Atmospheric Environment 39 (2005) 7137–7153 Measurements of ammonia emissions from oak and pine forests and development of a non-industrial ammonia emissions inventory in texas Golam Sarwar a,1 , Richard L. Corsi a, , Kerry A. Kinney a , Joel A. Banks a , Vince M. Torres a , Chuck Schmidt b a Center for Energy and Environmental Resources (R7100), The University of Texas at Austin, 10100 Burnet Road, Austin, TX 78758, USA b Independent Environmental Consultant, 19200 Live Oak Road, Red Bluff, CA 96080, USA Received 23 March 2005; accepted 14 August 2005 Abstract Estimates of non-industrial source ammonia emissions in Texas were developed through the use of published emission factors and activity data for those sources. A total of 64 non-industrial source emission sub-categories were addressed, each falling into one of seven major source categories: animal husbandry, fertilizer applications, on-road vehicles, non- road sources, municipal wastewater disposal, domestic sources, and natural soil and vegetation. Annual statewide ammonia emissions were initially estimated to be 921,000 metric tons, with greater than 50% originating from natural soil and vegetation. However, estimates for pine and oak forests were characterized as having a great deal of uncertainty. A series of field sampling events were conducted to determine ammonia fluxes from pine and oak forest floors in east Texas. Both dynamic and static chamber methods were used. The ammonia flux averaged 0.09 kg km 2 month 1 for pine forests and 0.13 kg km 2 month 1 for oak forests. These values are significantly lower than those previously measured and cited in the published literature. However, the ammonia fluxes measured in east Texas forests are reasonably consistent with those predicted using mechanistic models for evergreen pine and deciduous broadleaf forests in Alabama, California, Colorado, and Tennessee. Statewide annual ammonia emissions estimates, revised using the newly developed ammonia fluxes for oak and pine forests in Texas, dropped from 921,000 to 467,000 metric tons. The relative contribution of ammonia emissions from pine and oak forests dropped from 49% to less than 1%. Animal husbandry was predicted to be the dominant non- industrial source, accounting for approximately 77% of non-industrial source ammonia emissions. r 2005 Elsevier Ltd. All rights reserved. Keywords: Ammonia emissions; Soil; Forests; Inventory 1. Introduction Atmospheric chemical reactions are believed to be a major source of fine particulate matter (PM 2.5 ). An important contributor toward many of these reactions is ammonia (NH 3 ), which is emitted from a wide range of anthropogenic and natural sources. ARTICLE IN PRESS www.elsevier.com/locate/atmosenv 1352-2310/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.atmosenv.2005.08.016 Corresponding author. Tel.: +1 512 475 8617; fax: +1 512 471 1720. E-mail address: [email protected] (R.L. Corsi). 1 Current address: USEPA, Mail Drop E243-03, 109 T.W. Alexander Drive, RTP, NC 27711, USA.
Transcript
ARTICLE IN PRESS
1352-2310/$ - se
doi:10.1016/j.at
�Correspond
fax: +1512 471
E-mail addr1Current add
Alexander Driv
Atmospheric Environment 39 (2005) 7137–7153
www.elsevier.com/locate/atmosenv
Measurements of ammonia emissions from oak and pine forestsand development of a non-industrial ammonia emissions
inventory in texas
Golam Sarwara,1, Richard L. Corsia,�, Kerry A. Kinneya, Joel A. Banksa,Vince M. Torresa, Chuck Schmidtb
aCenter for Energy and Environmental Resources (R7100), The University of Texas at Austin, 10100 Burnet Road, Austin, TX 78758, USAbIndependent Environmental Consultant, 19200 Live Oak Road, Red Bluff, CA 96080, USA
Received 23 March 2005; accepted 14 August 2005
Abstract
Estimates of non-industrial source ammonia emissions in Texas were developed through the use of published emission
factors and activity data for those sources. A total of 64 non-industrial source emission sub-categories were addressed,
each falling into one of seven major source categories: animal husbandry, fertilizer applications, on-road vehicles, non-
road sources, municipal wastewater disposal, domestic sources, and natural soil and vegetation. Annual statewide
ammonia emissions were initially estimated to be 921,000 metric tons, with greater than 50% originating from natural soil
and vegetation. However, estimates for pine and oak forests were characterized as having a great deal of uncertainty. A
series of field sampling events were conducted to determine ammonia fluxes from pine and oak forest floors in east Texas.
Both dynamic and static chamber methods were used. The ammonia flux averaged 0.09 kg km�2 month�1 for pine forests
and 0.13 kg km�2 month�1 for oak forests. These values are significantly lower than those previously measured and cited in
the published literature. However, the ammonia fluxes measured in east Texas forests are reasonably consistent with those
predicted using mechanistic models for evergreen pine and deciduous broadleaf forests in Alabama, California, Colorado,
and Tennessee. Statewide annual ammonia emissions estimates, revised using the newly developed ammonia fluxes for oak
and pine forests in Texas, dropped from 921,000 to 467,000 metric tons. The relative contribution of ammonia emissions
from pine and oak forests dropped from 49% to less than 1%. Animal husbandry was predicted to be the dominant non-
industrial source, accounting for approximately 77% of non-industrial source ammonia emissions.
Atmospheric chemical reactions are believed to bea major source of fine particulate matter (PM2.5).An important contributor toward many of thesereactions is ammonia (NH3), which is emitted froma wide range of anthropogenic and natural sources.
ARTICLE IN PRESSG. Sarwar et al. / Atmospheric Environment 39 (2005) 7137–71537138
Tropospheric concentrations of NH3 are highlyvariable and dependent on proximity to sources,source strengths, meteorological conditions, andremoval mechanisms. Typical atmospheric resi-dence times for NH3 are on the order of 10 days,and ammonia mixing ratios over continents gen-erally range over two orders of magnitude, from 0.1to 10 parts per billion by volume (ppb) (Seinfeld andPandis, 1997). Dentener and Crutzen (1994) esti-mated global NH3 emissions to be 45 million metrictons per year, with approximately two-thirds of thistotal being attributed to anthropogenic activities.Nearly one-half of global ammonia emissions wereattributed to animal husbandry.
Chemical reactions involving NH3 to producesecondary PM2.5 depend on the presence and relativeconcentrations of atmospheric nitrates and sulfates. Inareas characterized by high ammonia and nitric acidconcentrations and low sulfate concentrations, gaseousammonia can react to form ammonium nitrate. In thepresence of sulfuric acid, increasing concentrations ofgaseous ammonia can react to form ammoniumsulfate. Whether reacted with nitrate or sulfate, theammonium ion (NH4
+) is often observed to be animportant component of tropospheric aerosols. Theconversion of ammonia to ammonium (NH4
+) is alsosignificant with respect to transport of NHx, since thedry deposition of ammonia gas is generally 5–10 timesfaster than dry or wet deposition of ammonium-containing particles (Bouwman et al., 1997).
There is significant evidence that natural soil is animportant contributor to global ammonia emissions(Dawson, 1977). For example, ammonium is foundat relatively high concentrations in rainwater.Gaseous concentrations of ammonia are also great-er over soils with high pH, a condition that shiftsthe acid-base equilibrium in soil from ammoniumion to ammonia. Atmospheric ammonia concentra-tions are greater over land than over oceans, andincrease with increasing soil temperature. However,measurements of ammonia emissions from naturalsoils are sparse and corresponding emission factorsare characterized by significant uncertainties. Thesefacts are particularly true for ammonia emissionsfrom forested areas, e.g., pine and oak forests thatcover large areas of east Texas. The primary sourceof nitrogen that is converted to ammonia is organicnitrogen associated with foliar litter. Thus, greateramounts of fresh litter deposition should lead toincreased ammonia emissions.
