Search

Abstract Volcanic eruptions can emit large quantities of sulphur dioxide (SO2) into the atmosphere, which can be harmful to people and the environment. Aircraft encounters with a volcanic SO2 cloud could represent a health hazard to crew and passengers onboard. In this study we have assessed concentration levels of volcanic SO2 in the atmosphere following eight historic eruptions and use four-dimensional dispersion model simulation data to calculate when and where the World Health Organisation (WHO) health protection guideline for SO2 of 500 μgm-3 over 10 minutes is exceeded. The time and area of exceedance varies and depends on the eruption characteristics: the amount, duration and height of the SO2 release. The WHO-based guideline value is exceeded for all historic eruptions considered. In several cases, the area delineated by the WHO-based guideline, here called the SO2 hazard area, can be considerably larger than the volcanic ash hazard area for the same eruption. SO2 hazard areas also often extend over a longer period of time compared to the equivalent ash advisories. For example, following the 2019 eruption of Raikoke, the SO2 hazard area reached up to 1.7 million km2 and the WHO-based guideline value was exceeded for about two weeks, while volcanic ash was considered hazardous to aviation for about five days. These results will help the aviation industry to better understand the potential risks to their passengers and crew from volcanic SO2, and aid in defining concentration thresholds for any potential volcanic SO2 forecasts for aviation.

Volcanic eruptions can cause significant disruption to society, and numerical models are crucial for forecasting the dispersion of erupted material. Here we assess the skill and limitations of the Met Office's Numerical Atmospheric-dispersion Modelling Environment (NAME) in simulating the dispersion of the sulfur dioxide (SO2) cloud from the 21–22 June 2019 eruption of the Raikoke volcano (48.3∘ N, 153.2∘ E). The eruption emitted around 1.5±0.2 Tg of SO2, which represents the largest volcanic emission of SO2 into the stratosphere since the 2011 Nabro eruption. We simulate the temporal evolution of the volcanic SO2 cloud across the Northern Hemisphere (NH) and compare our model simulations to high-resolution SO2 measurements from the TROPOspheric Monitoring Instrument (TROPOMI) and the Infrared Atmospheric Sounding Interferometer (IASI) satellite SO2 products. We show that NAME accurately simulates the observed location and horizontal extent of the SO2 cloud during the first 2–3 weeks after the eruption but is unable, in its standard configuration, to capture the extent and precise location of the highest magnitude vertical column density (VCD) regions within the observed volcanic cloud. Using the structure–amplitude–location (SAL) score and the fractional skill score (FSS) as metrics for model skill, NAME shows skill in simulating the horizontal extent of the cloud for 12–17 d after the eruption where VCDs of SO2 (in Dobson units, DU) are above 1 DU. For SO2 VCDs above 20 DU, which are predominantly observed as small-scale features within the SO2 cloud, the model shows skill on the order of 2–4 d only. The lower skill for these high-SO2-VCD regions is partly explained by the model-simulated SO2 cloud in NAME being too diffuse compared to TROPOMI retrievals. Reducing the standard horizontal diffusion parameters used in NAME by a factor of 4 results in a slightly increased model skill during the first 5 d of the simulation, but on longer timescales the simulated SO2 cloud remains too diffuse when compared to TROPOMI measurements. The skill of NAME to simulate high SO2 VCDs and the temporal evolution of the NH-mean SO2 mass burden is dominated by the fraction of SO2 mass emitted into the lower stratosphere, which is uncertain for the 2019 Raikoke eruption. When emitting 0.9–1.1 Tg of SO2 into the lower stratosphere (11–18 km) and 0.4–0.7 Tg into the upper troposphere (8–11 km), the NAME simulations show a similar peak in SO2 mass burden to that derived from TROPOMI (1.4–1.6 Tg of SO2) with an average SO2 e-folding time of 14–15 d in the NH. Our work illustrates how the synergy between high-resolution satellite retrievals and dispersion models can identify potential limitations of dispersion models like NAME, which will ultimately help to improve dispersion modelling efforts of volcanic SO2 clouds.

