NH3 Emissions for Regional Air Quality Modeling
Motivation and Questions Motivation NH3 plays a key role in secondary aerosol formation impact on health & regional haze NH3 deposits readily and contributes to N deposition in pristine ecosystems Questions What are the major sources of NH3 within the region? For the major sources, what are the uncertainties in the emission factors? For the major sources, how well do we know the activity levels? For the major sources, how do environmental or operating practices affect emission rates? NH3 emissions are uncertain, but how sensitive are model results to these uncertainties? Is near-source NH3 deposition an issue within a grid modeling framework?
Atmosphere/biosphere Nitrogen Cycle Anthropogenic Tg NH 3 / year Domesticated animals 21 29 Fertilizer 6-9 Biomass burning 2 6 Fossil fuel 0.3 2 Miscellaneous 0-7 Total Anthropogenic 30 50 Natural 15 23 Total 45-75
NEI NH3 Emissions for Washington Source Type NH3 (tons/yr) Ag Fertilizer Application 12,659 Cattle & calves operations 12,462 Dairy operations 5,718 Other animals 4,957 Biogenic sources 4,082 Waste treatment 2,602 Industrial Processes 1,392 Stationary Fuel Combustion 306 Waste burning 111 Solvent Use 0 WA Total 44,289
Fertilizer NH3 emissions factors (Chinkin et al. 2003) A. NH3---Battye-12(1%)--CMU-1%--Corsi-49(12-121)-EEA-4%
Beef and dairy cattle NH3 emission factors (lb/head/year; Chinkin et al. 2003) Dairy 48 to 74 lb/head/yr Beef 29 to 88 lb/head/yr
Recent Feedlot Emission Results (Flesch et al., 2007) Spring and summer 0.15 kg/head/day Extrapolated annual emission factor: 55 kg/head/yr or 120 lb/head/yr
Recent EI adjustments: Columbia River Gorge Visibility Modeling Analysis (Environ, 2007) Comparison of 2002 NEI and Environ NH3 GIS EI (1) Ammonia emissions from confined area feeding operations (CAFO), such as dairies, were understated by factors of 1.5 to approximately 3, depending on the type of manure handling conducted at each (i.e., flush, scrape, drylot/pasture, or deep-pit); (2) Ammonia emissions from fertilizer application were understated by upwards of a factor of three for anhydrous and aqueous ammonia application sources, and by a factor of 2.5 for nitrogen solution fertilizer application sources.
Tomorrow s Air Quality: AIRPACT-3 Daily Forecast System Airpact Terrain ht. MM5 numerical mesoscale meteorological model (UW) SMOKE: Sparse Matrix Operating Kernal for Emissions processing. CMAQ: Community Multi-scale Air Quality model: O 3 & toxics chemistry with 72 species and 214 reactions Aerosol dynamics & chemistry in Aitken, accumulation, & coarse modes Deposition of N, S, O3, & Hg species 12 km x 12 km grid cells, 21 layers Forecast to 64 hours daily Runs in ~1 hour on 4 nodes of a linux cluster PM species Nitrates Sulfates Organic aerosols Wind-Blown Dust (soon) PM2.5 total mass
AIRPACT-3 Dynamic Emissions & IC/BC 2005 anthropogenic emissions (SMOKE) MM5/MCIP Meteorology from UW Weather Forecast System Updated Biogenic Emission Model (BEIS3) WSU Dairy NH3 Emissions Module Wild and Prescribed Fire Emissions Gridded Emissions Dynamic Boundary Conditions: spatial & temporal details CMAQ Daisy-chain Initial Conditions
NH3 Emissions Dairy Module (Rumburg et al., 2006) based upon DOAS NH3 & tracer ratio measurements at the WSU dairy Housing, lagoon, and slurry application emissions Incorporates N balance approach and accounts for wind speed & temperature effects
Stalls Model Sensitivity Model Parameter Avg. Flux Avg. Daily Peak Daily Annual (mg cow -1 s -1 ) (kg day -1 ) (kg day -1 ) (kg year -1 ) Base Case 5.4 20 86 7,000 ph = 8.1 10 40 170 14,000 ph = 7.5 2.7 10 47 3,700 Temperature + 2 C 6.1 24 100 8,500 Temperature - 2 C 4.6 18 80 6,400 Milk Prod. = 50 kg day Milk Prod. = 35 kg day 6.5 22 97 7,900 4.6 20 86 7,000 Protein = 21% 6.0 23 100 8,300 Protein = 17% 4.7 18 80 6,500
Dairy Module Input Data Requirements Column Description Type A FIPS 5-digit Integer B Site-ID 5-digit Integer C Latitude Real D Longitude Real E Number of Milking cow Integer F Number of Dry cow Integer G Number of Heifer Integer H Number of Calves Integer I Lagoon area (m^2) Real J Pit area (m^2) Real K Tank area (m^2) Real L Housing type (Scrape Barn = 1) Integer M Application type (Big Gun = 1) Integer N Application area Real O Site name Text in quotes
Dairy NH3 Emission Module WA Dairy NH3 (ton/yr) Housing 8,634 Storage 5,441 Application 4,364 Total 18,439 OR Dairy NH3 (ton/yr) Housing 9,469 Storage 5,079 Application 3,420 Total 17,967
Source Type NH3 (tons/yr) Ag Fertilizer Application 12,659 Cattle & calves operations 12,462 Dairy operations 5,718 Other animals 4,957 Airpact NH3 Emissions for Washington Biogenic sources 4,082 Waste treatment 2,602 Industrial Processes 1,392 Stationary Fuel Combustion 306 Waste burning 111 Solvent Use 0 WA Total 44,289 or with WSU dairy module Dairy operations 18,439 WA Total 57,010 % increase 29%