The intent of the study described herein was todevelop a first estimate of non-point source
ammonia emissions in Texas. A total of 64 non-point sources of ammonia were considered in thisstudy. Each source required significant reviews ofexisting literature and relevant databases prior tothe selection of appropriate emission factors andsource activity data. Given the extensive nature ofthese tasks, it is impossible to describe all aspects ofthe study in this paper. Instead, we have describedthe project methodology in general terms, and havelisted several important references and databases.The reader is referred to the complete project reportfor a more extensive description of methodologiesand results (Corsi et al., 2000a). We do providedetails related to actual ammonia flux measure-ments from forest floors in east Texas and use theresults to facilitate the overall ammonia inventory.
2. Methodology
An extensive literature review was completed inorder to identify potential non-industrial sources ofammonia emissions; As well as to identify andassess relevant emission factors. Ten bibliographicdatabases were searched using ‘‘ammonia’’ and‘‘emissions’’ (inclusive) as keywords. Forty websites were also found to contain information relatedto ammonia emissions, 14 of which were identifiedas relevant to this project. Personal contacts werealso made with individuals known to be, or who areknown to have been, involved with ammoniaemissions estimates. In total, 655 publications wereidentified as containing information relevant to thisproject. Approximately 120 of these publicationswere selected for thorough review. Through thisprocess, it quickly became evident that a smallnumber of previous publications are frequentlyreferenced and used by others to estimate ammoniaemissions (Asman, 1992; Battye et al., 1994; Bouw-man et al., 1997; Buijsman et al., 1987; Gharib andCass, 1984; Klaassen, 1991; and Lee and Dollard,1994).
2.1. Selection of source categories
Based on a review of existing literature, a total ofseven non-industrial source emission categorieswere selected for this study. Sixty-four sub-cate-gories that fall within the major source categoriesare listed in Table 1. While it was obvious at thebeginning of this study that some sources would berelatively insignificant, e.g., rabbits and untreatedhuman waste, for completeness emission factors and
ARTICLE IN PRESS
Table 1
Source categories considered in this study
Primary source
category
Sub-categories associated with each primary source
temperate forest (dense oak [1], other [6]); urban land [1]
Note: Numbers in brackets correspond to the number of reported emission factors for each source.
G. Sarwar et al. / Atmospheric Environment 39 (2005) 7137–7153 7139
activity data were sought for all source sub-categories. This paper will focus entirely on thosesource sub-categories that were estimated to emitgreater than 1000 metric tons yr�1 (mtpy) of ammo-nia on a statewide basis. Relevant sources areitalicized in Table 1.
2.2. Emissions estimation
Emissions for individual source sub-categorieswere estimated as the product of an emission factorand relevant activity data for that source:
Ei ¼ ðEFÞiAi, (1)
where Ei is the emission rate for source sub-categoryi (kg yr�1), EFi the emission factor for source sub-category i (kg unit activity�1, e.g., kg vehicle miletraveled�1), and Ai the activity level (measure ofactivity) for source sub-category i (e.g., milestraveled yr�1). Emissions estimates for each of the64 source sub-categories listed in Table 1 weredetermined for each of the 254 counties in Texas.County-specific emissions were estimated by allo-cating activity data to each county. Althoughtemporal variations in ammonia emissions are likelyto be significant for some source categories, e.g.,
fertilizer applications and natural soil and vegeta-tion, attempts to estimate temporal variations inammonia emissions were beyond the scope of thisstudy. All emissions were estimated on an annualbasis. County-level emissions for each of the sevenprimary source categories were estimated by sum-ming over relevant source sub-categories for eachcounty. Statewide ammonia emissions for each ofthe source sub-categories and primary sourcecategories were estimated by summing over allcounties. Total ammonia emissions for Texas wereestimated by summing over all 64 source sub-categories for all 254 counties in Texas. At therequest of the sponsoring agency, emissions werestandardized to a 1996 base year by using activitydata specific to 1996.
2.3. Selection of emission factors
A total of 348 emission factors were found for the64 source sub-categories described above. Thenumber of emission factors for each source is listedin brackets in Table 1. A tiered approach was usedfor selecting source-specific emission factors. Incases where two or more emission factors wereavailable, an attempt was made to carefully assess
ARTICLE IN PRESSG. Sarwar et al. / Atmospheric Environment 39 (2005) 7137–71537140
the origin of each factor. Emission factors wereexcluded from further analysis for any one or moreof the following: (1) incomplete description of factordevelopment; (2) exclusion of all processes that aregenerally associated with a source; (3) developmentunder conditions that are inconsistent with Texas;and (4) statistical inconsistency. In most cases, theremaining list of emission factors was then used todetermine an arithmetic average emission factor foruse in this study. Care was taken to avoid ‘‘extraweighting’’ of emission factors that were alreadyincluded in those emission factors reported asaverages of others. Emission factors by Harvey etal. (1983) were used for all on-road vehicles basedon the large number of vehicles that were used indeveloping the factors as well as adherence toFederal Testing Procedures. An emission factordeveloped by Asman (1992) was used for ammo-nium nitrate fertilizer. For those sources in whichonly one published emission factor was found, thatemission factor was used by default in this study.For example, values reported by either Asman(1992) or Bouwman et al. (1997) were used for allfertilizers that had only one reported emissionfactor.
Finally, for those 12 source sub-categories with-out reported emission factors, values were assumedbased on other sources, e.g., horses were used formules, burros and donkeys, ducks for geese, dieselengines for agricultural vehicles. For septic tanks, itwas assumed that the ammonia concentration in theheadspace of the tank is always at equilibrium withthe ammonia concentration in the underlyingwastewater. Emissions were then estimated usingHenry’s law and the assumption that gas displace-ment is equal to wastewater discharge to the tank.The contribution of these 12 source sub-categoriesto total statewide ammonia emissions was estimatedto be only 0.05%.
It is important to note that many of the emissionfactors, particularly for livestock, were developedbased on studies completed in Europe. The authorsacknowledge that ammonia emissions may differbetween livestock raised in Europe and in Texas dueto differences in diet. These differences may besignificant for sources such as dairy cows, and maybe underestimated in this study due to higherprotein diets for livestock in the United States.However, because of a lack of actual data forlivestock in the United States, nonetheless Texas, wehave adopted several emission factors that weredeveloped based on studies in Europe.
2.4. Acquisition of activity data
A wide range of information sources was used toobtain necessary activity data. For example, the USDepartment of Commerce’s Census of Agricultureas well as the Texas Department of Agriculture’sreport on Texas Agricultural Statistics was informa-tion sources for animal husbandry. The Associationof American Plant Food Control Officials and theUniversity of Kentucky were sources of informationon commercial fertilizer sales and applications. TheTexas Department of Transportation providedactivity data related to vehicle miles traveled byvarious types of vehicles. The Texas Commission onEnvironmental Quality (TCEQ) provided datarelated to municipal wastewater flows on acounty-by-county basis. Population data wereobtained from the US Census for relevant domesticsource sub-categories, as were data obtained fromthe American Veterinary Medical Association. Soiland vegetation coverage was estimated using theBiogenic Emission Inventory System (PC-BEIS).
