthly then several AERSURFACE runs must be
- 3. 2 completed. The appropriate values of Bowen ratio are then input to AERMET for each seasonal or monthly period such that moisture conditions for each individual season or month versus climate normal can be considered. Previous studies have shown that AERMOD results are more sensitive to variations in surface roughness than variations in albedo or Bowen ratio.2 Therefore, one question to be explored in this paper is the sensitivity of AERMOD-predicted concentrations to the use of meteorological data processed using average moisture versus moisture conditions based on a comparison for each year, season, or month against climate normal precipitation. A secondary purpose of this paper is to explore the sensitivity of AERMOD-predicted concentrations to the use of annual, seasonal, or monthly varying land use parameters all assuming average moisture versus climate normal. METHODOLOGY To determine differences in AERMOD-predicted concentrations using differing moisture conditions and temporally varying land use parameters, the current version of AERMET (Version 13350) was run using one year (2012) of hourly meteorological observations for four different surface observing stations. The stations chosen for this analysis were National Weather Service (NWS) observation stations in Albany, New York (ALB), Nashville, Tennessee (BNA), Pocatello, Idaho (PIH), and Seattle, Washington (SEA). Hourly wind observations were supplemented using AERMINUTE (Version 11325) for each surface observing station. Upper air data from a sounding site located at or near the surface observing station were also used in AERMET. The upper air stations corresponding with the four surface stations were Albany, New York (ALY), Nashville, Tennessee (BNA), Boise, Idaho (BOI), and Quillayute, Washington (UIL). Monthly precipitation data were also gathered for the period covering 1983- 2012 for each surface observing station used to define 30th upper and 30th lower percentiles by month, season, and year. These percentiles were used to determine whether each month, season, or year was average, wet, or dry versus climate normal for input to each AERMET run. Meteorological Data Processing To define the land use characteristics and micrometeorological parameters for each station, the latest version (13016) of the AERSURFACE utility was applied to a digital mapping of land use and cover in accordance with the procedures identified in the AERMOD Implementation Guide3 and the AERSURFACE User’s Guide.1 Using publicly available digital land cover datasets and lookup tables of surface characteristics that vary by season and land cover type, the AERSURFACE tool generates realistic and reproducible surface characteristics for any site of interest that can then be directly imported into AERMET for generating AERMOD-ready meteorological datasets. As recommended by the AERSURFACE User’s Guide, the land use analysis was prepared using NLCD92 data. The AERSURFACE analyses were conducted using the following user inputs: In accordance with the AERMOD Implementation Guide, the default surface roughness study area radius of 1 km was used for the surface roughness length determination. Surface characteristics were defined assuming none of the sites exhibited continuous snow cover for most of the winter (i.e., seasonal category 3 in the AERSURFACE User’s
- 4. 3 Guide was used for winter rather than category 4). For all scenarios, twelve 30-degree sectors, starting at 0 degrees (North) were used to define surface roughness. All locations were assumed to be non-arid. AERMET was executed to generate AERMOD-ready meteorological datasets using annual, seasonal, and monthly varying land use parameters assuming both average moisture conditions and moisture conditions varying depending on the annual, seasonal, or monthly precipitation for each respective temporal period of 2012 versus the climate normal precipitation. Only one moisture condition was assigned for the entire year in the meteorological data files generated for the annual temporal period. This moisture condition may have been average for 2012 based on actual precipitation data; therefore, to avoid a situation in which two runs are executed using identical land use parameters, meteorological data files using both the dry and wet moisture characteristics were processed in AERMET to compare AERMOD-predicted concentrations using each moisture condition against the average moisture condition for the annual temporal period. AERMOD Setup The most recent version of AERMOD (Version 13350) was run for a set of hypothetical emissions sources using each of the meteorological data sets derived for this analysis. Three co- located emission sources were defined in AERMOD: a low level point source with 2 meter release height, a mid-level point source with 10 meter release height, and an elevated point source with 45 meter release height above ground level. A unit emission rate of 1 g/s was assumed for each emission source and ambient temperature and negligible exhaust flow