The Lagrangian particle dispersion model FLEXPART in its original version in the mid-1990s was designed for calculating the long-range and mesoscale dispersion of hazardous substances from point sources, such as those released after an accident in a nuclear power plant. Over the past decades, the model has evolved into a comprehensive tool for multi-scale atmospheric transport modeling and analysis and has attracted a global user community. Its application fields have been extended to a large range of atmospheric gases and aerosols, e.g., greenhouse gases, short-lived climate forcers like black carbon and volcanic ash, and it has also been used to study the atmospheric branch of the water cycle. Given suitable meteorological input data, it can be used for scales from dozens of meters to global. In particular, inverse modeling based on source–receptor relationships from FLEXPART has become widely used. In this paper, we present FLEXPART version 10.4, which works with meteorological input data from the European Centre for Medium-Range Weather Forecasts (ECMWF) Integrated Forecast System (IFS) and data from the United States National Centers of Environmental Prediction (NCEP) Global Forecast System (GFS). Since the last publication of a detailed FLEXPART description (version 6.2), the model has been improved in different aspects such as performance, physicochemical parameterizations, input/output formats, and available preprocessing and post-processing software. The model code has also been parallelized using the Message Passing Interface (MPI). We demonstrate that the model scales well up to using 256 processors, with a parallel efficiency greater than 75 % for up to 64 processes on multiple nodes in runs with very large numbers of particles. The deviation from 100 % efficiency is almost entirely due to the remaining nonparallelized parts of the code, suggesting large potential for further speedup. A new turbulence scheme for the convective boundary layer has been developed that considers the skewness in the vertical velocity distribution (updrafts and downdrafts) and vertical gradients in air density. FLEXPART is the only model available considering both effects, making it highly accurate for small-scale applications, e.g., to quantify dispersion in the vicinity of a point source. The wet deposition scheme for aerosols has been completely rewritten and a new, more detailed gravitational settling parameterization for aerosols has also been implemented. FLEXPART has had the option of running backward in time from atmospheric concentrations at receptor locations for many years, but this has now been extended to also work for deposition values and may become useful, for instance, for the interpretation of ice core measurements. To our knowledge, to date FLEXPART is the only model with that capability. Furthermore, the temporal variation and temperature dependence of chemical reactions with the OH radical have been included, allowing for more accurate simulations for species with intermediate lifetimes against the reaction with OH, such as ethane. Finally, user settings can now be specified in a more flexible namelist format, and output files can be produced in NetCDF format instead of FLEXPART's customary binary format. In this paper, we describe these new developments. Moreover, we present some tools for the preparation of the meteorological input data and for processing FLEXPART output data, and we briefly report on alternative FLEXPART versions.

Significant improvements in the way we can observe and model volcanic ash clouds have been obtained since the 2010 Eyjafjallajökull eruption. One major development has been the application of data assimilation techniques, which combine models and satellite observations such that an optimal understanding of ash clouds can be gained. Still, questions remain regarding the degree to which the forecasting capabilities are improved by inclusion of such techniques and how these improvements depend on the data input. This study explores how different satellite data and different uncertainty assumptions of the satellite and a priori emissions affect the calculated volcanic ash emission estimate, which is computed by an inversion method that couples the satellite retrievals and a priori emissions with dispersion model data. Two major ash episodes over 4 days in April and May of the 2010 Eyjafjallajökull eruption are studied. Specifically, inversion calculations are done for four different satellite data sets with different size distribution assumptions in the retrieval. A reference satellite data set is chosen, and the range between the minimum and maximum 4-day average load of hourly retrieved ash is 121 % in April and 148 % in May, compared to the reference. The corresponding a posteriori maximum and minimum emission sum found for these four satellite retrievals is 26 and 47 % of the a posteriori reference estimate for the same two periods, respectively. Varying the assumptions made in the satellite retrieval is seen to affect the a posteriori emissions and modelled ash column loads, and modelled column loads therefore have uncertainties connected to them depending on the uncertainty in the satellite retrieval. By further exploring our uncertainty estimates connected to a priori emissions and the mass load uncertainties in the satellite data, the uncertainty in the a priori estimate is found in this case to have an order-of-magnitude-greater impact on the a posteriori solution than the mass load uncertainties in the satellite. Part of this is explained by a too-high a priori estimate used in this study that is reduced by around half in the a posteriori reference estimate. Setting large uncertainties connected to both a priori and satellite mass load shows that they compensate each other, but the a priori uncertainty is found to be most sensitive. Because of this, an inversion-based emission estimate in a forecasting setting needs well-tested and well-considered assumptions on uncertainties for the a priori emission and satellite data. The quality of using the inversion in a forecasting environment is tested by adding gradually, with time, more observations to improve the estimated height versus time evolution of Eyjafjallajökull ash emissions. We show that the initially too-high a priori emissions are reduced effectively when using just 12 h of satellite observations. More satellite observations (> 12 h), in the Eyjafjallajökull case, place the volcanic injection at higher altitudes. Adding additional satellite observations (> 36 h) changes the a posteriori emissions to only a small extent for May and minimal for the April period, because the ash is dispersed and transported effectively out of the domain after 1–2 days. A best-guess emission estimate for the forecasting period was constructed by averaging the last 12 h of the a posteriori emission. Using this emission for a forecast simulation leads to better performance, especially compared to model simulations with no further emissions over the forecast period in the case of a continued volcanic eruption activity. Because of undetected ash in the satellite retrieval and diffusion in the model, the forecast simulations generally contain more ash than the observed fields, and the model ash is more spread out. Overall, using the a posteriori emissions in our model reduces the uncertainties in the ash plume forecast, because it corrects effectively for false-positive satellite retrievals, temporary gaps in observations, and false a priori emissions in the window of observation.