Two very important sources that were limited to asingle emission factor were pine and oak forests.Use of the single emission factor led to an ammoniaemission estimate of 257,000mtpy for pine forestand 197,000 mtpy for oak forests in Texas, whichaccounted over 50% of the statewide annualemissions (Corsi et al., 2000b). However, estimatesfor pine and oak forests were characterized ashaving a great deal of uncertainty. As such, a seriesof field sampling events were completed to deter-mine ammonia emissions from pine and oak forestsin Texas.
2.5. Site selection process
A set of criteria for selecting appropriate loca-tions for field measurements was developed, whichincluded the mix of tree species at a given locationand requirements for monitoring equipment andactivities. Forests were chosen that were represen-tative of typical east Texas pine and oak forests.Virtually all the pine forests in east Texas consist ofLoblolly Pine, and are relatively homogeneous.Therefore, selecting sites in pine forests for ammo-nia emissions monitoring was fairly straightfor-ward. The situation was not as simple for oakforests. There are 28 species of oak found in Texas,and they occur in a wide range of terrains andvegetation type groupings.
ARTICLE IN PRESSG. Sarwar et al. / Atmospheric Environment 39 (2005) 7137–7153 7141
A screening analysis of potential sites was basedon land use/land cover (LULC) and biomass densitydatabases (Yarwood et al., 1997,1999; Wiedinmyerand Allen, 1999; Wiedinmyer and Strange, 1998). Adetailed examination of vegetation data indicatesthat Post Oak (Quercus stellata) was the predomi-nant oak species in Texas, with approximately 40%of the total oak biomass. This was followed by thetwo Live Oak species (Q. virginiana and Q.
fusiformis) comprising approximately 17% of theoak biomass, and Shin Oak (Q. sinuata) with 10%.Thus, suitable Post Oak forests were identified forpossible study sites.
In addition, potential monitoring sites had tomeet several requirements for the emissions mea-surement equipment and procedures. The site hadto be within the interior of the forest, and farenough from urban, agricultural, or other devel-oped areas to minimize edge effects and theinfluence of activities within these other areas.Electrical service had to be available near themonitoring sites to provide power for the testequipment. A minimum of two monitoring siteshad to be available within each selected forest toallow for multiple samples, intended to compensatefor spatial variation in ammonia emission charac-teristics.
2.6. Site descriptions
Field sampling was completed in two pine forests(Davy Crockett National Forest and Sam HoustonNational Forest), and two oak forests (Purtis CreekState Park and Cooper Lake State Park located ineast Texas). Each of these forests is hereafterreferred to as a sampling ‘‘location’’. Each locationwas visited twice during the course of the study.Four sampling sites were selected at each location.A single sampling event was completed at eachsampling site. A sampling event involved set-up ofexperimental instrumentation, at least one andpossibly two sets of multi-hour flux chambermeasurements, collection of soil samples, andambient air sampling.
Both Purtis Creek and Cooper Lake State Parkwere found to contain significant post oak standsand were used in this study. Purtis Creek State Parkis located about 3.5 miles north of the city ofEustace, Texas and about 15 miles northwest ofAthens, Texas. The monitoring site was located at321210N, 96100W. The park (designated as Oak 1)surrounds a man-made lake. The trees were
primarily Post Oak and Hickory, with someMulberry, Eastern Red Cedar (Juniper), CedarElm, and occasional other Oaks. The soil was looseand sandy, with organic matter primarily in the top5 cm. The monitoring site at Cooper Lake StatePark was located at 331220N, 951400W. Cooper LakeState Park (designated as Oak 2) also surrounds aman-made lake.
Based on site visits, the Davy Crockett and SamHouston National Forests were determined to besuitable as pine forest sites. Both forests include amixture of pine trees including Loblolly Pine,Shortleaf Pine and Longleaf Pine. However, Lo-blolly Pine is the dominant species. Sam HoustonNational Forest is in New Waverly, Texas. Themonitoring site was located at 301330N, 951390W.The monitoring sites in the Sam Houston NationalForest (designated as Pine 1) are in the StubblefieldRecreation Area, located adjacent to an oxbow lakeof the San Jacinto River. The soil was loose andsandy, with organic matter primarily in the top5 cm, due to decomposition of leaf litter. DavyCrockett National Forest is located near the city ofCrockett, Texas. The monitoring site is located at311240N, 951100W. The monitoring sites in the DavyCrockett National Forest (designated as Pine 2) arein the Ratcliff Lake Recreation Area. The vegeta-tion and soil are similar to those found in the SamHouston forest.
2.7. Ammonia emissions measurement
The ammonia emission measurement systemconsisted of three components: an ammonia analy-zer (Thermo Environmental Instruments (TEI),Model 17C), an emission isolation flux chamber,and a zero air generator (Advanced PollutionInstrumentation, Model 701). A schematic diagramof the ammonia emission collection and analysissystem is shown in Fig. 1 (Adapted from Ecklund,1992). The air generator supplied clean dry air tothe flux chamber. The flux chamber had a cylind-rical, stainless steel body and an acrylic dome cover.During use, the bottom of the body wall wasinserted approximately 3 in into the soil, sealing thesoil inside the chamber from the surroundingenvironment. The sweep zero air was introducedinto the chamber by a perimeter distribution tube atground level. The tube had evenly spaced holeswhich created jets of sweep air directed towards thecenter of the chamber. The flow rate of the sweep airwas kept greater than the flow drawn by the
ARTICLE IN PRESS
FLOWMETER
INLET
PLEXIGLASS
TOP7*
11*
16*
CUT AWAY TO SHOWSWEEP AIR INLET LINEAND THE OUTLET LINE
ZERO AIRGENERATOR STAINLESS STEEL
OUTLET LINE
REAL TIMEANALYZER
THERMOCOUPLE
PRESSURERELEASE
TEMPERATUREREADOUT
Fig. 1. Schematic diagram of ammonia emission collection and analysis system (Adapted from Ecklund, 1992).
G. Sarwar et al. / Atmospheric Environment 39 (2005) 7137–71537142
analyzer sample pump; the excess air leaving thechamber through an open port at the top of thedome cover. The chamber was thus maintainedunder slight positive pressure with a continuousoutward flow of air through the open port, isolatingthe chamber from the ambient atmosphere so thatthe sample feed to the analyzer contained onlyammonia emitted from the soil. During monitoringevents, the zero air flow rate to the chamber waseither 5.0 or 3.0 liters per minute (lpm), dependingon the specific test protocol; the sample pump drew0.8 lpm through the analyzer. A portable canopywith removable sidewalls protected the instrumentsfrom rain and prevented overheating of the ammo-nia analyzer.
Ambient temperature, relative humidity, andbarometric pressure were measured with a compactweather station. The flux chamber was placed on thesoil next to the analyzer system. Teflon tubing (1/4-in diameter) was used to provide the zero air feedand connected the sample port to the analyzer. Theopen port at the top of the chamber dome allowedfor air exhaust, pressure equalization, and access forsoil temperature measurements. The sample line washeated to prevent condensation of moisture in the
tubing, and also included a filter to preventparticulate matter from entering the analyzer. Arotameter was used to control the flow rate of zerosweep air to the flux chamber.
