Regional Ammonia & Windblown Dust Models Status & Lessons Learned

1. Regional Ammonia & Windblown Dust ModelsStatus & Lessons Learned Gerard Mansell ENVIRON International WRAP Workshop on Regional Emissions & Air Quality Modeling Studies Denver, CO 29-30 July 2008

2. PRESENTATION OVERVIEW • Windblown Fugitive Dust • Fugitive Dust Sources • Regional WB Dust Modeling – WRAP WB Dust model • Methodology Development • Sensitivity Studies • Model Results • Specific Model Applications & Enhancements • Summary & Recommendations • Regional Ammonia Models • Inventory Development & AQ Modeling issues • WRAP NH3 Model • Model Development & Data Sources • Model Results • Other Regional NH3 Models • Carnegie Mellon University (CMU) NH3 Model • LADCo Process-Based NH3 Model • Summary & Recommendations

3. FUGITIVE DUST • What is fugitive dust? • Particulate matter (PM) emissions that can be suspended in air that are not ducted or capable of being ducted • Examples • Entrained paved and unpaved road dust • Construction and demolition activities • Mining and mineral processing activities • Agricultural operations • Windblown dust from disturbed surfaces • Others (unpaved shoulders, leaf blowers)

4. Fugitive Dust Source Emission Estimates • Traditional: AP-42 (USEPA) • Emissions = activity level x emission factor • Emissions are not spatially-resolved - need to use spatial surrogates, GIS overlays • CARB fugitive dust methodologies: http://www.arb.ca.gov/ei/areasrc/index7.htm • New: Emission Models • Emissions are calculated by an emissions model that attempts to capture the physical properties of the source • Example: windblown dust models • Better, spatially- and temporally-resolved emission estimates • Need significant input information • Technical expertise to exercise

5. Examples • Unpaved Roads • AP-42: emissions a function of VMT, silt content, speed, weight, number of wheels and mean precipitation • Controls: speed limits, water, chemical stabilizers, gravel, paving, access restrictions • Construction • CARB Emissions = activity (in acre-months) *EF (0.11 tons/acre-month); assumes watering • Spatial surrogates needed • South Coast using GIS-based approach • Controls: watering, phased grading, stabilization (unpaved roads, disturbed areas), track-out controls • Agriculture • Tilling (by crop): acreage, acre-passes • Harvest (by crop): acreage, crop-specific EF • Controls: Conservation Management Practices (CMPs) • Windblown Dust • Equation-based: WEQ, RWEQ, WEPS, etc. • Emissions models: ENVIRON/RMC, Draxler et al., Zender et al. (DEAD model), Shao • Controls: watering, stabilization, re-vegetation, access restrictions, CMPs

6. WRAP/RMC WINDBLOWN FUGITIVE DUST MODEL BACKGROUND • WRAP recognized the need for WB Dust inventory in regional haze modeling efforts – previously no dust from wind erosion available on regional-scale • Recognition of relative magnitude and importance with respect to other source categories in regional emission inventories • Consideration of EPA guidance & requirements • DEJF funded a number of studies to investigate and evaluate impacts of dust on visibility and air quality at Western Class I Areas • Causes of Haze (COHA) project • Causes of Dust (COD) • Analysis of Fine Fraction of PM in Fugitive Dust • Phase I & II WB Dust studies funded to develop methodology and modeling system to estimate WB Dust emissions applicable for Regional Haze Modeling

7. METHODOLOGY DEVELOPMENT • Phase I • Developed a general methodology and modeling approach/platform • Identified and evaluated data requirements and sources • Surface characteristics – LULC; Soil Texture • Meteorology – MM5; CALMET; other • Agricultural data – Crop types, calendars, CMPs (i.e., tilling, harvesting, irrigation, etc.) • Emission Factors and threshold velocity relationships • Utilized a simplified estimation methodology • Phase II • Reviewed and refined overall objectives • Conducted literature reviews • Global models – Zender, 2003; Draxler et al., 2001; Shao, 2001; Marticorena et al., 1997; Alfaro et al., 2003 • Field studies – Gillette et al, 1988, 1982; Gillette, 1988; Nickling & Gillies, 1989 • Other wind erosion models - Mac Dougall method; Alfaro, et al, 2003 • Identified and evaluated current updated databases – LULC • Developed and implemented revised emission estimation methodology • Sensitivity studies • Dust reservoir treatment and assumptions • Soil disturbance assumptions • Other Model Parameters • LULC (1992 NLCD vs. 2000 NALC) • Transport Fractions • Fine/Coarse PM ratios

8. METHODOLOGY DEVELOPMENT Phase II Estimation Approach Threshold Friction Velocities • Dust = f(LULC,z0,u*,u*th,SC) • u* = f(u,z0) • u*th = f(z0) • z0 = f(LULC) Emission Rates

