Search for: All records

Mercury (Hg), a neurotoxic heavy metal, is transferred to marine and terrestrial ecosystems through atmospheric transport. Recent studies have highlighted the role of vegetation uptake as a sink for atmospheric elemental mercury (Hg0) and a source of Hg to soils. However, the global magnitude of the Hg0 vegetation uptake flux is highly uncertain, with estimates ranging 1000–4000 Mg per year. To constrain this sink, we compare simulations in the chemical transport model GEOS-Chem with a compiled database of litterfall, throughfall, and flux tower measurements from 93 forested sites. The prior version of GEOS-Chem predicts median Hg0 dry deposition velocities similar to litterfall measurements from Northern hemisphere temperate and boreal forests (~0.03 cm s-1 yet it underestimates measurements from a flux tower study (0.04 cm s-1 vs. 0.07 cm s-1and Amazon litterfall (0.05 cm s-1 vs. 0.17 cm s-1). After revising the Hg0 reactivity within the dry deposition parametrization to match flux tower and Amazon measurements, GEOS-Chem displays improved agreement with the seasonality of atmospheric Hg0 observations in the Northern midlatitudes. Additionally, the modelled bias in Hg0 concentrations in South America decreases from +0.21 ng m-3 +0.05 ng m-3. We calculate a global flux of Hg0 dry deposition to land of 2276 Mg per year, approximately double previous model estimates. The Amazon rainforest contributes 29% of the total Hg0 land sink, yet continued deforestation and climate change threatens the rainforest's stability and thus its role as an important Hg sink. In an illustrative worst-case scenario where the Amazon is completely converted to savannah, GEOS-Chem predicts that an additional 283 Mg Hg per year would deposit to the ocean, where it can bioaccumulate in the marine food chain. Biosphere–atmosphere interactions thus play a crucial role in global Hg cycling and should be considered in assessments of future Hg pollution. 

As mathematical computing becomes more democratized in high-level languages, high-performance symbolic-numeric systems are necessary for domain scientists and engineers to get the best performance out of their machine without deep knowledge of code optimization. Naturally, users need different term types either to have different algebraic properties for them, or to use efficient data structures. To this end, we developed Symbolics.jl, an extendable symbolic system which uses dynamic multiple dispatch to change behavior depending on the domain needs. In this work we detail an underlying abstract term interface which allows for speed without sacrificing generality. We show that by formalizing a generic API on actions independent of implementation, we can retroactively add optimized data structures to our system without changing the pre-existing term rewriters. We showcase how this can be used to optimize term construction and give a 113x acceleration on general symbolic transformations. Further, we show that such a generic API allows for complementary term-rewriting implementations. Exploiting this feature, we demonstrate the ability to swap between classical term-rewriting simplifiers and e-graph-based term-rewriting simplifiers. We illustrate how this symbolic system improves numerical computing tasks by showcasing an e-graph ruleset which minimizes the number of CPU cycles during expression evaluation, and demonstrate how it simplifies a real-world reaction-network simulation to halve the runtime. Additionally, we show a reaction-diffusion partial differential equation solver which is able to be automatically converted into symbolic expressions via multiple dispatch tracing, which is subsequently accelerated and parallelized to give a 157x simulation speedup. Together, this presents Symbolics.jl as a next-generation symbolic-numeric computing environment geared towards modeling and simulation. 

Modern design, control, and optimization often require multiple expensive simulations of highly nonlinear stiff models. These costs can be amortized by training a cheap surrogate of the full model, which can then be used repeatedly. Here we present a general data-driven method, the continuous time echo state network (CTESN), for generating surrogates of nonlinear ordinary differential equations with dynamics at widely separated timescales. We empirically demonstrate the ability to accelerate a physically motivated scalable model of a heating system by 98x while maintaining relative error of within 0.2 %. We showcase the ability for this surrogate to accurately handle highly stiff systems which have been shown to cause training failures with common surrogate methods such as Physics-Informed Neural Networks (PINNs), Long Short Term Memory (LSTM) networks, and discrete echo state networks (ESN). We show that our model captures fast transients as well as slow dynamics, while demonstrating that fixed time step machine learning techniques are unable to adequately capture the multi-rate behavior. Together this provides compelling evidence for the ability of CTESN surrogates to predict and accelerate highly stiff dynamical systems which are unable to be directly handled by previous scientific machine learning techniques. 

Abstract Ambient fine particulate matter (PM 2.5 ) is the world’s leading environmental health risk factor. Reducing the PM 2.5 disease burden requires specific strategies that target dominant sources across multiple spatial scales. We provide a contemporary and comprehensive evaluation of sector- and fuel-specific contributions to this disease burden across 21 regions, 204 countries, and 200 sub-national areas by integrating 24 global atmospheric chemistry-transport model sensitivity simulations, high-resolution satellite-derived PM 2.5 exposure estimates, and disease-specific concentration response relationships. Globally, 1.05 (95% Confidence Interval: 0.74–1.36) million deaths were avoidable in 2017 by eliminating fossil-fuel combustion (27.3% of the total PM 2.5 burden), with coal contributing to over half. Other dominant global sources included residential (0.74 [0.52–0.95] million deaths; 19.2%), industrial (0.45 [0.32–0.58] million deaths; 11.7%), and energy (0.39 [0.28–0.51] million deaths; 10.2%) sectors. Our results show that regions with large anthropogenic contributions generally had the highest attributable deaths, suggesting substantial health benefits from replacing traditional energy sources. 