During field operations, the background signal ofthe TEI analyzer varied daily, and the instrumentzero levels were adjusted accordingly to compen-sate. The system was assembled with the bottom ofthe flux chamber covered with a sheet of vinylbefore it was set on the ground. Thus, only the zero-air sweep gas entered the flux chamber and wasdrawn into the sample line and analyzer. Theresulting response of the entire sample collectionand analysis system was used to verify that therewas no systematic error or bias in the results, andthat the system worked properly. This procedurewas also used to verify and adjust the instrumentbackground settings.
A period of 60–90 min after system startup wasrequired for the converters in the analyzer to reachoperating temperature and for the system tostabilize. Concurrently, the zero air fed into thechamber purged residual ammonia and nitrogenoxides (NOx) from the chamber, sampling lines, andanalyzer, ensuring that the system response was due
ARTICLE IN PRESSG. Sarwar et al. / Atmospheric Environment 39 (2005) 7137–7153 7143
to a true blank sample. The background test runcontinued until the indicated nitric oxide (NO) andNH3 levels stabilized and remained relatively steadyfor 5 min, at which time the system zero level wasset. The run continued for a minimum of fiveadditional minutes to ensure that stable zero-levelconcentration measurements of NO and NH3 weremaintained. If necessary, the zero setting wasrepeated. Once a reliable background adjustmentwas made, the analyzer and flux chamber systemwere considered to be ready for actual emissionmeasurements.
Primary flux measurements refer to the initialdynamic measurements made when the flux cham-ber was first placed on the soil at a test site, afterbackground adjustments had been made. Someambient air, primarily entrained in the surface soiland leaf litter, was trapped in the flux chamber itselfas it was placed on the soil; however, the constant,positive flow of sweep zero air through the chamberminimized this effect. The residual ambient air wasfully swept from the chamber after the first25–30 min of the primary flux test run; the emissionsafter this time were solely from soil and surfaceorganic matter (forest leaf litter). The primary fluxmeasurements continued until a steady reading wasreached.
To begin primary flux measurements, the flowrate of the sweep air was increased to 5.0 lpm,and the cover was temporarily kept on thebottom of the flux chamber to establish a base-line NH3 concentration reading. After approxi-mately 5min, the flux chamber was lifted offof the soil just enough to slide the cover out.The chamber was then pressed into the soil, withthe outer wall of the chamber inserted to a depthof 2–3 in. During primary flux measurements,hourly readings were made of the ambient tempera-ture, relative humidity, and barometric pressure.Temperature measurements were also made of thesweep air entering the flux chamber, and of the soilat the surface and depths of 5, 10, 15, 20, 25, and30 cm.
In general, the flow of the sweep air wassignificantly higher than the NH3 emission ratefrom the soil, and the progressive dilution causedthe indicated NH3 concentration to decay to belowdetection limit (1 ppb) within 3–4 h for all but oneevent. In ten of sixteen cases, once the indicatedconcentration reached a steady level below 0.5 ppb,the sweep air feed was shut off, the analyzer sampleline was disconnected, and the sample and vent
ports were capped. The chamber was thus sealed,and the area within was isolated from surroundingambient air. The chamber was sealed for a mini-mum of 2 h, during which time the NH3 concentra-tion from soil emissions again increased to ameasurable level. The peak ammonia concentrationin the sealed (static) chamber was used to determinean ammonia flux during this recovery period. Thistechnique is referred to herein as ‘‘static chambermeasurement’’.
After the flux chamber was sealed following theprimary flux run, the analyzer sample line wasconnected to the zero air supply for purging.Measurements were then made of the ammoniaconcentration in the ambient air, with the inlet ofthe ammonia analyzer sample line placed approxi-mately six feet above the ground. Following theambient air measurements, the sample lines andanalyzer were again purged with zero air. Thesample line was then reconnected to the fluxchamber, and the zero sweep air flow restarted inorder to measure the ammonia that had accumu-lated while the flux chamber was sealed. The initialconcentration measurement was taken as recupera-tion of ammonia levels due to emissions from thesoil. As with the primary runs, the secondary fluxmeasurements continued until a stable (typicallyzero) ammonia concentration measurement wasreached.
Several measures were taken to ensure the qualityand accuracy of the emissions data. These coveredthe physical aspects of the collection system,calibration of the analyzer, and sampling opera-tions. Because ammonia is a ‘‘sticky’’ compoundthat tends to adsorb to surfaces, the flux chamberwas made of electro-polished stainless steel and thesample lines were lined with PFA Teflon (the leastporous type) in order to minimize this effect. Inaddition, the sweep zero air was introduced into theflux chamber radially inward from the wall so thatthe main air flow direction was toward the samplingand exit ports at the top of the chamber near thecenterline. The sweep air flow rate gave a completechange of air in the chamber every 6min, yielding achamber response time to a step change inconcentration of 25–30 min. The sample line fromthe chamber to the analyzer was relatively short,approximately 2 m. At the 0.8 lpm sampling rate,the residence time of air in the line was slightly lessthan 5 s, giving a 20–25 s response time. A 65 Wcontinuous electrical heating strip kept the sampleline temperature elevated to prevent condensation
ARTICLE IN PRESSG. Sarwar et al. / Atmospheric Environment 39 (2005) 7137–71537144
of moisture and to minimize ammonia sorption onthe tubing walls.
The TEI ammonia analyzer was calibrated usinga TEI Model 146 Dynamic Gas Calibrator, withsupply gases of 20 parts per million by volume(ppm) NO and 10 ppm NH3, diluted with zero air.Part of the NO was converted in the calibrator toNO2 with internally generated ozone, controlledwith gas-phase titration. The analyzer was cali-brated using inlet concentrations of 100 ppb for NO,NO2, and NH3, matching the expected peak forestemission concentration. The analyzer works bypassing the sample stream through various combi-nations of high-temperature converters and scrub-bers to determine the concentrations of NO, NOx,and total Nx in the sample. The ammonia concen-tration is calculated from the difference between thetotal Nx and NOx measurements. The calibration ofthe analyzer was checked in the laboratory periodi-cally between and after forest measurement activ-ities; no significant changes were found.
Extensive calibration tests of the sampling andanalysis system (including the flux chamber) showedthat the time to reach equilibrium for a highammonia concentration, e.g., 100 ppb sample feed,was approximately 150–175 min. The decay time ofthe indicated ammonia level after switching back tozero air was shorter, typically 90–120 min. Actualpeak ammonia concentrations measured in the fieldwere less than 5 ppb, with 3–4 ppb peak levelstypical. Based on preliminary calibration tests it isreasonable to assume an upper bound of 25–30 minfor the system to reach equilibrium with a sampleconcentration of 10 ppb. The sample times for bothprimary and secondary flux periods were 3–5 h, wellbeyond the time necessary for any transient sorp-tion/desorption effects to be neutralized. In addi-tion, the emission rates were calculated using theintegrated measurements over the entire samplingperiod, which compensates for transient effects.These characteristics, combined with the practice tocontinue a flux measurement period until theindicated ammonia concentration was at the limitof detection (0.5 ppb or less) ensure that anyammonia which may have adsorbed to the chamberor sampling lines was removed by the sweep zero airas the test progressed. This effective purging of thesystem further ensured that any ammonia that mayhave adsorbed to the flux chamber walls during theperiod when the chamber was sealed was thendesorbed and collected by the analyzer during thesecondary flux measurement periods.
