A global network of nine Volcanic Ash Advisory Centres operates under the International Airways Volcano Watch established by the International Civil Aviation Organization to provide critical guidance on airborne volcanic hazards to national aviation regulators, commercial airlines, and airport authorities. Traditionally, these operational forecasting centres have issued guidance in the form of text-based Volcanic Ash Advisories and graphical Volcanic Ash Graphics outlining the expected cloud footprint as a simple two-dimensional polygon spanning up to 18 hours ahead. Under updated international service mandates set by the International Civil Aviation Organization, all advisory centres are required to transition toward delivering quantitative forecasts from November 2025. This upgraded operational framework, designated the Quantitative Volcanic Ash concentration service, requires centres to generate four-dimensional gridded datasets of ash concentrations, probabilities of exceeding defined concentration thresholds, and geometrically sophisticated three-dimensional polygon exchange objects. In a comprehensive operational report, researchers and meteorologists at the London Volcanic Ash Advisory Centre have detailed the scientific architecture and numerical dispersion setup implemented to deliver compliant quantitative and ensemble-based probabilistic forecasts.
The operational framework supporting the London centre relies fundamentally on the Numerical Atmospheric-dispersion Modelling Environment, known as NAME, which simulates atmospheric transport and dispersion within a Lagrangian mathematical framework. For deterministic operational runs, NAME is coupled offline with meteorological data extracted from the global configuration of the UK Met Office Unified Model. These driving numerical weather prediction fields are updated on a three-hourly temporal resolution across a horizontal grid spacing of approximately 10 kilometres, resolving 59 non-equally spaced vertical levels between the surface and 30 kilometres above sea level. To manage computational data volumes during routine operations, stratospheric conditions situated above 30 kilometres are assumed to remain sufficiently similar to those at the 59th vertical model level, allowing those top-level meteorological values to be persisted upward. Within NAME, particles are advected by the prescribed mean meteorological winds, while unresolved turbulent velocity structures are accounted for through random walk dispersion techniques, supplemented by specific parametrizations for mesoscale atmospheric motions that exceed turbulent scales but remain unresolved by driving weather grids.
Executing dispersion simulations requires initialising the atmospheric transport model with specific Eruption Source Parameters that characterize how volcanic particles are injected into the atmosphere. These essential parameters include the vent geographic coordinates, emission plume height, eruption start and duration timings, the total Mass Eruption Rate, and physical particle characteristics including size distributions, particle shape, and material density. Because direct near-real-time observations of active eruptions are notoriously difficult to measure during sudden crisis response, operational centres rely on early reports from state volcano observatories via Volcano Observatory Notifications for Aviation and satellite observations. In the complete absence of immediate observational constraints, the London centre applies standard default assumptions: ash is released as a uniform vertical line source between the volcano summit and plume top, while Mass Eruption Rates are derived using empirical scaling relationships from Mastin. The simulation tracks 15,000 computational particles per hour, assigning diameters between 0.1 and 100 micrometres, with a default airborne particle size distribution derived from the 1990 eruption of Mount Redoubt, an assumed particle density of 2300 kilograms per cubic metre, and an average particle sphericity of 0.5.
To represent physical grain populations accurately within the operational transport domain, a Distal Fine Ash Fraction scaling factor is systematically applied to the calculated Mass Eruption Rate. By default, this fraction is fixed at 5 percent of the total emitted mass to account for the restricted range of modelled grain sizes between 0.1 and 100 micrometres, as larger tephra fragments deposit rapidly near the volcanic vent due to gravitational settling. This operational scaling factor is also postulated to compensate for the current absence of explicit near-source removal mechanisms, such as particle aggregation, which can rapidly scavenge fine ash from volcanic plumes. While historical comparisons between satellite retrievals and modelled plumes supported this 5 percent baseline, recent evaluations have highlighted significant uncertainties and retrieval limitations in satellite data, prompting ongoing research into refining how observational constraints inform mass scaling. When preliminary observations become available, the London operational system dynamically updates eruption source parameters, adjusting historic emission profiles across preceding days without limiting the total number of discrete eruptive releases simulated within a single active crisis.
Beyond default empirical profiles, the London centre incorporates several advanced scientific schemes to refine eruption source terms as events evolve over time. When plume elevation is directly observed, an integrated one-dimensional steady-state buoyant plume model accounts for entrainment mechanisms driven by crosswinds and velocity differences, as well as atmospheric and magmatic moisture, to recalculate the corresponding mass eruption rate or deduce plume height from an estimated mass discharge. For massive explosive events producing radially spreading umbrella clouds, an umbrella cloud parameterization models gravity current dynamics spreading horizontally against ambient atmospheric wind fields. Furthermore, when downwind satellite retrievals are available, the Inversion Technique for Emissions Modelling, known as InTEM, functions as an operational data fusion tool. InTEM applies mathematical inversion methods that combine satellite observations, a priori parameter estimates, and their respective observational and numerical uncertainties to derive an optimal time-varying vertical distribution of mass emissions at the volcanic vent, significantly improving subsequent downwind dispersion calculations.
