Investigation of mercury emissions from burning of Australian eucalypt forest surface fuels using a combustion wind tunnel and field observations

Highlights

• Wildfire mercury release from eucalypt forests is currently overestimated by 60%.

• Under natural wildfire fuel moistures, 99% of release is elemental mercury.

• Fire progression impacts heavily; most mercury is released during flaming stage.

• Recommended emission ratio for GEM/CO is 0.58 ± 0.01 × 10⁻⁷.

Abstract

Environmental cycling of the toxic metal mercury (Hg) is ubiquitous, and still not completely understood. Volatilisation and emission of mercury from vegetation, litter and soil during burning represents a significant return pathway for previously-deposited atmospheric mercury. Rates of such emission vary widely across ecosystems as they are dependent on species-specific uptake of atmospheric mercury as well as fire return frequencies. Wildfire burning in Australia is currently thought to contribute between 1 and 5% of the global total of mercury emissions, yet no modelling efforts to date have utilised local mercury emission factors (mass of emitted mercury per mass of dry fuel) or local mercury emission ratios (ratio of emitted mercury to another emitted species, typically carbon monoxide). Here we present laboratory and field investigations into mercury emission from burning of surface fuels in dry sclerophyll forests, native to the temperate south-eastern region of Australia. From laboratory data we found that fire behaviour — in particular combustion phase — has a large influence on mercury emission and hence emission ratios. Further, emission of mercury was predominantly in gaseous form with particulate-bound mercury representing <1% of total mercury emission. Importantly, emission factors and emission ratios with respect to carbon monoxide and carbon dioxide, from both laboratory and field data all show that gaseous mercury emission from biomass burning in Australian dry sclerophyll forests is currently overestimated by around 60%. Based on these results, we recommend a mercury emission factor of 28.7 ± 8.1 μg Hg kg⁻¹ dry fuel, and emission ratio of gaseous elemental mercury relative to carbon monoxide of 0.58 ± 0.01 × 10⁻⁷, for estimation of mercury release from the combustion of Australian dry sclerophyll litter.

Introduction

The global nature of mercury (Hg) pollution has long been recognised (Driscoll et al., 2013). With natural sinks, sources and cycles, the unique physicochemical properties of this toxic metal allow for constant transfer between biological, terrestrial, aquatic and atmospheric reservoirs, making it ubiquitous throughout the environment (Fitzgerald et al., 1998; Selin, 2009; Mason et al., 2012; Krabbenhoft and Sunderland, 2013). Increases in mercury emission sources due to human activities have perturbed this natural cycle in a manner that has become a threat to human and ecosystem health (Streets et al., 2011, 2017; Amos et al., 2013). This threat is globally recognised in the Minamata Convention on Mercury (Kessler, 2013), aimed at reducing anthropogenic emissions of mercury to the environment. Article 19 of the convention addresses the need to understand mercury's complex natural cycling by calling for parties to the convention to, where possible, increase research and extend current monitoring efforts.

In the atmosphere, mercury exists largely in the form of gaseous elemental mercury (GEM), with the operationally-defined gaseous oxidised mercury (GOM) and particulate bound mercury (PBM) forms generally thought to comprise less than 10% of tropospheric atmospheric concentrations (Pirrone et al., 2010). The long atmospheric lifetime of GEM [estimated at between 5 and 12 months (Holmes et al., 2010; Horowitz et al., 2017; Lindberg et al., 2007)] means transport of mercury can take place through the atmosphere — but also in watercourses and the ocean — to regions far-removed from their sources. From the atmosphere, mercury is deposited to terrestrial surfaces and waterways, taken up by vegetation, and re-emitted in a complex natural cycle that is still not completely understood (Smith-Downey et al., 2010; Amos et al., 2013; Agnan et al., 2016). Atmospheric mercury may be taken into vegetation during photosynthesis (Rea et al., 2001) or deposited via dry or wet deposition processes onto vegetated surfaces, whereby it can be incorporated into the cell membrane through foliar uptake (Mason et al., 1995; Stamenkovic and Gustin, 2009; Hintelmann et al., 2002). Throughfall, litterfall and surface dry/wet deposition processes deliver atmospheric mercury to the underlying surface litter, whereby leaf decomposition and further atmospheric deposition enhance soil mercury levels due to binding of mercury to organic matter within the soil (Wright et al., 2016; Hartman et al., 2009). Vegetation type, coverage and growth rates, and atmospheric mercury concentrations all affect the rate at which mercury is stored within these components (Johnson and Lindberg, 1995; Ericksen and Gustin, 2004; Cobbett and Van Heyst, 2007).