2.8. Soil samples
Following flux measurements, soil samples werecollected from within the flux chamber, a second siteimmediately adjacent to the chamber, and at a thirdsite approximately 20–30 feet away from thechamber. The samples from inside and outside ofthe chamber were compared to determine the effectof the flux measurement procedures on moisture,ammonia, and other organic compounds present inthe soil. The third sample was used to determinelocal variability in soil properties. Samples weretaken of the surface leaf litter and of the soil atdepths of 5, 10, and 15 cm. Leaf litter and soilsamples were collected in 475 and 120 ml borosili-cate glass jars with Teflon seals, respectively. Thesample jars were then enclosed in two layers ofzipper-seal polyethylene bags and placed in a coolerwith ice for transport back to the laboratory. Thesamples remained refrigerated at 4 1C until analysis.The soil and leaf litter samples were analyzed forammonia and nitrogen using EPA Methods 350.1and 350.2 (EPA, 1983). The pH and moisturecontent of the soil and leaf samples were determinedbased on EPA Methods 150.1 and 160.3, respec-tively. In all cases, representative sub-samples of thesoil or leaf litter stored in the sampling jars wereselected for ammonia, pH and moisture contentanalyses.
2.9. Calculation of emission factor
The objective of flux chamber measurements wasto determine gaseous ammonia emissions per unitarea of forest floor. For primary flux chambermeasurements, emissions flux was determined usingEq. (2):
Ef ¼QðCout � CinÞ
A, (2)
where Ef is the ammonia flux (mg m�2 min�1), Q isthe volumetric flow rate of air through the chamber(m3 min�1), Cout is the concentration of ammoniaexiting chamber (mg m�3), Cin is the concentrationof ammonia entering with chamber inlet air(mgm�3), and A is surface area over which fluxchamber is placed (0.13 m2). For this study, Q was0.005 m3 min�1, the volume of the flux chamber was0.025 m3. Since a zero air supply-conditioning unitwas employed, Cin was always zero, a fact that wasconfirmed for each sampling event.
ARTICLE IN PRESS
Table 2
Summary of measured ammonia emission factors (Summer
Months)
Forest name Forest type Emission factor (kg km�2 month�1)
Dynamic Static
Sam Houston Pine o 1.2 —
1.2 —
o 1.2 0.07
o 1.2 0.11
Average 0.3–1.2 0.09
Davy crockett Pine o 1.2 0.05
o 1.2 0.10
o 1.2 0.12
o 1.2 0.07
Average o 1.2 0.08
Purtis creek Oak o 1.2 0.12
o 1.2 0.12
o 1.2 0.17
o 1.2 —
Average o 1.2 0.14
Cooper Lake Oak o 1.2 —
o 1.2 —
o 1.2 0.08
o 1.2 —
Average o 1.2 0.08
G. Sarwar et al. / Atmospheric Environment 39 (2005) 7137–7153 7145
The manufacturer-specified detection limit for theammonia analyzer used in this study is 1 ppb, whichcorresponds to an ammonia concentration ofapproximately 0.7 mg m�3 at 20 1C and 1 atmo-sphere. At 0.7 mgm�3, the minimum quantifiableemission factor is approximately 14 kg km�2 yr�1.In all cases, measurements were made until stableconcentrations were achieved. Any stable, non-zeroconcentration less than 0.7 mg m�3 was taken to beless than the manufacturer-specified detection limit,and corresponding emission factors were reportedaso14 kg km�2 yr�1 (o1.2 kg km�2 mos�1).
The ammonia emission rate per unit area of forestfloor during static chamber measurement wasestimated using Eq. (3):
Ef ¼m
DtA¼
CV
DtA, (3)
where, m is mass accumulated in head space overtime Dt (mg), Dt is time from initial sealing ofchamber to ammonia measurement (min), A issurface area of forest floor covered by chamber(m2), C is the ammonia concentration measured inchamber air after Dt (mgm�3), and V is chamber headspace volume (m3). For most static chamber experi-ments, the value of Dt was on the order of 150min ormore. Using this value and a minimum quantifiableconcentration of 0.7mgm�3, Eq. (3) leads to anapproximate lowest quantifiable static chamber fluxrate of 0.46 kgkm�2 yr�1. In fact, for static chambermeasurements, Ef was greater than 0.46kgkm�2 yr�1,but less than the primary flux chamber measurementquantifiable limit of 14kgkm�2 yr�1.
3. Results and discussion
3.1. Emission factors
A summary of emission factors resulting fromsummertime measurements in both pine and oakforests of east Texas is presented in Table 2. Resultsassociated with both dynamic and static chamberexperiments are presented in Table 2. Dynamicexperiments were attempted at each site. Emissionfactors for 15 of the 16 dynamic experimentswere less than 14 kg NH3 km�2 yr�1 (1.2 kg NH3
km�2 month�1). The emission factor for the remain-ing dynamic experiment was 14 kg NH3 km�2 yr�1
(1.2 kg NH3 km�2 month�1). It is important torecognize that all field experiments were completedduring the summer months of 2001, and some
temporal variations in ammonia emissions arepossible.
Although static chamber monitoring was notemployed at every site, it did lead to consistentresults among those sites where it was employed.Emission factors derived from static chambermeasurements for pine forests ranged between0.05–0.12 kg NH3 km�2 month�1 and in oakforests between 0.08–0.17 kg NH3 km�2 month�1.Arithmetic mean emission factors based on theuse of the static chamber method at SamHouston and Davy Crockett National Forestswere 0.09 kg NH3 km�2 month�1 and 0.08 kgNH3 km�2 month�1, respectively. The arithmeticmean emission factor over four static chamberexperiments in oak forests was 0.13 kgNH3 km�2 month�1. The reader is cautioned thatstatic flux measurements are not generally employedor described in the published literature. Never-theless, due to the limitations associated withdynamic chamber experiments, we opted to usestatic chamber results as being representative ofsummertime emission factors for pine and oakforests in east Texas.
ARTICLE IN PRESSG. Sarwar et al. / Atmospheric Environment 39 (2005) 7137–71537146
Before comparing the emission factors derivedfrom this study with previous measurements orestimates, it is instructive to review the state ofknowledge related to ammonia emissions fromundisturbed soil, particularly as related to forestlitter and soil. Of particular significance are theconflicting arguments made about ammonia emis-sions from undisturbed soil and from forests, andthe sparseness of measured emissions. Kim (1973)was the first to report measurements of ammoniaemissions from forest floors. Our work is the secondstudy focusing on the measurements of ammoniaemissions from forest floors. Most other investiga-tors either measured or modeled ammonia concen-trations above the forest canopy.
Using the results of a simple ammonia emissionsmodel for undisturbed non-fertilized soil, Dawson(1977) argued that undisturbed land is likelythe primary source of global ammonia emissions.However, Dawson (1977) acknowledged that,as of 1977, emission from uncultivated andunfertilized vegetated land had not been measured.Fifteen years later, Langford et al. (1992) noted thatgaseous ammonia fluxes in unmodified forests werevirtually non-existent. Schlesinger and Hartley(1992) indicated that little is known about thevolatile loss of ammonia from non-agricultural soilsin which ammonium (NH4
+) is derived frommineralization of organic nitrogen. Asman et al.(1998) observed that emissions from B-Napus
canopies following the deposition of leaf litterappear to be ‘‘significant’’. They recommended thatdecomposition of leaf litter as a ground source ofammonia emissions needs further investigation.However, Pryor et al. (2001) suggested that soilconditions, particularly surface soil pHo6.5, pre-cludes a large efflux of ammonia emissions fromforest floors.