The conservation field is experiencing a rapid increase in the amount, variety, and quality of spatial data that can help us understand species movement and landscape connectivity patterns. As interest grows in more dynamic representations of movement potential, modelers are often limited by the capacity of their analytic tools to handle these datasets. Technology developments in software and high-performance computing are rapidly emerging in many fields, but uptake within conservation may lag, as our tools or our choice of computing language can constrain our ability to keep pace. We recently updated Circuitscape, a widely used connectivity analysis tool developed by Brad McRae and Viral Shah, by implementing it in Julia, a high-performance computing language. In this initial re-code (Circuitscape 5.0) and later updates, we improved computational efficiency and parallelism, achieving major speed improvements, and enabling assessments across larger extents or with higher resolution data. Here, we reflect on the benefits to conservation of strengthening collaborations with computer scientists, and extract examples from a collection of 572 Circuitscape applications to illustrate how through a decade of repeated investment in the software, applications have been many, varied, and increasingly dynamic. Beyond empowering continued innovations in dynamic connectivity, we expect that faster run times will play an important role in facilitating co-production of connectivity assessments with stakeholders, increasing the likelihood that connectivity science will be incorporated in land use decisions. 

Abstract. We present an updated mechanism for tropospheric halogen (Cl + Br + I) chemistry in the GEOS-Chem global atmospheric chemical transportmodel and apply it to investigate halogen radical cycling and implications for tropospheric oxidants. Improved representation of HOBr heterogeneouschemistry and its pH dependence in our simulation leads to less efficient recycling and mobilization of bromine radicals and enables the model toinclude mechanistic sea salt aerosol debromination without generating excessive BrO. The resulting global mean tropospheric BrO mixingratio is 0.19 ppt (parts per trillion), lower than previous versions of GEOS-Chem. Model BrO shows variable consistency and biases in comparison tosurface and aircraft observations in marine air, which are often near or below the detection limit. The model underestimates the daytimemeasurements of Cl2 and BrCl from the ATom aircraft campaign over the Pacific and Atlantic, which if correct would imply a very largemissing primary source of chlorine radicals. Model IO is highest in the marine boundary layer and uniform in the free troposphere, with a globalmean tropospheric mixing ratio of 0.08 ppt, and shows consistency with surface and aircraft observations. The modeled global meantropospheric concentration of Cl atoms is 630 cm−3, contributing 0.8 % of the global oxidation of methane, 14 % of ethane,8 % of propane, and 7 % of higher alkanes. Halogen chemistry decreases the global tropospheric burden of ozone by 11 %,NOx by 6 %, and OH by 4 %. Most of the ozone decrease is driven by iodine-catalyzed loss. The resulting GEOS-Chem ozonesimulation is unbiased in the Southern Hemisphere but too low in the Northern Hemisphere. 

Abstract. Air quality models have not been able to reproduce the magnitude of theobserved concentrations of fine particulate matter (PM2.5) duringwintertime Chinese haze events. The discrepancy has been at least partlyattributed to low biases in modeled sulfate production rates, due to the lackof heterogeneous sulfate production on aerosolsin the models. In this study, we explicitly implement four heterogeneous sulfate formationmechanisms into a regional chemical transport model, in addition togas-phase and in-cloud sulfate production. We compare the model results withobservations of sulfate concentrations and oxygen isotopes, Δ17O(SO42-), in the winter of 2014–2015, the latter of whichis highly sensitive to the relative importance of different sulfateproduction mechanisms. Model results suggest that heterogeneous sulfateproduction on aerosols accounts for about 20 % of sulfate production inclean and polluted conditions, partially reducing the modeled low bias insulfate concentrations. Model sensitivity studies in comparison with theΔ17O(SO42-) observations suggest that heterogeneoussulfate formation is dominated by transition metal ion-catalyzed oxidation of SO2. 

Formaldehyde (HCHO) is generated from direct urban emission sources and secondary production from the photochemical reactions of urban smog. HCHO is linked to tropospheric ozone formation, and contributes to the photochemical reactions of other components of urban smog. In this study, pollution plume intercepts during the Wintertime INvestigation of Transport, Emissions, and Reactivity (WINTER) campaign were used to investigate and characterize the formation of HCHO in relation to several anthropogenic tracers. Analysis of aircraft intercepts combined with detailed chemical box modeling downwind of several cities suggests that the most important contribution to observed HCHO was primary emission. A box model analysis of a single plume suggested that secondary sources contribute to 21 ± 10% of the observed HCHO. Ratios of HCHO/CO observed in the northeast US, from Ohio to New York, ranging from 0.2% to 0.6%, are consistent with direct emissions combined with at most modest photochemical production. Analysis of the nocturnal boundary layer and residual layer from repeated vertical profiling over urban influenced areas indicate a direct HCHO emission flux of 1.3 × 10¹⁴molecules cm⁻²h⁻¹. In a case study in Atlanta, GA, nighttime HCHO exhibited a ratio to CO (0.6%–1.8%) and was anti‐correlated with O₃. Observations were consistent with mixing between direct HCHO emissions in urban air masses with those influenced by more rapid HCHO photochemical production. The HCHO/CO emissions ratios determined from the measured data are 2.3–15 times greater than the NEI 2017 emissions database. The largest observed HCHO/CO was 1.7%–1.8%, located near co‐generating power stations.