Mapping the drivers of uncertainty in atmospheric selenium deposition with global sensitivity analysis

Abstract

Abstract. An estimated 0.5–1 billion people globally have inadequate intakes of selenium (Se), due to a lack of bioavailable Se in agricultural soils. Deposition from the atmosphere, especially through precipitation, is an important source of Se to soils. However, very little is known about the atmospheric cycling of Se. It has therefore been difficult to predict how far Se travels in the atmosphere and where it deposits. To answer these questions, we have built the first global atmospheric Se model by implementing Se chemistry in an aerosol–chemistry–climate model, SOCOL-AER (modeling tools for studies of SOlar Climate Ozone Links – aerosol). In the model, we include information from the literature about the emissions, speciation, and chemical transformation of atmospheric Se. Natural processes and anthropogenic activities emit volatile Se compounds, which oxidize quickly and partition to the particulate phase. Our model tracks the transport and deposition of Se in seven gas-phase species and 41 aerosol tracers. However, there are large uncertainties associated with many of the model's input parameters. In order to identify which model uncertainties are the most important for understanding the atmospheric Se cycle, we conducted a global sensitivity analysis with 34 input parameters related to Se chemistry, Se emissions, and the interaction of Se with aerosols. In the first bottom-up estimate of its kind, we have calculated a median global atmospheric lifetime of 4.4 d (days), ranging from 2.9 to 6.4 d (2nd–98th percentile range) given the uncertainties of the input parameters. The uncertainty in the Se lifetime is mainly driven by the uncertainty in the carbonyl selenide (OCSe) oxidation rate and the lack of tropospheric aerosol species other than sulfate aerosols in SOCOL-AER. In contrast to uncertainties in Se lifetime, the uncertainty in deposition flux maps are governed by Se emission factors, with all four Se sources (volcanic, marine biosphere, terrestrial biosphere, and anthropogenic emissions) contributing equally to the uncertainty in deposition over agricultural areas. We evaluated the simulated Se wet deposition fluxes from SOCOL-AER with a compiled database of rainwater Se measurements, since wet deposition contributes around 80 % of total Se deposition. Despite difficulties in comparing a global, coarse-resolution model with local measurements from a range of time periods, past Se wet deposition measurements are within the range of the model's 2nd–98th percentiles at 79 % of background sites. This agreement validates the application of the SOCOL-AER model to identifying regions which are at risk of low atmospheric Se inputs. In order to constrain the uncertainty in Se deposition fluxes over agricultural soils, we should prioritize field campaigns measuring Se emissions, rather than laboratory measurements of Se rate constants.

Sulfur Amino Acid Status Controls Selenium Methylation in Pseudomonas tolaasii: Identification of a Novel Metabolite from Promiscuous Enzyme Reactions

“…A number of studies have measured volatile Se in marine environments (reference 44 and references therein), whereas measurements above terrestrial environments (e.g., wetlands, forests, and soils) are scarcer ( 12 , 45 , 46 , 47 ). Large regional- to continental-scale data sets of total Se concentrations in soil as well as environmental variables (e.g., climate and soil physicochemical properties) have been used to identify drivers of Se concentrations in soil in top-down approaches (references 43 and 48 and references therein). Bottom-up and top-down approaches have shown considerable differences (a factor of ∼2 in Se fluxes) in global atmospheric emissions ( 43 ).…”

“…Bottom-up and top-down approaches have shown considerable differences (a factor of ∼2 in Se fluxes) in global atmospheric emissions ( 43 ). Uncertainty in Se deposition maps in top-down approaches mainly arose from uncertainties in global emission fluxes because of the high variability of past flux measurements ( 48 ). Furthermore, the spatial distribution of terrestrial Se emissions is largely unknown ( 43 ).…”

“…Here, DMDSe and DMSeS were the only species found. The first global Se deposition model, however, considered only DMSe, as it is usually the dominant emitted volatile Se compound ( 48 ). Identifying appropriate proxies of the S status of soils (e.g., total S, S oxyanions, and thiols) that correlate with Se volatilization to certain species (DMSe, DMDSe, and DMSeS) may increase the accuracy of global-scale distribution modeling in terms of the underpredictions still found ( 54 ).…”

Atmosphere–ocean–aerosol–chemistry–climate model SOCOLv4.0: description and evaluation

“…Surface albedo is parameterized according to Brovkin et al (2013). The radiative properties of tropospheric aerosols are now represented by the Max Planck Institute Aerosol Climatology version 2 simple plume implementation (MACv2-SP) parameterization (Stevens et al, 2017;Fiedler et al, 2017). MACv2-SP is formulated in terms of nine spatial plumes associated with different major anthropogenic source regions.…”

Atmosphere-Ocean-Aerosol-Chemistry-Climate Model SOCOLv4.0: description and evaluation

“…The base version of Stenke et al (2013), however, further branched into two significant sub-versions SOCOL-MPIOM with interactive ocean (Muthers et al, 2014) and SOCOL-AER with interactive aerosol microphysics (Sheng et al, 2015), each of them receiving further, independent, upgrades (Arsenovic et al, 2018;Feinberg et al, 2019), and several smaller variants, such as an improved tropospheric ozone budget (Revell et al, 2015, Revell et al, 2018, detailed methane sources and sinks (Feinberg et al, 2018), and atmospheric selenium cycling (Feinberg et al, 2020).…”