A new, more physically based wet removal scheme for aerosols has been implemented in the Lagrangian particle dispersion model FLEXPART. It uses three-dimensional cloud water fields from the European Centre for Medium-Range Weather Forecasts (ECMWF) to determine cloud extent and distinguishes between in-cloud and below-cloud scavenging. The new in-cloud nucleation scavenging depends on cloud water phase (liquid, ice or mixed-phase), based on the aerosol's prescribed efficiency to serve as ice crystal nuclei and liquid water nuclei, respectively. The impaction scavenging scheme now parameterizes below-cloud removal as a function of aerosol particle size and precipitation type (snow or rain) and intensity.Sensitivity tests with the new scavenging scheme and comparisons with observational data were conducted for three distinct types of primary aerosols, which pose different challenges for modeling wet scavenging due to their differences in solubility, volatility and size distribution: (1) 137Cs released during the Fukushima nuclear accident attached mainly to highly soluble sulphate aerosol particles, (2) black carbon (BC) aerosol particles, and (3) mineral dust. Calculated e-folding lifetimes of accumulation mode aerosols for these three aerosol types were 11.7, 16.0, and 31.6 days respectively, when well mixed in the atmosphere. These are longer lifetimes than those obtained by the previous removal schem, and, for mineral dust in particular, primarily result from very slow in-cloud removal, which globally is the primary removal mechanism for these accumulation mode particles.Calculated e-folding lifetimes in FLEXPART also have a strong size dependence, with the longest lifetimes found for the accumulation-mode aerosols. For example, for dust particles emitted at the surface the lifetimes were 13.8 days for particles with 1 µm diameter and a few hours for 10 µm particles. A strong size dependence in below-cloud scavenging, combined with increased dry removal, is the primary reason for the shorter lifetimes of the larger particles. The most frequent removal is in-cloud scavenging (85 % of all scavenging events) but it occurs primarily in the free troposphere, while below-cloud removal is more frequent below 1000 m (52 % of all events) and can be important for the initial fate of species emitted at the surface, such as those examined here.For assumed realistic in-cloud removal efficiencies, both BC and sulphate have a slight overestimation of observed atmospheric concentrations (a factor of 1.6 and 1.2 respectively). However, this overestimation is largest close to the sources and thus appears more related to overestimated emissions rather than underestimated removal. The new aerosol wet removal scheme of FLEXPART incorporates more realistic information about clouds and aerosol properties and it compares better with both observed lifetimes and concentration than the old scheme.