rate was assumed for each point source. Receptors used in this analysis include a square boundary grid, located at least 500 meters from the emission sources, with 100 meter grid spacing, a 100 meter spaced Cartesian grid extending 1.5 kilometers from the emission sources, and a 500 meter spaced Cartesian grid extending 3.5 kilometers from the emission sources. Flat terrain was assumed in all analyses. RESULTS In the following sections, differences between AERMOD-predicted concentrations are evaluated for each of three averaging periods (1-hour, 24-hour, and annual) and each stack height (2 meters, 10 meters, and 45 meters) comparing meteorological data sets generated using 1) average moisture versus actual moisture for annual, seasonal, and monthly temporal periods and 2) annual temporal period versus seasonal and monthly temporal periods all using average moisture conditions. Possible explanations for differences seen between AERMOD-predicted concentrations for different meteorological data sets are then provided. Moisture Condition Comparison The percentage differences of first high AERMOD-predicted concentrations for model runs executed using meteorological data sets of varying moisture condition are shown in Table 1 for 1-hour maximum, Table 2 for 24-hour maximum, and Table 3 for annual average concentrations. For the seasonal and monthly varying land use parameter cases, the comparison is made between
- 5. 4 AERMOD-predicted concentrations based on runs using meteorological data generated using “average” moisture condition for all seasons or months and meteorological data generated using moisture conditions varying based on the percentile of 2012 seasonal or monthly precipitation versus climate normal. The values shown are the percentage difference between the AERMOD- predicted concentration for the moisture condition case of interest and the average moisture condition case normalized to the average moisture condition case. For example, the percentage difference for the Annual Wet versus Annual Average cases is calculated as: % Difference = (Annual Wet Conc. – Annual Average Conc.) / Annual Average Conc x 100 Positive values indicate that the maximum concentration for the moisture condition case of interest is greater than for the average moisture condition and negative values indicate that maximum concentration for the moisture condition case of interest is less than for the average moisture condition case. For the annual land use parameter cases, AERMOD-predicted concentrations using average moisture conditions are compared with concentrations predicted using both wet and dry moisture conditions instead of using only a single moisture condition based on 2012 precipitation versus climate normal conditions. This additional case is included to avoid comparing AERMOD concentrations generated using identical meteorological data sets if 2012 happened to be an average year for precipitation for the station of interest. Table 1. 1-Hour Maximum Concentration - Moisture Condition Percent Difference Stack Height Met Station Annual Wet vs Annual Average Annual Dry vs. Annual Average Seasonal Actual vs. Seasonal Average Monthly Actual vs. Monthly Average ALB 0.0 0.0 0.0 0.0 BNA 0.0 0.0 0.0 0.0 PIH 0.0 0.0 0.0 0.0 SEA 0.0 0.0 0.0 0.0 ALB 0.0 0.0 0.0 0.0 BNA 0.0 0.0 0.0 0.0 PIH 0.0 0.0 0.0 0.0 SEA 0.0 0.0 0.0 0.0 ALB 4.4 -6.8 0.0 0.0 BNA 4.7 -2.0 -4.8 -0.1 PIH 4.5 -7.9 -0.7 0.6 SEA 4.6 -6.2 4.5 -4.6 45 meter 2 meter 10 meter
- 6. 5 Table 2. 24-Hour Maximum Concentration - Moisture Condition Percent Difference Table 3. Annual Average Concentration - Moisture Condition Percent Difference Maximum predicted 1-hour concentrations for the 2 meter and 10 meter stack height scenarios are identical for all moisture conditions at all meteorological observation stations, as shown in Table 1. The maximum predicted 1-hour concentrations for each of the meteorological stations occur during nighttime hours for the two lower level stack height scenarios. Turbulence during these hours is generated only through mechanical effects, which are only dependent on the surface roughness out of the three parameters (albedo, Bowen ratio, and surface roughness) output by AERSURFACE. Therefore, this observed trend is reasonable and expected. Differences in AERMOD-predicted concentrations are observed for the 24-hour and annual averaging periods for the lower stack height scenarios, but these differences are small with the greatest difference seen for 24-hour average concentrations being 0.3% and the greatest Stack Height Met Station Annual Wet vs Annual Average Annual Dry vs. Annual Average Seasonal Actual vs. Seasonal Average Monthly Actual vs. Monthly Average ALB 0.1 -0.1 0.1 -0.1 BNA 0.0 0.0 0.0 0.0 PIH 0.0 0.0 0.0 0.0 SEA 0.0 0.0 0.0 0.0 ALB 0.1 -0.3 0.0 0.2 BNA 0.0 0.0 0.0 0.0 PIH 0.2 -0.2 -0.2 0.2 SEA 0.1 -0.2 -0.2 0.1 ALB 10.2 -3.1 6.8 0.0 BNA 2.9 -4.5 0.0 3.2 PIH 0.7 -1.3 0.9 0.0 SEA 6.3 -8.5 5.1 -7.7 2 meter 10 meter 45 meter Stack Height Met Station Annual Wet vs Annual Average Annual Dry vs. Annual Average Seasonal Actual vs. Seasonal Average Monthly Actual vs. Monthly Average ALB 0.2 -0.2 0.0 0.0 BNA 0.1 -0.1 -0.1 0.0 PIH 0.1 -0.1 0.0 0.0 SEA 0.1 -0.2 0.0 0.0 ALB 0.7 -0.9 0.2 0.1 BNA 0.4 -0.4 -0.3 0.0 PIH 0.1 -0.1 0.0 0.0 SEA 0.5 -0.8 0.2 -0.5 ALB 4.2 -5.4 1.2 0.9 BNA 4.0 -3.5 -3.3 0.1 PIH 1.8 -3.5 -0.8 -0.9 SEA 2.4 -3.8 0.9 -3.0 2 meter 10 meter 45 meter