9. SENSITIVITY STUDIES Sensitivity simulations performed to evaluate effects of soil disturbance and dust reservoir assumptions Assumptions associated with scenario (b) were used in all subsequent model applications for the WRAP – loose undisturbed soils; wind event duration limited to 10 hrs/day

10. OTHER MODELING INVENTORY PARAMETERS • Land Use/Land Cover (LULC) • 1992 National Land Cover Database (1992 NLCD) • 2001 NLCD is now currently available – high spatial resolution, more detailed classifications • 2000 North America Land Cover database (2000 NALC)

11. OTHER MODELING INVENTORY PARAMETERS • LULC Summary & Comparison LULC critical input as it determines surface roughness  threshold velocity  wind erosion potential

12. OTHER MODELING INVENTORY PARAMETERS • Fugitive Dust Transport Fractions • Transport fractions used to model affects of near-source removal of dust emissions through gravitational settling and/or deposition to surfaces or captured in the surround canopy or on physical structures • Reduces the amount of dust emitted to the atmosphere for regional air quality modeling. • Applied to gridded modeling inventory • Transport fractions updated in all fugitive dust emission inventories

13. MODEL RESULTS

14. How Important is Fugitive Dust in Western States? • Of the 68 State Implementation Plans (SIPs) and Natural Event Action Plans (NEAPs) from Western States: • 30 cite dust as the primary pollutant to control • 15 site dust as the second highest pollutant • Other PM10 contributors include on- and off-road mobile emissions (e.g. diesel PM) and pre-cursors such as NOx and SOx from combustion sources and ammonia

15. OTHER APPLICATIONS AND RELATED STUDIES • State/local NEI & AQ modeling inventory development (HI, AR, AZ, WY, NV, FCAQTF, Columbia Gorge AQ Study, many others) • NEAP applications (Salt Lake City, UT area) • PM Nonattainment Area Maintenance Plans (Phoenix/Maricopa Co.; Imperial Valley, CA) • WRAP Dust Definition & New Mexico Pilot Dust Regional Haze State Implementation Plan for the Salt Creek Wilderness Area

16. APPLICATIONS FOR PM10 MAINTAINENCE PLANS • Phoenix/Maricopa Co., AZ • Used Phase II Methodology • LULC based on more detailed local data – MAG LULC data • STATSGO Soil texture data updated/augmented with SSURGO database • Revised agricultural adjustment info based on input from MAG staff • Consideration given to temporal variation of disturbance levels based on crop calendars • Model revised to consider LULC-specific Z0 values & disturbance levels for all LULC categories • Sensitivity simulations implemented for variations in soil disturbance for various LULC categories. • Significant improvements over “standard” model application realized • Imperial Valley, CA • Used Phase I Methodology • LULC based on more detailed data – CA DWR merged with NLCD • Revised ag adjustment info based on DWR data – multi-cropping, inter-cropping, etc.. • Sensitivity simulations implemented varying threshold friction velocities. • Specific adjustments to LULC data based on local input • LULC-based Z0 values & soil disturbance adjusted for Imperial Sand Dunes and surrounding areas

18. APPLICATIONS FOR WRAP DUST DEFINITION & NM PILOT DUST REGIONAL HAZE SIP • Study Objectives • Collect & summarize fugitive dust PM10 emissions to support Salt Creek Wilderness Pilot PM10 SIP project • Provide fugitive dust PM10 emissions for use in Dust Definition Case Study • Develop inventory refinements based on WRAP modeling data, revised spatial allocation, local data, back-trajectory residence time analyses, etc. • Inventory Refinements and Evaluation • Focus on emission sources within 100-km buffer zone around Salt Creek • Revised allocation of county-level dust emissions to 100-km buffer zone using 2000 NALC data • Extract and evaluate modeling (gridded) data within buffer zone • Estimate wind blown dust emissions from unpaved road using data from NMDOT and WRAP WB Dust model • Estimate wind blown dust emissions from O&G well sites using WRAP WB Dust model • Assess dust emission impacts at Salt Creek using back-trajectory analyses

19. APPLICATIONS FOR WRAP DUST DEFINITION & NM PILOT DUST REGIONAL HAZE SIP • Most significant source: Windblown from shrub/grass lands, much of which has been or can be grazed by cattle (mixed source) • Major sources: Other windblown dust (mixed sources); agriculture, construction, road dust (anthropogenic) • Conceptual Model for Salt Creek identified an uninventoried major source: oil and gas production areas / unpaved roads Satellite photograph of oil and gas production facilities. Lightly shaded areas are well-pads, which are connected by access roads

