Emission fluxes of ammonia, N2O, and NO from the terrestrial biosphere are of great interest for atmospheric chemistry. They are determined by the biogeochemical cycling of nitrogen. In turn, the deposition of ammonia and nitrate is an important source of nitrogen to the terrestrial biosphere. Ideally, the emissions would be computed in a coupled atmosphere–land model tracking the chemical cycling of nitrogen in the atmospheric and terrestrial reservoirs. Simpler parameterizations are generally used in atmospheric models in which emission is computed as a function of soil nitrogen availability, temperature, and soil moisture. For example, Hudman et al. (2012) parameterize the soil emission Ei of NO for different biomes i as the product of functions describing respectively the dependences on soil nitrogen enrichment (N), temperature (T), soil moisture measured by the fraction of water-filled pore space (θ), and time since the last precipitation event (l):
Ei = f1, i(N) × f2(T) × f3(θ) × f4(l)
(9.4)
Here, N includes contributions from fertilizer input and atmospheric deposition (thus coupling emissions to deposition). The temperature function is a measure of microbial activity. The soil moisture function peaks for θ = 0.2–0.3; at lower values of θ bacterial activity is limited by water availability, while at higher values the clogging of soil pores leads to anaerobic conditions where emission of N2O and N2 dominates over emission of NO. The pulsing function f4 describes the observed surge of emissions upon precipitation after an extended dry period (dry season), when water-stressed bacteria reactivate to mobilize excess nitrogen.
Figure 9.5 shows the global annual soil emissions of NO computed by Hudman et al. (2012) from (9.4). Emissions are high in agricultural areas of northern mid-latitudes, reflecting the heavy use of fertilizer. Dry grasslands in South America and Africa also have high emissions, largely driven by the pulsing at the end of the dry season. Some of the soil emissions of NO may be oxidized to NO2 within the canopy, and this NO2 may then deposit to leaves, thus limiting export to the above-canopy atmosphere. However, the leaves may also be a source of NO2. These canopy effects are very poorly understood and often not included in models.
Figure 9.5 Annual emission of NO from soils.
From Hudman et al. (2012).
9.2.2 Open Fires
Open fire emissions include contributions from wildfires, prescribed fires, land clearing, and agricultural management. These emissions are often labeled in the literature as biomass burning, but that leaves ambiguity as to whether biofuels are included. Most fires are set by humans, although some wildfires are triggered by lightning. Even when set by humans, fires are not generally classified as “anthropogenic” in emission inventories because they may have happened anyway even without human intervention. In fact, human intervention may be to suppress wildfires.
Fires emit mostly CO2, CO, and H2O, but also many other trace species. Emissions depend on the type of vegetation, the vegetation density, and the fire intensity. Fire information is usually available from satellites and ground surveys as area burned ak over a time period Δt for a vegetation type k. From there, one can compute the emission rate Ei,k of species i from the fire as the product of the area burned per unit time dak/dt, the fuel load Γk [kg biomass m–2], the fraction of fuel combusted or burning efficiency βk, and an emission factor Fi,k [g species emitted per kg fuel burned]:
(9.5)
The burning efficiency depends on the fire intensity and on meteorological conditions, and also varies between different ecosystem components. Emission factors are determined from laboratory fire experiments or from sampling of fire plumes, generally using CO2 as the normalization factor. They can be very different between successive flaming and smoldering stages of a fire. For example, NOx emission factors are much higher in the flaming stage while CO emission factors are much higher in the smoldering stage. The different stages of a fire are generally not resolved in models because of lack of detailed temporal information. In most cases, models use mean emission factors compiled from data for different vegetation types (Table 9.2).