- 7. 6 difference seen for annual average concentrations being 0.9%. For the elevated stack (45 meter) release scenario, AERMOD-predicted concentrations are impacted by differences in moisture condition for each of the meteorological stations evaluated and for all averaging periods. A review of the time of day of the maximum 1-hour concentrations shows these events occurring during daytime hours, when buoyant turbulence conditions are dominant. AERMOD-predicted buoyant turbulence is a function of Bowen ratio; therefore, this observed trend is reasonable and expected. The magnitude of differences between AERMOD-predicted concentrations based on moisture condition varied from no difference up to just over a 10% difference in one scenario. The trend in model-predicted concentrations with varying surface moisture (wet, dry, and average) using annually varying surface parameter values is also of note. For all stations and stack heights, concentrations are either equal for all moisture conditions or show a trend for highest concentrations with the wet moisture condition and lowest concentrations for the dry moisture condition. Higher surface moisture (and therefore higher Bowen ratio) means that more solar energy will evaporate surface water rather than heating the surface and thus the lower atmosphere compared with average or dry moisture conditions. With higher surface moisture, the lower levels of the atmosphere will be heated less efficiently resulting in lower average convective mixing heights, less buoyant turbulence, and less plume dispersion compared with average or dry surface moisture conditions. Temporal Comparison The percentage differences of first high AERMOD-predicted concentrations for model runs executed using meteorological data sets of varying temporal (monthly, seasonal, and annual) land use parameters are shown in Table 4 for 1-hour maximum, Table 5 for 24-hour maximum, and Table 6 for annual average concentrations. The values shown are the percentage difference between the AERMOD-predicted concentration for the temporal case of interest and the annual temporal case normalized to the annual temporal case. Average moisture conditions are used to generate meteorological observation station-specific surface parameters for all cases. For example, the percentage difference for the Seasonal Average versus Annual Average cases is calculated as: % Difference = (Seasonal Average Conc. – Annual Average Conc.) / Annual Average Conc x 100 Positive values indicate that the maximum concentration for the temporal case of interest is greater than for the annual temporal case and negative values indicate that maximum concentration for the temporal period case of interest is less than for the annual temporal case.
- 8. 7 Table 4. 1-Hour Maximum - Temporal Percent Difference Table 5. 24-Hour Maximum - Temporal Percent Difference Stack Height Met Station Seasonal Average vs. Annual Average Monthly Average vs. Annual Average ALB 0.3 0.3 BNA -0.3 -0.3 PIH 166.4 166.4 SEA -0.3 -0.3 ALB 0.8 0.8 BNA 1.8 1.8 PIH 3.5 3.5 SEA -0.5 -0.5 ALB 2.2 2.2 BNA 0.2 0.2 PIH -16.0 -16.0 SEA -2.2 -2.2 2 meter 10 meter 45 meter Stack Height Met Station Seasonal Average vs. Annual Average Monthly Average vs. Annual Average ALB -0.9 -0.9 BNA -3.0 -3.0 PIH 34.8 34.8 SEA 2.2 2.2 ALB 3.0 3.0 BNA 1.6 1.6 PIH 1.5 1.5 SEA 0.0 0.0 ALB 0.3 0.3 BNA 4.1 4.1 PIH -1.9 -1.9 SEA -0.6 -0.6 2 meter 10 meter 45 meter
- 9. 8 Table 6. Annual Average - Temporal Percent Difference The percentage differences shown in Tables 4, 5, and 6 indicate that there is relatively good agreement between the AERMOD-predicted maximum concentrations when comparing the annual temporal period versus the seasonal or monthly temporal periods, with the greatest difference being 4.1%. The outlier in this analysis is the PIH surface observing station. A closer investigation of the hours at which maximum 1-hour average concentrations are predicted for the PIH meteorological data set shows a considerable difference between the surface roughness value present in the data sets processed using monthly and seasonal temporal resolution (0.058 meters) versus the data set processed using annual temporal resolution (0.143 meters). The sector upwind of the meteorological station for the hour with maximum predicted 1-hour concentrations at PIH is characterized by row crops using the AERSURFACE classification scheme on the NLCD92 file for the 1 km sector upwind of the PIH observation location. A review of the tables accompanying the AERSURFACE User’s Guide shows that this land use category has a greater seasonal variation in surface roughness than any other land use type, ranging from 0.02 meters in fall and winter (when fields are bare) to 