20. APPLICATIONS FOR WRAP DUST DEFINITION & NM PILOT DUST REGIONAL HAZE SIP PM10 Dust Emission Inventory Refinements • Unpaved road dust • WRAP WB Dust model applied to estimate additional dust from unpaved roads • Landuse input data developed using road mileage and assumed road width • Assumed barren land; distributed uniformly across grid cells • Sensitivity simulations used to evaluate assumed soil disturbance levels • Oil & Gas Well Sites • WRAP WB Dust model applied to estimate additional dust from unpaved roads • Landuse input data developed using well site location, status and acreage • Assumed barren land; allocated to grid cells using site location • Sensitivity simulations used to evaluate assumed soil disturbance levels • Undisturbed; 100% disturbed; 50% disturbed

21. APPLICATIONS FOR WRAP DUST DEFINITION & NM PILOT DUST REGIONAL HAZE SIP Unpaved roads based on NMDOT unpaved road miles by County. Assumed 27 ft width. Area distributed uniformly across counties. Oil & Gas well sites based on acres per site. Location based on geographic coordinates

22. RECOMMENDATIONS & FUTURE WORK • Numerous assumptions are implemented in the model – need to be reviewed on a case-by-case basis for applicability • Results are highly dependent of accurate and detailed databases • Model limitations are related to regional-scale nature of input data – can be resolved through implementation at small local scale with detailed surface characteristics databases • Additional further research is recommended • Continued investigation and evaluation of surface characterizations – LULC; Surface roughness parameters; soil disturbance • Identification of local detailed databases • Local scale applications and evaluations • Further investigation into various model assumptions – reservoir characteristics, soils, disturbance levels • Refinement of agricultural data and adjustments – temporal & spatial variations

23. AMMONIA EMISSIONS & AQ MODELING NH3 Emission Estimation & Modeling Issues • Emission Rates (Emission Factors) • AP-42: E = A*EF • Process-based: N mass balance • Spatial Resolution • Locational accuracy of emission sources • Mobile  roadways • Agricultural/Livestock  ag land surrogates • Domestic  population distribution • Feedlots • Impacts on AQ Modeling • Point source vs. distributed area source • AQ Model grid resolution • Temporal Patterns • Temporal profiles • Meteorology • Activity data

24. REGIONAL AMMONIA MODELS WRAP NH3 Model Overview • Developed as GIS-based modeling system • Incorporates environmental parameters – soil pH, met data (winds, temperatures) • Source categories include: • Livestock • Fertilizers • Native Soils • Domestic Sources • Wild Animals • Based on 2002 activity data (no future year projections done to date) • Monthly activity data for fertilizers; annual for livestock, domestic, wild animals • Activity data for soil emissions based on LULC (2000 NALC) • Hourly emission estimates based on temporal variations of met data & temporal profiles • No Mexico or Canada

25. MODEL DEVELOPMENT & DATA SOURCES • Emission estimates calculated as: • Enh3= EF * Activity • EF based on “whole animal” approach • Activity data • annual; monthly • Emission factors adjusted for environmental effects (soil pH; meteorology) • Fertilizer application • Natural soils • Temporal variation based on environmental parameters & temporal profiles • Livestock • Fertilizer application • Natural soils • Gridded emissions spatially allocated based on LULC • 2000 NALC data; 2000 Census populations • Model developed in Arc/INFO GIS

26. MODEL DEVELOPMENT & DATA SOURCES Activity Data • Livestock Headcounts • National Agricultural Services Statistics (NASS, 2003) • Fertilizer Usage • Assoc. of American Plant Food Control Officers (AAPFCO, 2003) • USDA • NASS • Natural Soils • Land cover acreage (2000 NALC; 1992 NLCD) • Domestic • 2000 US Census • Pet per capita ratios from Dickson et al., 1991 • Wild Animals • CMU NH3 model

27. MODEL DEVELOPMENT & DATA SOURCES Livestock Emission factors (kg/head/yr) Fertilizer Emission factors (%N)

28. MODEL DEVELOPMENT & DATA SOURCES Soil Emission Factors (kg/km2-yr) Domestic Emission Factors

29. Emission Factors Natural soil emission factors adjusted to account for soil conditions (Potter, et al., 2003) EFadj = EF * (1-M){1/[1+10(0.09+2730/T-c*pH)]} M = soil moisture T = soil temp (K) C = constant determines sensitivity to pH (=1.3) Fertilizer emission factors adjusted for soil pH (Potter, et al. 2001) EFadj = EF * (0.3125pH – 1.01) Normalized for 4% at pH=6.5 Temporal Allocation Livestock Monthly profiles recommended by Chinkin et al. (2003) Diurnal profiles based on Russell & Cass (1986) Ei~ [2.36(Ti-273)/10] ViA Fertilizers Monthly variation from fertilizer usage data Diurnal profiles based on Russell & Cass (1986) Native Soils Temporal variation based on environmental parameters EFadj = EF * (1-M){1/[1+10(0.09+2730/T-c*pH)]} Domestic & Wild Animal Sources Temporally invariant MODEL DEVELOPMENT & DATA SOURCES