The emission factors determined in this study aresignificantly lower than any measured previously forpine and oak forests. However, it is important tonote that the forests used in this study were notartificially fertilized, while several of those thatformed the basis for previous studies had beenamended with urea-nitrogen (Camire and Bernier,1981; Marshall and DeBell, 1980; Overrein, 1968).The remaining two emission factors (Kim, 1973;Langford and Fehsenfeld, 1992) for pine/coniferousforests differ by a factor of 500, suggesting thedifficulties and potential errors associated withselecting a single emission factor for estimatingammonia emissions from forests.
The emission factors reported by Kim (1973) forpine and oak forests in South Korea are approxi-mately four orders of magnitude greater than theemission factors determined in this study. Theemission factor reported by Langford and Fehsen-feld (1992) for coniferous forests is 32 times greaterthan the summertime pine forest emission factordetermined for this study. A small numberof researchers have predicted ammonia fluxes fromforest soils based on mathematical modelsthat differ significantly in complexity. Bouwmanet al. (1997) developed a simple model to estimateglobal emissions of ammonia and estimated0.03 gm�2 yr�1 (2.5 kg km�2 month�1) ammoniaemissions from temperate forests, a value approxi-mately 20–30 times greater than those measured inthis study for east Texas. However, the Bouwmanet al. model did not account for soil pH effects andmay therefore underestimate emissions from alka-line soils, while potentially over-estimating emis-sions from acidic soils, e.g., soils in east Texas.
Dawson (1977) developed a more sophisticatedmodel than Bouwman et al. (1997) to estimateglobal emissions of ammonia from un-disturbedland. His model accounted for pH effects onammonia/ammonium equilibrium partitioning insoil and considered the degree of exchangeableammonium resulting from a balance of litterdecomposition and nitrification. Resulting emis-sions from un-disturbed soil within 30–401N lati-tude were predicted to be 9.2 billion kg�1 yr�1 overan area of 15.57 million km2. This leads to anemission factor of 49 kg km�2 month�1, approxi-mately 500 times greater than forest floor emissionfactors derived in this study for east Texas. Dawson(1977) did not separate forest emissions from otherland types.
Langford et al. (1992) used the Dawson (1977)model to estimate ammonia emissions from forestfloors using soil-specific properties (surface pH andammonium concentration) from three forests in theUnited States. For a montane coniferous forest inColorado (pH ¼ 5.22; NH4
+¼ 17.3 g kg�1) they
estimated an emission factor (normalized here to amonthly average) of 0.25 kg km�2 month�1. For atemperate coniferous forest in Alabama (pH ¼ 5.3;NH4
+¼ 5.6 g kg�1) they estimated an emission
factor of 0.10 kg km�2 month�1. For a temperatedeciduous forest in Tennessee (pH ¼ 4.7;NH4
+¼ 17.9 g kg�1) they estimated an emission
factor of 0.05 kg km�2 month�1. These emissionfactors in US forests bound the values that were
ARTICLE IN PRESSG. Sarwar et al. / Atmospheric Environment 39 (2005) 7137–7153 7147
derived based on our measurements in east Texas.While the assumed soil surface temperature of 20 1Cwas less than those observed in this study, the soilpH and ammonium concentration levels werereasonably consistent with those measured in eastTexas.
Potter et al. (2001) employed the NASA–CASAmodel within a GIS framework and completed arigorous modeling evaluation of ammonia emissionsfrom native soils in California. The model accountsfor area water balances, soil pH and moisturecontent, litter fall, nutrient allocation, soil nitrogenmineralization, seasonal carbon fixation, surfacetemperature, and soil ammonia emissions. Whilerecent ammonia flux estimates for California aregreater than those measured in east Texas, Potter etal. (2001) predicted significant seasonal variations inammonia emissions, with 20-fold or greater differ-ences in predicted emissions between differentmonths. Soil moisture was predicted to have asignificant influence on ammonia emissions, withmuch lower emissions from moist soils owing tolower gas diffusivities. Soil pH was also predicted tohave a significant effect on ammonia emissions, withthe highest emissions predicted to occur for condi-tions of high pH and low moisture content. Inseveral California counties that are dominated byevergreen forests the soil pH ranged from 5.5 to6.19. For comparison, six of the eight pine forestlocations that we studied in east Texas had soilsurface (5 cm depth) pH of less than or equal to 5.5,and 7 of 8 oak forest locations had soil surface pHless than 5.3. Using the same algorithm that wasemployed in the NASA–CASA model, we estimateda 20-fold reduction in ammonia flux for a pH dropfrom 6 to 5 at a surface temperature of 30 1C,everything else being equal.
Based on a version A (moderate pH effects)model application, Potter et al. (2001) estimatedannual ammonia emissions from evergreen needle-leaf forests in California to be 810 mtpy over a totalarea of 12.4 million hectares. This translates to anaverage emissions flux of 0.54 kg km�2 month�1, sixtimes greater than the summertime pine forestemission factor determined in this study for eastTexas. Similarly, the average emission flux fordeciduous broadleaf forests was estimated to be2.6 kg km�2 month�1, 20 times greater than thesummertime oak forest emission factor determinedin this study for east Texas.
The relatively small differences in emissionfactors between sites in this study preclude a
rigorous evaluation of the effects of soil/literproperties on ammonia emissions from forests ineast Texas. However, the relatively low emissionsfactors are consistent with low values of soil pH.
3.2. Soil properties and their effects
The mean soil temperature, pH, moisture contentand ammonia concentration observed at each of theforest monitoring sites are summarized in Table 3.The highest moisture levels were found within theleaf litter and lower, but relatively constant,moisture contents were observed within 0–15 cmsoil depths. The soil at Davy Crockett NationalForest had the highest moisture content. Incontrast, the soils at the remaining sites wererelatively dry, with mean soil moisture contentsranging from 2.8% to 7.5%. In general, soils withmoderate moisture content are expected to releasethe greatest fraction of ammonia. If the soil is toodry, microbial activity may be inhibited and the soilcan adsorb ammonia directly; if the soil is too wet, itinhibits the diffusion of ammonia to the soil surface.
The NH3–N concentration measured in the soiland leaf litter samples varied from 0.4 to47.3 mg kg�1. As expected, the average ammoniaconcentration in the leaf litter across all the forests(i.e., 18 mg kg�1) was greater than the concentrationin the underlying soil (e.g., 5.3 mg kg�1 for the0–5 cm soil horizon). Results from these analysesindicate that, in general, the ammonia concentra-tions in the soil continued to decline with depthfrom the surface. These variations with depthsupport the common assumption that the surfacelitter and near-surface soils are the most importantregions to consider when quantifying ammoniaemissions.
As expected, soil temperatures were greatest atthe surface and declined with depth. Since fluxexperiments were conducted over different periodsof the day and night, the associated soil tempera-tures observed during the experiments followdiurnal variations. The minimum soil temperatureobserved was 22.7 1C and the maximum observedon a hot afternoon when the ambient temperaturewas nearly 36 1C was 30.9 1C. In general, hightemperatures are expected to increase ammoniavolatilization; however, the low pH of the soilspresent in Texas forests seems to be an over-riding factor that ultimately suppresses ammoniaemissions.