Within the atmospheric transport domain, the NAME dispersion code explicitly parametrizes multiple distinct physical sink mechanisms responsible for removing airborne ash. Gravitational sedimentation is computed using a Reynolds number-dependent drag relation established by Ganser, which accounts for particle non-sphericity and yields terminal settling velocities within approximately 20 percent across expected atmospheric regimes. When microscopic particles enter the sub-micrometre range and approach the mean free path of ambient air molecules, the model introduces the Cunningham Correction Factor to adjust for diminished aerodynamic drag within the transition flow regime. Turbulent surface deposition is calculated through an electrical resistance analogy combined with gravitational sedimentation. Wet deposition processes are governed by size-dependent scavenging coefficients that represent both within-cloud rainout, where volcanic ash particles become incorporated into forming cloud droplets prior to falling as precipitation, and below-cloud washout, where falling hydrometeors physically collide with and sweep ash out of the air column. Processes such as explicit aggregation, gravitational convective instabilities, hydrometeor generation, and direct particle-particle interactions currently remain unrepresented in routine operations.
The service specifications defined by the International Civil Aviation Organization mandate substantial increases in output data resolution compared to historical legacy products. Under the new Quantitative Volcanic Ash framework, advisory centres must deliver gridded datasets across a horizontal resolution of 0.25 degrees latitude and longitude, a vertical structure spanning 12 standardized Flight Levels spaced at 50-Flight-Level intervals between mean sea level and Flight Level 600, and a three-hourly temporal resolution providing slices at 0, 3, 6, 9, 12, 15, 18, 21, and 24 hours post-issue. For direct digital consumption by aviation stakeholders, gridded fields are distributed via an Application Programming Interface using NetCDF formats. In contrast to legacy advisory graphics, which compressed predicted concentrations into three very coarse vertical layers and relied on simple bounded polygons, the new framework introduces standardized three-dimensional prism objects encoded in the machine-readable Meteorological Information Exchange Model, or IWXXM, format. These digital exchange polygons possess unlimited vertices, allowing precise delineation of complex cloud boundaries across multiple prescribed concentration thresholds.
To evaluate the quantitative forecasting pipeline, researchers conducted deterministic demonstration simulations based on a hypothetical eruption scenario at Hekla volcano in Iceland using baseline operational configurations. In comparison with traditional Volcanic Ash Graphics that encompass broad geographic zones with a single polygon enclosing all ash exceeding roughly 0.2 milligrams per cubic metre, the quantitative forecast resolved fine-scale structural variation across atmospheric flight corridors. At an 18-hour forecast horizon, deterministic outputs demonstrated that core regions containing ash concentrations equal to or exceeding 10 milligrams per cubic metre remained confined to a narrow southeastward corridor between Flight Level 150 and Flight Level 350, with sharply diminishing concentrations above and below these flight altitudes. The accompanying three-dimensional IWXXM polygon object captures the maximum lateral extent of ash exceeding specified concentration thresholds while precisely recording the lowest and highest altitudes attained by the plume across all vertical levels.
Recognizing the substantial uncertainty inherent in numerical dispersion modelling, the quantitative framework introduces probabilistic ash forecasts driven by ensemble weather predictions. Forecasting uncertainties originate from multiple interacting sources, primarily numerical weather prediction wind and precipitation fields, initial eruption source parameters, and model physics parametrizations. To represent atmospheric variability, the London centre couples NAME with the Met Office Global and Regional Ensemble Prediction System global configuration, known as MOGREPS-G. This meteorological ensemble generates 18 distinct analysis and forecast members using hybrid four-dimensional ensemble variational data assimilation, resolving weather fields at a horizontal grid spacing of roughly 20 kilometres across mid-latitudes and 59 vertical model levels. Operating 18 parallel NAME simulations across these ensemble members allows forecasters to compute uncalibrated model probabilities of exceeding prescribed ash concentrations based on the exact proportion of ensemble members surpassing a given threshold at any spatial grid point.
