The current state of understanding of the dynamics of volcanic plumes, which have benefited from the addition of innovative techniques measuring physical and chemical parameters relatively close to vents (e.g. see review in Sparks et al., 1997; Vergniolle and Gaudemer, 2015; Costa et al., 2016), is now sufficiently advanced to envision a very fruitful collaboration with atmosphere scientists. Volcanoes release large quantities of gases which are atmospheric pollutants (e.g. SO2, halogens), as well as liquid/solid fragments of various compositions, sizes and amounts, all of which lead to various types of natural hazards when dispersed in the atmosphere even far from the volcanic source: poor air quality, acid rain, ash plumes threatening aircraft traffic, ozone depletion, climate forcing.
With the further concern of modelling atmospheric dispersion of volcanic products (a topic specifically addressed by WP3), the central questions of WP2 are: to what height is the volcanic plume injected, what are the relative mass flux of ash and volcanic gases, how are the volcanic products vertically distributed in the atmosphere and what is the lateral distribution of the volcanic products in the spreading top of the volcanic column before dispersion by atmospheric currents takes over?
This is of primary importance since atmospheric properties (pressure, temperature, humidity, wind speed and direction, static stability conditioning vertical diffusion) greatly varies with height in both the troposphere and stratosphere and thus the dispersion of volcanic products can vary markedly according to the height to which they are injected.
During volcanic eruptions, a multi-phase mixture (various gas species and solid/liquid aerosol particles) is expelled into the atmosphere with ejection velocities from a few 10’s up to 350 m/s and maximum heights of several hundreds of metres to some tens of km according to the style of eruption (e.g. Woods, 1988, 1993). The large range of parameters and scales, and the variability of the volcanic processes adds an extra layer of complexity, because different volcanoes or even eruptive episodes at the same volcano potentially require various assumptions to account for different regimes.
Volcanological studies are generally focused on the eruption dynamics at the vent but also at depth (conduit, reservoir), using measurements just above the vent, such as e.g. infrasonic (gas volume, gas pressure, a proxy for the gas velocity), seismic (e.g. depth of the source, processes in conduit/reservoir), thermal (ejection temperatures and velocities from thermal images), Doppler radar (gas and fragment velocity, and grain-size distribution), halogens, SO2 and aerosol particles flux (e.g. Balcone-Boissard et al., 2008, 2010, 2012 and references therein), gas composition and pyroclast sample collection (textures in relation to geochemical data). These variables, recorded as quasi-continuous time-series, can be used for modelling volcanic plumes above the vent. Numerical fluid-mechanical models and scaled-analogue experiments provide outputs such as shape of the eruptive column, velocity, temperature, ash concentration and flux as functions of height above the vent and of time, as well as the plume maximum height. Previous models have been validated for strong volcanic plumes (with high mass flux at the source) by comparing field estimates of mass flux (kg/s, from the spatial distribution of ash deposit thicknesses and grains size with the distance to the vents) with those provided by the model (e.g. Carazzo et al., 2008a, 2008b, Girault et al., 2014, Costa et al., 2016). However volcanic plumes of intermediate height – corresponding to smaller mass flux at the vent – are more complicated to understand and many effects should be added in the models, such as the effect of air humidity, initial gas overpressure, the thermal and mechanical coupling of gas and pyroclasts in the volcanic plume (Fig. 2), the dynamics of the umbrella cloud during lateral emplacement at its maximum height, as well as the often transient nature of such eruptions (e.g. Scase, 2009). Especially once water vapor saturation is reached in a buoyant volcanic updraft, a cloud forms with further release of heat, and the full complexity of cloud dynamics, microphysics (i.e. interactions between condensation nuclei – fine solid and liquid particles – and the various forms of hydrometeors: liquid water droplets, ice particles, snow, graupel, hail, etc…), and heterogeneous chemistry should be taken into account. Furthermore, recent attempts to couple a view of the physical processes at the vent with the development of the updraft in the atmosphere up to its maximum height, performed within the framework of the ANR STRAP, has shown how complex the physical processes are at the vent and the need to test the validity of existing atmospheric models in regards to the geological records.
In general, the scope of model studies in pure volcanology research ends at the stage of determining the plume maximum height and regime (e.g. buoyant plume versus collapsing fountain) for a given eruptive mass flux and under given atmospheric ambient profiles. On the other hand, even state-of-the-art operational atmospheric dispersion models deal with the vertical plume motion in a crude way: often, a plume maximum height and an eruption mass rate are determined from observations and/or empirical models, and a certain quantity of volcanic products (SO2 and/or ash concentration) is then manually introduced at a given height in the dispersion model. Such an approach clearly fails to adequately deal with the full range of thermo-chemical processes present in the rising plume with obvious consequences for the accuracy and applicability of the lateral dispersion models.
But much further advantage could be taken from all the detailed updraft quantities on the vertical by volcanic plume models, to correctly and dynamically seed atmospheric dispersion models with volcanic products at any vertical levels above the vent, and also to take into account the feedback of atmosphere properties on the volcanic plume vertical development. Great benefits could thus be taken in volcanic plume modelling and atmospheric dispersion if state-of-the-art models from both communities – volcanology and atmospheric sciences – could be fully coupled, to obtain a single numerical prediction system modelling both the volcanic plume dynamics and its atmospheric dispersion in the same time, with sufficiently low numerical cost allowing its utilization in forecast mode. Such an integrated model will open the way to the use of atmospheric observables beyond the plume height to constrain the volcanic sources.