ARTICLE IN PRESS
Table 3
Mean soil properties at forest monitoring sites
Site Sam Houston Davy Crockett Purtis Creek Cooper lake Range observed
Min Max
Temperature (1C)
Soil: surface 26.2 25.3 27.7 28.9 23.7 30.9
Soil: 5 cm down 25.1 23.8 25.5 26.2 22.7 27.1
Soil: 10 cm down 24.8 24.4 25.1 25.9 23.3 26.6
Soil: 15 cm down 24.6 24.1 24.4 25.7 23.1 26.1
pHa
Leaf litter 5.3 5.3 5.3 5.9 4.2 6.3
Soil: 5 cm down 5.5 5.2 5.1 4.8 4.3 6.1
Soil: 10 cm down 5.7 5.4 5.6 b b b
Soil: 15 cm down 5.6 5.4 5.6 b b b
Moisture Content (%)c
Leaf litter 5.0 27.9 22.8 14.2 11.6 36.3
Soil: 5 cm down 6.8 12.5 2.8 7.5 1.8 20.0
Soil: 10 cm down 6.7 10.9 2.8 b 1.4 17.3
Soil: 15 cm down 6.5 11.4 3.1 b 2.2 14.8
NH3–N (mgkg�1)a
Leaf litter 20.7 11.5 27.7 10.5 0.7 47.3
Soil: 5 cm down 10.3 3.8 1.8 5.3 0.4 23.1
Soil: 10 cm down 5.5 5.4 b b 5.4 5.5
Soil: 15 cm down 2.8 b b b 2.8 2.8
aSoils inside flux chamber.bSample not collected.cSoils immediately adjacent to flux chamber.
G. Sarwar et al. / Atmospheric Environment 39 (2005) 7137–71537148
3.3. Ambient ammonia concentrations in forest
canopy
The mean of in-canopy ambient ammonia con-centration measured at each site varied over arelatively narrow range of 1.5–2.9 ppb, reasonablyconsistent with a value of 1.7 ppb reported by Pryoret al. (2001) measured above deciduous forests insouthern Indiana during the fall. However, thearithmetic mean value measured in east Texas is afactor of 4–6 greater than springtime measurementsover the same Indiana forests and mean summer-time (day) surface concentrations in forests asreported by Langford et al. (1992).
3.4. Omission of canopy effects
Several recent studies have indicated that canopyvegetation (leaves and needles) can significantlyaffect ammonia emissions from forests due to twophenomena: re-absorption of ammonia and vegeta-tive emissions of ammonia. The net effects of forestcanopies on ammonia emissions appear to be
dynamic, with canopies sometimes serving as a netsink and sometimes serving as a net source ofammonia. Andersen et al. (1993) measured ammo-nia fluxes above spruce forests in Denmark. Theyobserved the canopy to be a net source of ammoniaemissions during 10 of 34 experiments, and a netsink during 24 experiments. Wyers and Erisman(1998) studied the exchange of ammonia overconiferous forests in the Netherlands for over twoyears. They observed vertical ammonia fluxes tovary in direction (to and from the top of thecanopy). They discussed stomatal release, drying ofleaf surfaces, and ammonium aerosol evaporationas potential ammonia sources within forests. Pryoret al. (2001) also measured ammonia fluxes over adeciduous broadleaf forest in southern Indiana.They observed that, on average, the forest canopywas a sink of ammonia. They did observe a reverseflux (canopy as net source of emissions) on somespring days, with a magnitude of up to0.2 mg m�2 h�1. They attributed ammonia sourcesto stomatal release, leaf drying, and evaporation ofammonium nitrate particles.
ARTICLE IN PRESSG. Sarwar et al. / Atmospheric Environment 39 (2005) 7137–7153 7149
In none of the aforementioned studies did theresearchers actually measure ammonia emissionsfrom forest litter and soil. It is also important torecognize that fluxes to and from the forest canopieswere all made above the canopy. Thus, canopy‘‘sink effects’’ all relate to uptake from the overlyingambient atmosphere as opposed to the underlyingterrestrial environment. Measurements of ammoniaconcentrations within the forest canopy, includingabove the soil surface, were not found in thepublished literature, nor was any discussion of thepossibility of horizontal transport of ammonia inthe air space located above forest soil but below themain portion of the vegetative canopy.
Forest canopies may have a significant influenceon net ammonia emissions from forests. However,the current knowledge base related to canopyemissions and uptake do not allow for accuratepredictions of these phenomena. Given the rela-tively low emission fluxes measured for pine andoak forests in Texas, canopy uptake would not havea significant effect on overall non-industrial sourceammonia emissions in Texas. However, the impactof canopy emissions cannot be ascertained based onthe results of this study.
3.5. Statewide ammonia emissions estimates for non-
industrial sources
Updated statewide non-industrial source emis-sions estimates for ammonia are listed in Table 4 foreach primary source category and source sub-category that exceeds 1000 mtpy. The predictedstatewide total ammonia emissions for the year1996 were 467,000mtpy. The dominant source waspredicted to be animal husbandry, which wasestimated to contribute 77% of total ammoniaemissions. Within this category, generic cows, beefcows, and horses alone were estimated to contribute65% of non-point source ammonia emissions inTexas. As described earlier in this paper, the readeris cautioned that ammonia emissions estimates forlivestock, and especially cows, are subject tosignificant uncertainty due to their development inEurope where the diet of livestock can differconsiderably from those in the United States.Fertilizer applications were estimated to accountfor almost 8% of non-point source ammoniaemissions in Texas, and were distributed betweensix different types of fertilizer. Natural soil/vegeta-tion was estimated to be responsible for almost 7%of total non-point ammonia emissions in Texas, a
value that is close to that predicted by Dentener andCrutzen (1994) for the natural soil and vegetationcontribution to global ammonia emissions. Withinthis category, scrublands and grasslands wereestimated to contribute about 5% of non-pointsource ammonia emissions. Domestic sources wereestimated to contribute about 6% of non-pointsource ammonia emissions in Texas, with emissionsfrom dog and cat urine contributing nearly 2/3 ofthose emissions. On-road vehicles were estimated tocontribute only 2% of non-point source ammoniaemissions, with nearly all of these emissionsoriginating from light-duty gasoline engines withthree-way catalysts. Municipal wastewater wasestimated to contribute less than 1% of non-pointsource ammonia emissions in Texas.
We utilized emissions factors for pine and oakforests from a 1972 study in South Korea (Kim,1973) in our previous estimates of annual ammoniaemissions in Texas (Corsi et al., 2000b). The impactof replacing Kim’s emission factors with thosemeasured during this study is enormous. Predictedstatewide non-point source ammonia emissions arereduced by almost a factor of two, from921,000mtpy to 467,000mtpy. Predicted statewideammonia emissions from pine forests, locatedprimarily in east Texas, are reduced from257,000mtpy to only 16 mtpy. Similarly, predictedstatewide ammonia emissions from oak forests arereduced from 197,000mtpy to 22 mtpy. Revisedammonia emissions estimates for oak and pineforests in Texas are small and not included in Table4. Reductions in predicted ammonia emissions frompine and oak forests have a significant effect on theoverall contribution of natural soil and vegetationand animal husbandry to statewide non-pointsource ammonia emissions. The natural soil andvegetation source category drops from 52% to just3% of statewide non-point source emissions. Theoverall contribution of animal husbandry to non-point source ammonia emissions increases by afactor two, from 39% to 77%.