Aerosols have important impacts on air quality and climate, but the processes affecting their removal from the atmosphere are not fully understood and are poorly constrained by observations. This makes modelled aerosol lifetimes uncertain. In this study, we make use of an observational constraint on aerosol lifetimes provided by radionuclide measurements and investigate the causes of differences within a set of global models. During the Fukushima Dai-Ichi nuclear power plant accident of March 2011, the radioactive isotopes cesium-137 (137Cs) and xenon-133 (133Xe) were released in large quantities. Cesium attached to particles in the ambient air, approximately according to their available aerosol surface area. 137Cs size distribution measurements taken close to the power plant suggested that accumulation-mode (AM) sulfate aerosols were the main carriers of cesium. Hence, 137Cs can be used as a proxy tracer for the AM sulfate aerosol's fate in the atmosphere. In contrast, the noble gas 133Xe behaves almost like a passive transport tracer. Global surface measurements of the two radioactive isotopes taken over several months after the release allow the derivation of a lifetime of the carrier aerosol. We compare this to the lifetimes simulated by 19 different atmospheric transport models initialized with identical emissions of 137Cs that were assigned to an aerosol tracer with each model's default properties of AM sulfate, and 133Xe emissions that were assigned to a passive tracer. We investigate to what extent the modelled sulfate tracer can reproduce the measurements, especially with respect to the observed loss of aerosol mass with time. Modelled 137Cs and 133Xe concentrations sampled at the same location and times as station measurements allow a direct comparison between measured and modelled aerosol lifetime. The e-folding lifetime τe, calculated from station measurement data taken between 2 and 9 weeks after the start of the emissions, is 14.3 days (95 % confidence interval 13.1–15.7 days). The equivalent modelled τe lifetimes have a large spread, varying between 4.8 and 26.7 days with a model median of 9.4 ± 2.3 days, indicating too fast a removal in most models. Because sufficient measurement data were only available from about 2 weeks after the release, the estimated lifetimes apply to aerosols that have undergone long-range transport, i.e. not for freshly emitted aerosol. However, modelled instantaneous lifetimes show that the initial removal in the first 2 weeks was quicker (lifetimes between 1 and 5 days) due to the emissions occurring at low altitudes and co-location of the fresh plume with strong precipitation. Deviations between measured and modelled aerosol lifetimes are largest for the northernmost stations and at later time periods, suggesting that models do not transport enough of the aerosol towards the Arctic. The models underestimate passive tracer (133Xe) concentrations in the Arctic as well but to a smaller extent than for the aerosol (137Cs) tracer. This indicates that in addition to too fast an aerosol removal in the models, errors in simulated atmospheric transport towards the Arctic in most models also contribute to the underestimation of the Arctic aerosol concentrations.

Vertical profiles of the optical (extinction and backscatter coefficients, lidar ratio and Ångström exponent), microphysical (mean effective radius, mean refractive index, mean number concentration) and geometrical properties as well as the mass concentration of volcanic particles from the Eyjafjallajökull eruption were retrieved at selected heights over Athens, Greece, using multi-wavelength Raman lidar measurements performed during the period 21–24 April 2010. Aerosol Robotic Network (AERONET) particulate columnar measurements along with inversion schemes were initialized together with lidar observations to deliver the aforementioned products. The well-known FLEXPART (FLEXible PARTicle dispersion model) model used for volcanic dispersion simulations is initiated as well in order to estimate the horizontal and vertical distribution of volcanic particles. Compared with the lidar measurements within the planetary boundary layer over Athens, FLEXPART proved to be a useful tool for determining the state of mixing of ash with other, locally emitted aerosol types. The major findings presented in our work concern the identification of volcanic particles layers in the form of filaments after 7-day transport from the volcanic source (approximately 4000 km away from our site) from the surface and up to 10 km according to the lidar measurements. Mean hourly averaged lidar signals indicated that the layer thickness of volcanic particles ranged between 1.5 and 2.2 km. The corresponding aerosol optical depth was found to vary from 0.01 to 0.18 at 355 nm and from 0.02 up to 0.17 at 532 nm. Furthermore, the corresponding lidar ratios (S) ranged between 60 and 80 sr at 355 nm and 44 and 88 sr at 532 nm. The mean effective radius of the volcanic particles estimated by applying inversion scheme to the lidar data found to vary within the range 0.13–0.38 μm and the refractive index ranged from 1.39+0.009i to 1.48+0.006i. This high variability is most probably attributed to the mixing of aged volcanic particles with other aerosol types of local origin. Finally, the LIRIC (LIdar/Radiometer Inversion Code) lidar/sunphotometric combined inversion algorithm has been applied in order to retrieve particle concentrations. These have been compared with FLEXPART simulations of the vertical distribution of ash showing good agreement concerning not only the geometrical properties of the volcanic particles layers but also the particles mass concentration.