Table 9.2 Emission factors for open fires
Chemical species Savanna and grassland Tropical forest Extra-tropical forest Crop residue Pasture maintenance
CO2 1686 1643 1509 1585 1548
CO 63 93 122 102 135
CH4 1.9 5.1 5.68 5.82 8.71
NMVOCs 12.4 26 27 25.7 44.8
H2 1.7 3.36 2.03 2.59 –
NOx 3.9 2.55 1.12 3.11 0.75
N2O – – 0.38 – –
Organic aerosol 2.62 4.71 9.1 2.3 9.64
Black carbon 0.37 0.52 0.56 0.75 0.91
Emission factors [g kg−1] for species emitted from combustion of different types of biomass. NOx is given as NO.
From the review of Akagi et al. (2011).
Figure 9.6 shows an inventory of CO emissions from open fires in September 2000. September is the end of the dry season in the southern tropics, and fire activity is particularly intense there. Most tropical fires are from agricultural management, in particular savanna burning. There is also a contribution from land clearing. Fires at northern mid-latitudes include contributions from wildfires (as in Siberia and Canada), prescribed burning (e.g., the southeast USA), and agricultural waste burning (e.g., West Asia).
Figure 9.6 Emission of CO from open fires in September 2000.
Source: ECCAD database (Granier et al., 2011; Lamarque et al., 2010).
Plumes from large fires are buoyant due to the heat released by combustion and can thus be lofted to the free troposphere above the PBL. This lofting is important to recognize in models because it affects the subsequent transport and chemistry of the fire plumes, and allows smoke particles to reach the free troposphere without being scavenged by precipitation. The height reached by the plume is determined by the fire size and intensity, and by the thermodynamic stability of the background atmosphere. Latent heat release can increase the height reached by the plume and lead to the formation of deep convective clouds, a process called pyroconvection. A standard plume-rise formulation used in atmospheric models is that of Freitas et al. (2007). Figure 9.7 illustrates its application to a boreal fire in central Canada (large fire size and intensity) and a grassland fire in Texas (small fire size and intensity). The boreal fire plume rises to 3 km altitude while the grassland fire plume remains in the PBL.
Figure 9.7 Plume rise from a large boreal forest fire in Canada (a and b) and from a small grassland fire in Texas (c and d). The left panels show the plume rise inferred from aerosol retrievals by the MISR satellite instrument. The right panels show results from the 1-D plume rise model of Freitas et al. (2007), where the point of zero vertical velocity marks the top of the plume.
From Val Martin et al. (2012).
9.2.3 Volcanoes
Volcanoes play a fundamental role in the cycling of elements on geologic timescales by transferring material from the lithosphere to the atmosphere. On the shorter perspective of atmospheric sources and sinks, volcanoes are of most interest as sources of ash and sulfur gases (mainly SO2 and H2S). Volcanoes often release material in the free troposphere. Large volcanic eruptions inject material into the lower stratosphere, and the resulting long-lived sulfate aerosol has important implications for climate and for stratospheric ozone.
Volcanic emissions can be non-eruptive or eruptive. Non-eruptive emissions are released at the volcano mouth while eruptive emissions are injected to higher altitude. Eruptive emissions are usually brief and variable, although some volcanoes can be in continuous eruption for many years. Worldwide databases of volcanic eruptions are available with eruption dates and strengths measured by the logarithmic volcanic explosivity index (VEI). The VEI is an integer measure that ranges from 0 (non-explosive) to 8 (colossal). Volcanic emissions and injection heights are commonly assigned in models as a function of VEI or using direct observations. Satellite observations have greatly increased the ability to map volcanic SO2 emissions (Schnetzler et al., 2007; Figure 9.8).
Figure 9.8 SO2 plume from the Kasato
chi volcanic eruption in the Aleutians observed by the OMI satellite instrument on August 8, 2008.
From Wang et al. (2013).