There are several reasons why the previousestimates for natural soil/vegetation are suspect.First, only one emission factor was obtained forpine and oak forests from the published literature.These factors were based on research involving alimited amount of experimental data that werecollected in 1972. Significant advances in analyticalinstrumentation have occurred in the past threedecades. The emission factors were collected in pineand oak forests in South Korea, and it is not clear
ARTICLE IN PRESS
Table 4
Sources emitting greater than 1000 metric tons yr�1
Fertilizer application kgmetric ton�1 of N applied 37
N-P-K 49 4 12
Anhydrous ammonia 49 6 10
Other nitrogen solutions 30 6 6
Urea 121 3, 9, 13 5
Ammonium sulfate 140 4, 9 3
Ammonium nitrate 24 4 1
Natural Soil/Vegetation kgm�2 yr�1 31
scrublands 0.0001 6, 18 12
grasslands 0.00004 6, 18 10
other temperate forests 0.0004 6, 15, 18 6
Urban land area 0.0004 15 2
Domestic sources kg/head/yr 27
Dogs 2.18 7,8,10,13,15 12
Humans 0.44 3,5,7,8,10,13,16,19 8
Cats 0.69 7,8,10,13,15 5
On-road vehicles kg vehicle mile traveled�1 10
Light-duty gasoline engines
w/ 3-way catalysts 0.0001 20 10
Municipal wastewater kgmillion gallons wastewater�1 6
POTWs 8.6 14 6
Statewide total 467
Notes: For some source categories only one reference is provided; emission factor from this reference was directly used in this study for
these categories. Where multiple references are provided, an average emission factor was calculated from the listed references and was used
in this study. (1—ApSimon et al., 1987; 2—Kruse et al., 1986; 3—Dianwu and Anpu, 1994; 4—Asman, 1992; 5—Battye et al., 1994; 6—
Bouwman et al., 1997; 7—Lee and Dollard, 1994; 8—Heisler et al., 1988; 9—Buijsman et al., 1987; 10—Dickson et al., 1991; 11—Sadeghi
and Dickson, 1992; 12—Roe and Strait, 1998; 13—Sutton et al., 1995; 14—Warn et al., 1990; 15—Gharib and Cass, 1984; 16—Moller and
Schieferdecker, 1989; 17—Kruse et al., 1989; 18—Schlesinger and Hartley, 1992; 19—Atkins and Lee, 1993; 20—Harvey et al., 1983).
G. Sarwar et al. / Atmospheric Environment 39 (2005) 7137–71537150
from the original source as to whether the soil orcanopy conditions in the forests that were sampledwere consistent with those that are observed in theforests of Texas.
3.6. Rural versus urban counties
As described above the dominant non-industrialsource of ammonia emissions is predicted to be
associated with animal husbandry. This is particu-larly true in rural counties, many of which arecharacterized by an animal husbandry contributionthat exceeds 85%. We also predicted the relativecontributions of various source categories to non-industrial ammonia emission in urbanized countiesin Texas. Results are listed in Table 5. Bexar Countycontains the greater San Antonio area, DallasCounty the City of Dallas, El Paso County the City
ARTICLE IN PRESS
Table 5
Contribution to non-industrial ammonia emissions in urbanized counties
County NH3 Emissions
(metric tons yr�1)
Domestic (%) Animal
husbandry (%)
On road
vehicles (%)
Wastewater (%) Fertilizer (%)
Bexar 4800 39 37 14 9 2
Dallas 7100 41 27 18 14 1
El Paso 2800 35 41 9 7 9
Harris 10,900 44 26 15 10 5
Tarrant 4500 45 29 17 7 3
Travis 2600 34 38 15 10 3
G. Sarwar et al. / Atmospheric Environment 39 (2005) 7137–7153 7151
of El Paso, Harris County the City of Houston,Tarrant County the City of Ft. Worth, and TravisCounty the City of Austin.
As expected, the results for urbanized counties inTexas are considerably different than those for ruralcounties. Contributions due to animal husbandryare still relevant, but approximately 1/2 to 1/3 of thecontribution in most rural counties. Contributionsdue to fertilizer usage are relatively small, except forEl Paso County, which has a significant base ofagricultural activity along the Rio Grande River.On-road vehicles are predicted to emit slightly moreammonia than publicly owned treatment works(wastewater).
Domestic sources exceed those of animal husban-dry in four of the six urban counties listed in Table5, and exceed animal husbandry when emissionsfrom these counties are summed. As listed inTable 1, domestic sources are individually smallbut extremely numerous in urban areas, e.g.,domestic cats. Interestingly, we predict the majorcontributors to domestic sources to be humans(perspiration and respiration), and dogs and cats(primarily urine). In fact, dogs and cats arepredicted to contribute nearly 2/3 of domesticnon-industrial ammonia emissions in urban coun-ties, on the order of, or greater than, emissions fromon-road vehicles and wastewater combined. Forexample, in urban areas cats and dogs are bornthrough uncontrolled reproduction at rates esti-mated as high as seven times the birth rate ofAmericans, but the numbers are characterized by ahigh degree of uncertainty and likely vary consider-ably between cities based on the effectiveness ofspray/neuter programs. This surprising result in-dicates a need for additional studies to reduceuncertainties in emissions estimates for this sourcecategory.
4. Conclusions
A summer field study was conducted to determineammonia fluxes from pine and oak forest floors ineast Texas. Both dynamic and static chambermethods were employed. Dynamic chamber experi-ments proved to be too insensitive to determineammonia fluxes. Thus, static chamber results wereused to estimate ammonia fluxes from forest litterand soil. The ammonia flux averaged0.09 kg km�2 month�1 for pine forests and0.13 kg km�2 month�1 for oak forests during thesummertime monitoring period. These values aresignificantly lower than those previously measuredand cited in the published literature, and areapproximately four orders of magnitude less thanemission factors that were employed for pine andoak forests in a previous non-industrial sourceammonia emissions inventory for Texas. Lowammonia emissions from pine and oak forest floorsin east Texas are likely influenced greatly by theacidic nature of forest litter and surface soils. Anestimate of non-point source ammonia emissions inTexas was developed. Animal husbandry is pre-dicted to be the largest non-point source ofammonia emissions in Texas, with cattle predictedto emit greater than 280,000 metric tons ofammonia in 1996. Fertilizer application is thesecond largest non-point source of ammonia emis-sions. Temporal variations were not considered inthis study and emissions from fertilizer are likely tobe much greater than annual average emissionsduring specific months.
Acknowledgments
This study was funded by the Texas Commissionon Environmental Quality. The authors wish to
ARTICLE IN PRESSG. Sarwar et al. / Atmospheric Environment 39 (2005) 7137–71537152
acknowledge Mr. Steve Anderson for his guidance.The authors acknowledge contributions from thefollowing students of The University of Texas atAustin: Stacey Fredenberg, Marie Dondelle,Katherine Dombrowski, Satoshi Takahama, andWidianto. The authors also acknowledge the TexasParks and Wildlife Department, and the UnitedStates Department of Agriculture for their approvalof setting up the instruments and for takingmeasurements in the parks.