The discrete mathematical structure of an 18-member meteorological ensemble imposes clear operational constraints on probability resolution. A single member exceeding a threshold establishes the minimum non-zero probability of approximately 6 percent, with subsequent probability increments stepping through roughly 11, 17, and 22 percent up to complete agreement at 100 percent. In demonstration runs for the hypothetical Hekla scenario at Flight Level 150 to 200, the ensemble confirmed high confidence that dense ash exceeding 10 milligrams per cubic metre remained confined to the central cloud core without impacting Norway, while revealing a 10 to 30 percent probability that dilute ash exceeding 0.2 milligrams per cubic metre could extend over northern Scotland and coastal Norway up to 62.5 degrees north. Similar retrospective evaluations of the 2019 Raikoke eruption demonstrated that ensemble spread accurately reflected meteorological dispersion uncertainty, with 9 of the 18 ensemble members showing no ash exceeding 0.2 milligrams per cubic metre south of the Aleutian Islands below 50 degrees north, while the remaining members tracked distinct southward transport pathways.
Independent evaluations by atmospheric researchers have demonstrated that dispersion runs driven by forecast ensemble meteorology consistently outperform single deterministic runs when validated against analyzed atmospheric states. These studies indicate that MOGREPS-G provides appropriate spread and produces well-calibrated ash concentration forecasts throughout the troposphere, although dispersion predictions exhibit an increasing tendency toward over-confidence when plumes ascend into the stratosphere. Operationally, handling high-resolution 4D concentration grids across 18 ensemble members within real-time event response timelines has required extensive code optimisation and high-performance computing infrastructure. The London centre continues developing technical pathways to expand ensemble sampling, including time-lagged ensemble strategies that combine successive forecast cycles to achieve an effective sample size of 36 members, thereby halving the minimum probability resolution to approximately 3 percent and suppressing correlated forecast errors.
A critical limitation of the initial operational capability is that probabilistic calculations reflect only meteorological variability, while eruption source parameters remain deterministic. In real-world volcanic crises, observational constraints on plume height and discharge rates vary dramatically between events and across eruptive pulses, introducing profound uncertainties in downwind mass concentrations. Ongoing scientific work at the London centre is establishing a unified statistical framework designed to integrate parameter uncertainties alongside ensemble meteorology, particularly addressing complex interactions where uncertain injection heights intersect strong vertical wind shear. Furthermore, model verification remains fundamentally constrained by existing satellite observing systems, which typically measure column-integrated mass loadings in grams per square metre rather than vertically resolved concentrations in milligrams per cubic metre. As the aviation community prepares for the formal rollout of the Quantitative Volcanic Ash service, comprehensive verification across future eruptive events will remain essential to refining operational accuracy and guiding risk-informed flight routing.
Subject of Research: Operational quantitative and probabilistic volcanic ash dispersion forecasting for aviation
Article Title: Atmospheric dispersion modelling at the London VAAC: development of quantitative and probabilistic volcanic ash forecasts
Article References: Beckett, F. M., Bates, C. E., Evans, B. R., Hort, M. C., Jones, A. R., Kristiansen, N. I., Sherratt, V., Smedley, O., Stebbing, N., Trzeciak, T. M., & Tucker, C. (2026). Atmospheric dispersion modelling at the London VAAC: development of quantitative and probabilistic volcanic ash forecasts. Journal of Applied Volcanology, 15(1), Article 11. https://doi.org/10.1186/s13617-026-00166-7
Image Credits: AI Generated
DOI: 10.1186/s13617-026-00166-7
Keywords: Volcanic Ash Advisory Centre, London VAAC, Quantitative Volcanic Ash, NAME dispersion model, MOGREPS-G ensemble, aviation safety, atmospheric dispersion, eruption source parameters, International Civil Aviation Organization, IWXXM format, flight levels, InTEM inversion, volcanic ash forecasting, numerical weather prediction, mass eruption rate, sedimentation drag, ensemble forecasting
Cite Scienmag News
Russell Cooper. (November 18, 2026). London VAAC Develops Quantitative and Probabilistic Forecasts for Volcanic Ash. Scienmag. https://scienmag.com/london-vaac-develops-quantitative-and-probabilistic-forecasts-for-volcanic-ash/
Russell Cooper. "London VAAC Develops Quantitative and Probabilistic Forecasts for Volcanic Ash." Scienmag, 18 November 2026, https://scienmag.com/london-vaac-develops-quantitative-and-probabilistic-forecasts-for-volcanic-ash/. Accessed 20 November 2026.
Russell Cooper. "London VAAC Develops Quantitative and Probabilistic Forecasts for Volcanic Ash." Scienmag. November 18, 2026. https://scienmag.com/london-vaac-develops-quantitative-and-probabilistic-forecasts-for-volcanic-ash/