9.2.4 Anthropogenic Emissions
Anthropogenic emissions span a wide range of processes of which combustion, industrial leaks, and agricultural activities are the most important. The “anthropogenic” label in the literature can be ambiguous and inconsistent. For example, some anthropogenic inventories include prescribed and agricultural fires while others do not. Anthropogenic inventories typically include emissions of ammonia from agricultural fertilizer, but may not include emissions of NOx from the same process. Regional inventories may include emissions from aircraft in airports but not in the air. They may include ship emissions in ports but not at sea. Because of definitional problems such as these, care is needed when using anthropogenic emission inventories. It is important to ascertain which sources are included.
Anthropogenic emissions are usually better quantified than other emissions because activity rates are available as economic data and emission factors are documented for air quality management purposes. Emission inventories commonly distinguish between area sources and point sources. Area sources include vehicles and other individually small sources for which emissions are distributed over the activity area with best estimates of emission factors. Point sources are concentrated discharges from localized sources such as power plants. These emissions are often released by smokestacks hundreds of meters above the surface, and height information may be provided in the inventory. Large point sources may have continuous emission monitoring devices installed in their stacks to comply with air quality regulations, in which case the emissions are particularly well quantified.
Anthropogenic emission inventories are produced by various groups and agencies to serve air quality management and climate modeling needs. They may cover the whole world or limited geographical domains. Regulatory models used for air quality management typically construct their own highly detailed emission inventories over regional domains separating individual sources and with temporal resolution as fine as hourly. At the other end, many inventories are available only as gridded annual totals. In such cases, temporal information on emissions (diurnal, weekday/weekend, seasonal, interannual) needs to be independently provided using scaling factors.
Figure 9.9 shows as an example a global inventory of NOx anthropogenic emissions in 2008. Emissions are mainly from fossil fuel combustion and peak in the densely populated regions of developed countries. Emissions over the oceans are from ships and aircraft.
Figure 9.9 Anthropogenic NOx emissions from the MACCity inventory at 0.5° x 0.5° resolution for the year 2008.
Courtesy: C. Granier, Centre National de la Recherche Scientifique (CNRS).
9.2.5 Mechanical Emissions: Sea Salt and Dust
Wind stress on the Earth’s surface causes mechanical emission of aerosol particles including sea salt, mineral dust, pollen, and plant debris. Sea salt and dust are dominant components of the coarse-mode (supermicron) aerosol over ocean and land, respectively, and generally make important contributions to total aerosol mass concentrations and optical depth. Pollen and plant debris have more localized influences.
Sea Salt Aerosol
Emission of sea salt particles is mostly driven by the entrainment of air into seawater by wave-breaking. The resulting air bubbles rise and burst at the sea surface, injecting particles into the air. The emission flux is a strong function of wind speed. A commonly used emission parameterization is that of Monahan et al. (1986), modified by Gong (2003) to better fit observations, and by Jaeglé et al. (2011) to include dependence on sea surface temperature (SST):
(9.6)
with
(9.7)
Here, dE/dr is the emission flux size distribution function [particles m–2 s–1 μm–1] at 80% relative humidity (RH), r is the particle radius [μm] at 80% RH (about twice the dry radius), u10 [m s–1] is the wind speed at 10 m above the surface, and TC [°C] is the SST. Figure 9.10 shows the resulting number size distribution of the emitted particles, featuring a peak at 0.1 μm consistent with observations (Gong, 2003). The dependence of emissions on SST reflects the strong sensitivity of seawater viscosity to temperature: Warmer waters are less viscous, allowing for faster rise of small bubbles and hence a larger particle source.
Figure 9.10 Emission flux size distribution function computed with the Gong (2003) parameterization for three different wind speeds (6, 9, 17 m s–1) and compared to the parameterizations of Monahan et al. (1986) and Vignati et al. (2001). The original Monahan et al. (1986) parameterization features an increase in emission with decreasing radius below 0.1 μm that is inconsistent with observations.
Adapted from Gong (2003).