0.2 meters in spring and summer (when field contain growing crops). The maximum 1-hour concentration observed for the PIH station occurs during nighttime conditions, during which only mechanical turbulence is assumed to occur in AERMOD. High AERMOD-predicted concentrations occur for the low level point source for the seasonal/monthly temporal case (with low surface roughness) compared with the annual temporal case (with higher surface roughness). With higher surface roughness, there is greater mechanical turbulence assumed for the hour in AERMOD. This greater mechanical turbulence acts to mix the low level release away from ground level. The converse is seen for the elevated point source case, where higher AERMOD-predicted concentrations are seen with the annual temporal case (with higher surface roughness) as the plume released from the stack is mixed more efficiently toward ground level with the greater mechanical turbulence. It is also notable that, when using a single moisture condition for all periods, the use of seasonal and monthly temporal periods produce identical model-predicted concentrations. The tables of Stack Height Met Station Seasonal Average vs. Annual Average Monthly Average vs. Annual Average ALB 0.6 0.6 BNA 0.1 0.1 PIH 1.0 1.0 SEA 0.7 0.7 ALB 0.1 0.1 BNA 0.7 0.7 PIH -0.6 -0.6 SEA 2.6 2.6 ALB -0.1 -0.1 BNA 0.2 0.2 PIH -1.3 -1.3 SEA -1.7 -1.7 2 meter 10 meter 45 meter
- 10. 9 albedo, Bowen ratio, and surface roughness used in AERSURFACE are based on seasonal averages. Thus, albedo and surface roughness will be the same for a particular hour in an AERMOD-ready meteorological data file generated using a monthly temporal period as for a file generated using a seasonal temporal period. If the moisture condition does not change, then Bowen ratios would also be the same using monthly and seasonal temporal periods. Thus, there is expected to be little difference in land use parameters in AERMOD-ready meteorological data sets processed using seasonal or monthly temporal frequency unless a particular month’s precipitation versus climate normal is not consistent with the seasonal value versus climate normal (e.g., if a wet month occurs during an average or dry season). CONCLUSIONS The AERSURFACE utility allows users to define land use parameters (albedo, Bowen ratio, and surface roughness) with differing temporal resolution (i.e., annual, seasonal, and monthly) and with Bowen ratios adjusted based on the annual, seasonal, or monthly actual precipitation versus climate normal. U.S. EPA provides guidance in the AERSURFACE User’s Guide concerning appropriate methodologies for defining whether a particular period is wet, dry, or average versus climate normal. However, guidance is not readily available concerning the choice of the temporal resolution of land use parameters that is appropriate for use in AERMET. The case study presented here highlights the impacts of varying the moisture condition and temporal definition of land use parameters through a hypothetical scenario with point source releases at varying heights above ground level and using meteorological data from locations across the United States. In the cases investigated, variations in the surface moisture condition used in AERSURFACE had less impact for low level releases. In these situations maximum concentrations often occur in stable conditions when mechanical turbulence, which is not a function of Bowen ratio, is dominant. Variations in surface moisture had greater impact for elevated releases. In these situations maximum concentrations more frequently occur in unstable conditions when buoyant turbulence, which is a function of Bowen ratio, is dominant. Variations in maximum AERMOD-predicted concentrations between meteorological data sets with differing temporal definitions of land use parameters were greatest in situations where surface roughness varied considerably as a function of season, such as land used for cultivation of row crops. Therefore, AERSURFACE users should exercise caution in selecting appropriate temporal resolution for land use parameters in situations where land use types with a wide seasonal variation in surface roughness are dominant. Additional guidance from U.S. EPA concerning the appropriate choice of the temporal definition of land use parameters (i.e., annual, seasonal, or monthly) when using AERSURFACE is warranted. This recommendation is due to the greater sensitivity of AERMOD-predicted concentrations based on variations in temporal period as compared with variations in moisture condition in AERMET, for which guidance is available.
- 11. 10 REFERENCES 1. AERSURFACE User’s Guide, U.S. EPA, Research Triangle Park, North Carolina, January 2008 (Revised 1/16/2013). 2. Hill, J, I. Donaldson, D. Harrison, “An Evaluation of AERMOD Model Sensitivity to Variations in Landuse Characteristics”, A&WMA Specialty Conference Guideline on Air Quality Models, October 2009. 3. AERMOD Implementation Guide. U.S. EPA, Research Triangle Park, North Carolina. Revised March 2009. KEYWORDS AERMOD, AERSURFACE, Bowen ratio