30. Emission Factor Adjustments & Impacts

31. MODEL DEVELOPMENT & DATA SOURCES Temporal Allocation • Livestock Diurnal Profiles

32. MODEL RESULTS Gridded NH3 Emissions Total Livestock – Annual 2002

33. MODEL RESULTS • Comparison with CMU Model

34. OTHER REGIONAL NH3 MODELS CMU NH3 Model -- http://www.envinst.cmu.edu/nh3/ • Windows-based emission model • Uses AP-42 type estimation approach • Incorporates GIS data for LULC (USGS data) • Spatial Resolution • National Level • State Level • County Level (ArcINFO/ArcView compatible) • Sub-county (200 m resolution) based on land use GIS data (ArcINFO/ArcView compatible) • Temporal Resolution • Annual • Monthly (for fertilizer application)

35. OTHER REGIONAL NH3 MODELS LADCo Process-based NH3 Model http://www.ladco.org/reports/rpo/MWRPOprojects/Emissions/Technical_Paper1.pdf • Consider and analyze all physical, chemical and biochemical processes and reactions that take place and influence ammonia emission rate, • Employ processed based mechanistic and empirical models (new and existing) • Keep mass balances for the flow of nitrogen through each component of an animal waste management system.

36. LADCo Process-based NH3 Model • NH3 Animal Allocation Processor • NH3 Farm Emissions Model: • Animal excretion model • Housing emissions model • Feedlot emissions model • Storage emissions model, and • Land emissions model • Animal species considered: • Dairy cows • Beef cattle • Swine • Poultry (layers, broilers, and turkeys) • Commercial Fertilizers

37. LADCo Process-based NH3 Model Animal Allocation Processor (AAP) • Distribute county-level animal head counts to defined Manure Management Trains (MMTs) • Spatially allocate MMTs to grid cells using gridded surrogates (agricultural land) • Format input data for Farm Emission Model (FEM) • Actual Farm Data • Commercial Fertilizers Farm Emission Model (FEM) • FEM computes NH3 emissions with animal numbers by each Manure Management Train (MMT) for each grid cell: • Reads in ASCII outputs from AAP • Reads in meteorology file from CONCEPT met tables • Run Animal Excretion Model • Based upon MMTID: • Run Housing Emissions Model • Run Storage/Feedlot Emissions Model • Run Land Emissions Model • Output total NH3 based on animal type & MMTID • Output commercial fertilizer NH3 emissions estimates

38. AAP Input Data Sources Livestock Data 2002 and 1997 Census of Agriculture Data EPA MMT Distributions Revised MMT by ISU for Midwest states FEM Defaults from UCD and ISU EPA Animal Population Category  FEM Categories from UCD and ISU Commercial Fertilizers Carnegie Mellon University (CMU) NH3 Model County-level fertilizer application rates by month for 2002 FEM Input/Output Input Data AAP ASCII outputs Meteorological Data Lat/Lon Coordinates Wind velocity and direction Relative Humidity, Rain, Frictional velocity, etc. Output Data Based upon animal type & MMTID Format: CONCEPT ready format NIF 3.0 format ASCII csv format LADCo Process-based NH3 Model

39. Animal Housing and Management Practices Example Manure Management Train (MMT)

40. LADCo Process-based NH3 Model Data Requirements • Animal excretion sub-module: Nitrogen excretion from animals is influenced by the age, species, and diet of animals. • Required data include animal populations, age, and feed ratios. • Animal housing sub-module: Emissions depend on the specific housing design and practices. • Collection and storage of manure varies by indoor and outdoor storage. • Required data include specific housing design and operation, animal populations by age and species and climatic conditions. • Storage sub-module: Emissions associated with manure storage varies by type of manure • Dry manure storage is typical of beef cattle feedlot, dairy corrals, high-rise layer facility, broiler and turkey facilities. • Wet manure storage is commonly used in swine and layer facilities. • Emissions are based on the type of storage facilities and manure. • Required data include type and number of storage facilities and environmental data. • Land application module: Emission rates from different land application vary by type of manure, crop management practices and climatic conditions. • Required data include nutrient content of manure by animal type, specific application and crop management practices and environmental conditions. Required data for each sub-module are not typically available with the type of detail required for a region encompassing the entire U.S. • Default values for each sub-module, based on information in the Midwest are used. • Data can vary considerably across the US, a set of ranges for these parameters are provided.

41. LADCo Process-based NH3 Model - Results

42. LADCo Process-based NH3 Model - Results

43. SUMMARY & RECOMMENDATIONS • …