Figure 9.11 shows the global mass flux of sea salt aerosol computed from (9.6) using assimilated meteorological data. Emission is highest at southern mid-latitudes where winds are strongest, though this maximum is mitigated by cold SSTs. Warm waters of the tropics have higher emission than would be computed solely from a wind speed dependence.
Figure 9.11 Annual mean mass emission flux of sea salt aerosol.
From Jaeglé et al. (2011).
Mineral Dust
Mineral dust is emitted by sandblasting of soils, a process called saltation. Wind lifts large sand particles (diameter Ds > 50 μm) that travel over only short horizontal distances before falling back to the surface by gravity. As the sand particles fall, they eject dust particles of diameter Dd small enough to be transported over long distances in the atmosphere. These fine particles are classified as clay (Dd < 2 μm) and silt (2 < Dd < 50 μm).
Experimental data show that the dust emission flux is proportional to the horizontal saltation flux from the transport of sand particles (Gillette, 1979). Over bare soils, the saltation flux Q [kg m−1 s−1] can be expressed as a function of the friction velocity u∗[m s−1] (defined in Chapter 8) and a threshold friction velocity u*t by
(9.8)
Here ρa is the air density [kg m−3], g is the acceleration of gravity [m s−2], S a preference source term, and c a constant of proportionality derived from wind-tunnel experiments and typically taken to be c = 2.61. The preference source term S accounts for accumulated erodible sediments in a given grid square due, for example, to topography or run-off areas; Ginoux et al. (2001) assume that large amounts of sediments accumulate primarily in valleys and depressions and adopt the empirical formulation
(9.9)
where z denotes the mean altitude of the model grid cell under consideration, while zmin and zmax represent the maximum and minimum elevations in the surrounding 10° × 10° area (typical size of a hydrological basin). The threshold friction velocity u∗t is a function of the sand particle diameter Ds, and represents the capacity of the soil to resist wind erosion. Its value is determined by factors such as soil texture, soil moisture, and the presence of vegetation and other roughness elements (Xi and Sokolik, 2015). In the case of dry soils, u∗t has a value of about 0.2 m s−1 for Ds = 100 μm.
The emission flux of dust particles resulting from the bombardment of saltating particles (sand grains) of size Ds is calculated by assuming that the flux of dust [kg m−2 s−1] corresponding to a particle size bin i (i = 1, I) of increment ΔDi and mean diameter Di, is given by
(9.10)
The sandblast efficiency α [m−1] can be derived from theoretical considerations (Shao, 2004) or from wind tunnel experiments. The dust emission from size bin Di is then given by integrating Ê(Di, ΔDi, Ds) between the lower and upper limits d1 and d2 of the size of the saltating particles
Here p(Ds) is the size distribution of the sand particles (often assumed to be a composite of log-normal distributions). The total emission rate of dust is obtained by summing the emission for all I bins:
Darmenova et al. (2009) review different physical parameterizations adopted in dust emission models.
The above formulation requires detailed information on soil characteristics that may not be available
. Simpler formulations are used in global models (Ginoux et al., 2001; Zender et al., 2003). Ginoux et al. (2001, 2012) compute the dust emission flux as:
(9.11)
Here, u10 denotes the 10 m wind speed, u10,t is a threshold, fA is the fractional area of land suitable for saltation, and S is an adjustable global scaling factor to match dust observations. Dust emission in this formulation has a cubic dependence on wind speed, and is therefore controlled by gusty conditions that are poorly resolved in atmospheric models. The scaling factor S is intended to correct for this effect and varies with the model grid resolution. Global models typically choose S to yield a global dust emission of about 1500 Tg a–1 as this is found to provide a good fit to observations.
Figure 9.12 shows the global distribution of natural and anthropogenic dust emissions estimated by Ginoux et al. (2012). Natural emission is dominated globally by the Sahara and also has substantial contributions from the Middle East, the Gobi desert, and the North American West. There are large anthropogenic dust emissions from dry and eroded agricultural areas.
Modeling of Atmospheric Chemistry Page 50