Biomass burning releases mercury from these stores back into the atmosphere through volatilisation of mercury within biomass during combustion and through thermal desorption of mercury bound within the soil matrix (Melendez-Perez et al., 2014). Herein we limit the definition of biomass burning to free-burning vegetation fires (both intentionally and accidentally ignited) and exclude burning of biomass for industrial/cultural purposes (e.g. wood burners and stoves). The release of mercury from biomass burning is an important yet complex and poorly understood component of the global mercury cycle as it can lead to redistribution of mercury to sensitive ecosystems where methylation may occur, or it can result in direct human exposure to mercury through inhalation of biomass burning plumes (Simone et al., 2015). Mercury is often completely volatilised from combusted biomass (Friedli et al., 2003) and this emitted mercury is largely in the form of GEM (Friedli et al., 2001, 2003), however Obrist et al. (2007) showed that increasing PBM levels are associated with increasing fuel moisture and decreasing fire intensity. The lower atmospheric lifetime of PBM leads to changing mercury deposition patterns in response to emission partitioning (Simone et al., 2016). The extent to which thermal desorption takes place in the soil is related to the intensity of the fire, as low intensity, slow-moving fires may heat the soil to higher temperatures than faster moving, higher-intensity fires (Webster et al., 2016). As such, release of mercury is not only dependent on the loading of mercury within the fuels but also on fire behaviour.

Ecosystem-scale estimates of mercury release from biomass burning are typically achieved by applying a mixture of empirical and remotely-sensed data. Burned areas are often derived from satellite data products, from which emission of mercury (or other chemical species) can be estimated by applying an empirically-derived emission factor (ratio of emitted species per mass of dry fuel), or an emission ratio with reference to another chemical species along with its emission factor. These approaches are outlined generally below (Aalde et al., 2006):

$$E_x=A\cdot L\cdot BE\cdot E{F}_x$$

$$E_x=A\cdot L\cdot BE\cdot E{F}_y\cdot E{R}_{x/y}$$

where $E_x$ is the emitted mass of species x, A the area burned, L the fuel loading in mass per area, BE the burning efficiency and $E{F}_x$ the emission factor for species x. Where emission factors for species x are poorly constrained or not known, these can be determined by using an emission ratio $E{R}_{x/y}$ and applying this to the known emission factor for species y. Emission ratios for mercury are generally derived from ground- or aerial-based measurements of smoke plumes and are typically reported with respect to carbon monoxide (CO), although ratios with carbon dioxide (CO₂) have been presented by Brunke et al. (2001). The use of emission ratios is advantageous as, due to turbulent mixing in the plume, it provides an average enhancement across the horizontal and vertical extent of the fire. Emission factors are instead based on fuel mercury concentrations and empirically-derived estimates of release during combustion. These provide the most direct estimate of mercury release from specific vegetation types and from soils when the amount of biomass burned is known, yet require significant sampling to obtain data suitable at an ecosystem scale (Andreae and Merlet, 2001).

Australia is a particularly fire-prone continent, and global-scale models of mercury emission from biomass burning estimate that emissions over Australia represent between 1 and 5% of the global total (Friedli et al., 2009; Simone et al., 2015). Based on the National Oceanic and Atmospheric Administration's Advanced Very High Resolution Radiometer (NOAA-AVHHR) satellite data, an average of 41 million ha (5% total land mass) burned annually in the years 1997–2011 [Fig. 1 (Maier, 2016),]. Home to 32 major vegetation groups (Australian Government Department of the Environment and Energy (DEE), 2016) and spanning a broad range of climates, Australia's ecosystems are subject to varying degrees of fire frequency and intensity. Tropical savannah in northern Australia may undergo burning every 1–2 years (Meyer et al., 2012), whilst temperate forests in south-eastern Australia typically experience burning every 15 + years (Gill et al., 1981), which can result in greater uptake of mercury over a longer growing period. Simplification of vegetation types across continental scales is necessary in modelling biomass burning mercury emissions, a global example of which is the terrestrial ecoregion [ (Olson and Dinerstein, 2002), (Bailey, 1995), see Fig. 1]. To date, the most extensive investigation into vegetation mercury content across Australia was performed by Packham et al. (2009) (see Table 1), yet these have not been used in any subsequent mercury modelling efforts. Modelling estimates of mercury emissions from biomass burning in Australia have instead so far only been obtained using empirically-derived emission factors or emission ratios from studies undertaken in the Northern Hemisphere (Friedli et al., 2003, 2009). The resulting estimates of annual release over Australia are currently poorly constrained, spanning a range between 7 Mg Hg a⁻¹ and 129 Mg Hg a⁻¹ (Friedli et al., 2009; Nelson et al., 2009, 2012; Packham et al., 2009; Cope et al., 2009; Simone et al., 2015).

In response to the general lack of knowledge surrounding mercury in Australian vegetation — and the complete lack of Australian-derived emission ratios or emission factors in Australian mercury emission modelling — this paper reports on and compares two different studies of mercury release from the burning of Australian native forest surface fuels. In this paper we limit our analysis to the eucalypt-dominated dry sclerophyll forests, the most widespread forest type in south-eastern Australia (Montreal Process Implementation Group for Australia and National Forest Inventory Steering Committee, 2013). Native to Australia, eucalypts have been cultivated globally and can now be found on all inhabited continents. Experimental burns of dry sclerophyll surface fuels took place in a combustion wind tunnel designed for the study of combustion of vegetation fuels (Sullivan et al., 2013), which provided the unique advantage that the influence of fire propagation on emissions could be investigated. Results from this laboratory-scale study are then compared with observations of biomass burning plumes from the Cape Grim Baseline Air Pollution Station (CGBAPS) in Tasmania. These data add to the growing knowledge surrounding mercury in vegetation and its release during biomass burning, and will contribute to constraining uncertainty regarding natural mercury cycling over the Australian continent and the Southern Hemisphere in general.