Caesium-137 (137Cs) and iodine-131 (131I) are radionuclides of particular concern during nuclear accidents, because they are emitted in large amounts and are of significant health impact. 137Cs and 131I attach to the ambient accumulation-mode (AM) aerosols and share their fate as the aerosols are removed from the atmosphere by scavenging within clouds, precipitation and dry deposition. Here, we estimate their removal times from the atmosphere using a unique high-precision global measurement data set collected over several months after the accident at the Fukushima Dai-ichi nuclear power plant in March 2011. The noble gas xenon-133 (133Xe), also released during the accident, served as a passive tracer of air mass transport for determining the removal times of 137Cs and 131I via the decrease in the measured ratios 137Cs/133Xe and 131I/133Xe over time. After correction for radioactive decay, the 137Cs/133Xe ratios reflect the removal of aerosols by wet and dry deposition, whereas the 131I/133Xe ratios are also influenced by aerosol production from gaseous 131I. We find removal times for 137Cs of 10.0–13.9 days and for 131I of 17.1–24.2 days during April and May 2011. The removal time of 131I is longer due to the aerosol production from gaseous 131I, thus the removal time for 137Cs serves as a better estimate for aerosol lifetime. The removal time of 131I is of interest for semi-volatile species. We discuss possible caveats (e.g. late emissions, resuspension) that can affect the results, and compare the 137Cs removal times with observation-based and modeled aerosol lifetimes. Our 137Cs removal time of 10.0–13.9 days should be representative of a "background" AM aerosol well mixed in the extratropical Northern Hemisphere troposphere. It is expected that the lifetime of this vertically mixed background aerosol is longer than the lifetime of fresh AM aerosols directly emitted from surface sources. However, the substantial difference to the mean lifetimes of AM aerosols obtained from aerosol models, typically in the range of 3–7 days, warrants further research on the cause of this discrepancy. Too short modeled AM aerosol lifetimes would have serious implications for air quality and climate model predictions.

Volcanic aerosols resulting from the Eyjafjallajökull eruption were detected in south-eastern Italy from 20 to 22 April 2010, at a distance of approximately 4000 km from the volcano, and have been characterized by lidar, sun/sky photometer, and surface in-situ measurements. Volcanic particles added to the pre-existing aerosol load and measurement data allow quantifying the impact of volcanic particles on the aerosol vertical distribution, lidar ratios, the aerosol size distribution, and the ground-level particulate-matter concentrations. Lidar measurements reveal that backscatter coefficients by volcanic particles were about one order of magnitude smaller over south-eastern Italy than over Central Europe. Mean lidar ratios at 355 nm were equal to 64 ± 5 sr inside the volcanic aerosol layer and were characterized by smaller values (47 ± 2 sr) in the underlying layer on 20 April, 19:30 UTC. Lidar ratios and their dependence with the height reduced in the following days, mainly because of the variability of the volcanic particle contributions. Size distributions from sun/sky photometer measurements reveal the presence of volcanic particles with radii r > 0.5 μm on 21 April and that the contribution of coarse volcanic particles increased from 20 to 22 April. The aerosol fine mode fraction from sun/sky photometer measurements varied between values of 0.85 and 0.94 on 20 April and decreased to values between 0.25 and 0.82 on 22 April. Surface measurements of particle size distributions were in good accordance with column averaged particle size distributions from sun/sky photometer measurements. PM1/PM2.5 mass concentration ratios of 0.69, 0.66, and 0.60 on 20, 21, and 22 April, respectively, support the increase of super-micron particles at ground. Measurements from the Regional Air Quality Agency show that PM10 mass concentrations on 20, 21, and 22 April 2010 were enhanced in the entire Apulia Region. More specifically, PM10 mass concentrations have on average increased over Apulia Region 22%, 50%, and 28% on 20, 21, and 22 April, respectively, compared to values on 19 April. Finally, the comparison of measurement data with numerical simulations by the FLEXPART dispersion model demonstrates the ability of FLEXPART to model the advection of the volcanic ash over the 4000 km from the Eyjafjallajökull volcano to Southern Italy.