50 POC PEC Airpact-3 Evaluation, Aug-Nov 2004 (Chen, 2007) Modeled (µg/m 3 ) 40 30 20 10 0 0 10 20 30 EPA_AQS IMPROVE SWCAA 40 50 Modeled (µg/m 3 ) 8 6 4 2 0 0 2 4 6 EPA_AQS IMPROVE SWCAA 8 Measured (µg/m 3 ) Measured (µg/m 3 ) Modeled (µg/m 3 ) 20 15 10 5 EPA_AQS SWCAA PSO4 Modeled (µg/m 3 ) 5 4 3 2 1 EPA_AQS SWCAA PNH4 Modeled (µg/m 3 ) 10 8 6 4 2 EPA_AQS SWCAA PNO3 0 0 5 10 15 20 0 0 1 2 3 4 5 0 0 2 4 6 8 10 Measured (µg/m 3 ) Measured (µg/m 3 ) Measured (µg/m 3 )
Columbia Gorge Visibility Modeling Analyses (Environ, 2007) Source attribution results for PNH4 Nov. 2004
Questions remain What are the major sources of NH3 within the region? Fertilizer vs livestock which is larger? Biogenic natural soil emissions very few measurements. For the major sources, what are the uncertainties in the emission factors? Factor of two or more in available factors For the major sources, how well do we know the activity levels? For the major sources, how do environmental or operating practices affect emission rates? Seasonal changes, diurnal changes NH3 emissions are uncertain, but how sensitive are model results to these uncertainties--how accurate does an emission inventory need to be? Is near-source NH3 deposition an issue within a grid modeling framework? What is required to improve our ability to construct more accurate emission inventories for NH3?
Animal Ammonia Emissions Pathways for NH 3 emissions from animal waste Hydrolysis of urea in urine with urease Anaerobic biological breakdown organic nitrogen amino acids NH 3 Dairy cows have the largest per animal emissions (Bouwmann et al., 1997) Developed countries ~ 20 kg NH 3 cow -1 yr -1 Developing countries ~ 8 kg NH 3 cow -1 yr -1 European measurements and farming practices Filters, flasks, and passive samplers
Housing Emissions Model Calculates liquid NH3 concentration based upon cow factors Uses milk production, dietary Na, K, and N intake, body weight, and week of lactation -> urinary urea Assumes excess of urease from manure Converts urea to ammonia-n Partitions ammonia-n between NH 3 and NH 4 + based upon ph, (ph = 7.8 - lagoon ph) Calculates volatilization based upon temperature
Lagoon Emissions Model Calculates lagoon ammonia-n concentration based upon temperature Partitions ammonia-n between NH 3 and NH 4+ based upon ph, ph = 7.8 Calculates volatilization based upon temperature Uses Gaussian plume model to calculate downwind concentrations
Lagoon Modeling 600 500 Normalized mean error = 21% NH 3 Modeled Concentration (ppbv) 400 300 200 100 0 0 100 200 300 400 500 600 NH 3 Measured Concentration (ppbv)
Stalls Modeled versus Measured Concentrations 2500 Normalized Mean Error = 30% 2000 NH 3 Modeled Concentration (ppbv) 1500 1000 500 0 0 500 1000 1500 2000 2500 NH 3 Measured Concentration (ppbv)
Slurry Application Modeling 600 500 Theoretical Model Error = 59% Empirical Model Error = 22% NH 3 Concentration [ppbv] 400 300 200 Measured Concentration Theoretical Model Empirical Model 100 0 10 20 30 40 Decay Time (Hours)
Lagoon Emissions Model Stalls Convective Mass Transfer Coefficient hm Free Air Stream NH3(g) NH3 NH3 + NH4 + NH 4 + ph NH 3 Temperature Bacteria Bacteria Organic N Organic N
Waste Application Model Theoretical Model Slurry volatilization from upper layer of soil Slurry infiltration - HYDRUS 1D Slurry adsorption to soil Slurry temperature - radiation & canopy light model Empirical Model Exponential fit to the data Useful for air quality models
Slurry Application Convective Mass Transfer Coefficient hm Free Air Stream Adsorption NH3(g) NH3 Soil (-) particles NH 4 + ph NH 3 Temperature Infiltration
Tomorrow s Air Quality: AIRPACT-3 Daily Forecast System Airpact Terrain ht. MM5 numerical mesoscale meteorological model (UW) SMOKE: Sparse Matrix Operating Kernal for Emissions processing. CMAQ: Community Multi-scale Air Quality model: O 3 & toxics chemistry with 72 species and 214 reactions Aerosol dynamics & chemistry in Aitken, accumulation, & coarse modes Deposition of N, S, O3, & Hg species 12 km x 12 km grid cells, 21 layers Forecast to 64 hours daily Runs in ~1 hour on 4 nodes of a linux cluster PM species Nitrates Sulfates Organic aerosols Wind-Blown Dust (soon) PM2.5 total mass
Airpact-3 Aerosol Results / http://www.lar.wsu.edu/airpact-3
Modeling sensitivity to NH3 emissions: Columbia Gorge Modeling Analysis (Environ, 2007) Sensitivity Runs 1. Zero PGE Boardman from BART-level emissions 2. Zero ammonia emissions East of Gorge 3. Zero on-road mobile emissions in Portland 4. Zero major industrial (point) emissions Portland 5. Zero major point emissions in the Gorge