During the Eyjafjallajökull eruption (14 April to 24 May 2010), the volcanic aerosol cloud was observed across Europe by several airborne in situ and ground-based remote-sensing instruments. On 18 and 19 May, layers of depolarizing particles (i.e. non-spherical particles) were detected in the free troposphere above the Puy de Dôme station, (PdD, France) with a Rayleigh-Mie LIDAR emitting at a wavelength of 355 nm, with parallel and crossed polarization channels. These layers in the free troposphere (FT) were also well captured by simulations with the Lagrangian particle dispersion model FLEXPART, which furthermore showed that the ash was eventually entrained into the planetary boundary layer (PBL). Indeed, the ash cloud was then detected and characterized with a comprehensive set of in situ instruments at the Puy de Dôme station (PdD). In agreement with the FLEXPART simulation, up to 65 μg m−3 of particle mass and 2.2 ppb of SO2 were measured at PdD, corresponding to concentrations higher than the 95 percentile of 2 yr of measurements at PdD. Moreover, the number concentration of particles increased to 24 000 cm−3, mainly in the submicronic mode, but a supermicronic mode was also detected with a modal diameter of 2 μm. The resulting optical properties of the ash aerosol were characterized by a low scattering Ångström exponent (0.98), showing the presence of supermicronic particles. For the first time to our knowledge, the combination of in situ optical and physical characterization of the volcanic ash allowed the calculation of the mass-to-extinction ratio (η) with no assumptions on the aerosol density. The mass-to-extinction ratio was found to be significantly different from the background boundary layer aerosol (max: 1.57 g m−2 as opposed to 0.33 ± 0.03 g m−2). Using this ratio, ash mass concentration in the volcanic plume derived from LIDAR measurements was found to be 655 ± 23 μg m−3 when the plume was located in the FT (3000 m above the sea level – a.s.l.). This ratio could also be used to retrieve an aerosol mass concentration of 579 ± 60 μg m−3 on 19 April, when LIDAR observations detected the ash cloud at 3000 m a.s.l. in correspondence with model simulations (FLEXPART). On 22 April, another ash plume entered the BL, and although it was more diluted than during the May episode, the French research aircraft ATR42 that passed over Clermont-Ferrand in the PBL confirmed the presence of particles with a supermicronic mode, again with a modal diameter at 2 μm. This data set combining airborne, ground-based and remote sensing observations with dispersion model simulations shows an overall very good coherence during the volcanic eruption period, which allows a good confidence in the characteristics of the ash particles that can be derived from this unique data set.

The April–May, 2010 volcanic eruptions of Eyjafjallajökull, Iceland caused significant economic and social disruption in Europe whilst state of the art measurements and ash dispersion forecasts were heavily criticized by the aviation industry. Here we demonstrate for the first time that large improvements can be made in quantitative predictions of the fate of volcanic ash emissions, by using an inversion scheme that couples a priori source information and the output of a Lagrangian dispersion model with satellite data to estimate the volcanic ash source strength as a function of altitude and time. From the inversion, we obtain a total fine ash emission of the eruption of 8.3 ± 4.2 Tg for particles in the size range of 2.8–28 μm diameter. We evaluate the results of our model results with a posteriori ash emissions using independent ground-based, airborne and space-borne measurements both in case studies and statistically. Subsequently, we estimate the area over Europe affected by volcanic ash above certain concentration thresholds relevant for the aviation industry. We find that during three episodes in April and May, volcanic ash concentrations at some altitude in the atmosphere exceeded the limits for the "Normal" flying zone in up to 14 % (6–16 %), 2 % (1–3 %) and 7 % (4–11 %), respectively, of the European area. For a limit of 2 mg m−3 only two episodes with fractions of 1.5 % (0.2–2.8 %) and 0.9 % (0.1–1.6 %) occurred, while the current "No-Fly" zone criterion of 4 mg m−3 was rarely exceeded. Our results have important ramifications for determining air space closures and for real-time quantitative estimations of ash concentrations. Furthermore, the general nature of our method yields better constraints on the distribution and fate of volcanic ash in the Earth system.