In mesoscale models (resolution
A critical factor in successfully monitoring and forecasting volcanic ash and
gases dispersion is the height reached by eruption clouds, which is mainly
controlled by the eruptive mass flux
The convective scale of a volcanic plume corresponds to the unstable region where intense but localised sensible and latent heat fluxes released by pyroclasts, gases and lava near eruptive vents generate convection which transports energy and pollutants to higher altitudes through buoyant plumes. Throughout the course of this convection, mixing of the plume with the atmosphere takes place at different levels of altitude through entrainment and detrainment. This process allows for the distribution of pollutants over a certain vertical range.
Piton de la Fournaise (PdF) is one of the world's most active volcanoes
Simulations of atmospheric plumes from intense heat source points have been
performed using the Méso-NH
Simulations of buoyant eruptive columns, chemistry dispersal in the proximal environment and the volcanic cloud tracking at regional scale can be based on similar numerical and conceptual approaches as the ones used for the study of forest fire plumes. However, volcanic eruptive vents usually cover small areas and in (at best) kilometric-resolution models used for air-quality purposes (simulation or forecasts), the localised heat source is diluted in the model grid; hence, no convection is explicitly generated.
Several types of atmospheric movements are sub-grid processes, and they are incorporated into atmospheric models through appropriate parameterisation schemes. In order to determine the evolution of volcanic plumes in the atmosphere, numerical models need to consider two different scales:
an implicit/convective scale corresponding to the convective
plume above the erupting volcano, whose processes are sub-grid even
at fine resolutions ( an explicit/dispersion scale that corresponds to the dispersion
of the volcanic plume in the atmosphere.
In mesoscale models used for regional dispersion of pollution plumes (target
resolution
Due to the computational efficiency of a one-dimensional (1-D) model and the
ability to isolate a column of atmosphere for study, 1-D modelling is an
ideal configuration to develop and test parameterisations
Simultaneously, a three-dimensional (3-D) large eddy simulation (LES) is
performed (10
An eruption took place on 2 January 2010 around 10:20 UTC at the summit of
PdF located at 2632
January 2010 summit eruption of Piton de la Fournaise: the
60
The vertical plume above the crater (Fig.
It is well understood that a volcanic eruption plume enters into an
atmosphere that has a pre-existing stratification in terms of temperature,
moisture content and wind the gas thrust region, where the dynamics is dominated by the
exit velocity at the vent and the flow near the vent is driven
upward by its initial kinetic momentum; buoyancy-driven convective region which covers most of the
height of the plume; and umbrella cloud region, where vertical motion is small and the
plume disperses horizontally due to wind impacts.
For the purpose of modelling volcanic clouds using Méso-NH, we are
predominately interested in the convective region of the volcanic cloud. For
the kind of effusive eruption under consideration in this study, the gas
thrust region extends only over few metres (see, e.g. Fig.
Temperature (
The current updraft model used in Méso-NH defined by
The basic idea of the EDMF (eddy diffusivity/mass flux) approach is to represent
vertical transport of matter and energy that occurs at the sub-grid scale in
numerical simulations of the convective boundary layer (CBL) with resolutions of
turbulent eddies convective updrafts and compensating downdrafts.
Turbulent transport is commonly parameterised with the eddy-diffusivity (ED)
method, corresponding fluxes being written in the form of
A grid box can contain multiple convective updrafts. For simplicity
a single updraft is considered carrying the properties of the ensemble
of updrafts. This is known as the mass-flux (MF) approach. The
fraction of the total area of a grid box that is covered by the
updraft is known as the fractional updraft area (
Both ED and MF approaches have been combined in a single EDMF parameterisation such that nonlocal
sub-grid transport due to strong updrafts is taken into account by MF,
while the remaining transport is taken into account by ED
The two key parameters determining the mass-flux profile are
entrainment (
The mass-flux evolves along the vertical at a rate given by the
difference between the
In
The updraft buoyancy acceleration is evaluated in relation to the difference of
virtual potential temperature (
Variables and values used for LES and SCM models.
Firstly, in the current EDMF parameterisation
Secondly, the updraft fraction area is simply initialised as the ratio
of the fissure surface (
Now, as
Equation (
Figure displaying the input data, the mass flux at ground
level (
Entrainment of ambient air through turbulent mixing plays a central role in
the dynamics of eruption plumes, primarily because the plume density is
controlled by the mixing ratio between ejected gas/material and ambient
air
In this sub-section we present the modifications to the input method
of
The importance of adjusting the ground level
Let
Let
If
The interconnection in terms of the simulation domain between the three sets of simulations performed: Spin-up, SCM and LES. The single cell corresponding to the fissure is tagged for LES.
Finally, using Eqs. (
For our chosen case study, three sets of simulations were run as
depicted in Fig.
Section Section Section
The Méso-NH model (version MNH-4-9-3) is used in this study; it is
a mesoscale non-hydrostatic atmospheric model able to simulate
convective motion and flow over sharp topography. This model has been jointly
developed by Laboratoire d'Aérologie (UMR5560 UPS/CNRS) and Centre
National de Recherches Météorologiques – Groupe d'études de
l'Atmosphère Météorologique, CNRM-GAME (UMR3589
CNRS/Météo-France), and is designed to simulate atmospheric
circulations from small-scale (type – LES) to synoptic-scale phenomena
Different sets of parameterisations have been introduced for cloud
microphysics
A 3-D spin-up simulation is performed to generate the
background profiles which are used for SCM and LES. Two, two-way grid-nested
domains with horizontal mesh sizes of 4 and 1
Figure
Meteorological profiles from the 3-D spin-up model (1
The vertical structure of trade winds over Réunion Island was
investigated by
It should be noted that the wind profiles as obtained from the spin-up simulation
appeared to be unrealistic since the wind near the ground
(7–8
An LES model has such a high resolution that it can resolve not only convective motions but also the largest eddies (responsible for the major part of the turbulent transport). This section describes the set-up of the LES simulation considered as reference used to validate the EDMF parameterisation for volcano-induced convection.
Table
LES model configuration.
The surface mass and heat fluxes representing the volcanic mass and energy
source in the LES are prescribed for one single surface cell (i.e.
Let
Steady surface fluxes are used as volcanic input in LES runs. Their values
are summarised in Table
Table
SCM model configuration.
The adapted EDMF model in Sect.
Common to both LES (Sect. the wind profiles obtained from the spin-up simulation are not used as
background conditions, instead a prescribed uniform wind field is used
(u radiative processes are neglected.
Finally, the comparison method of LES and SCM simulation results is as
follows: fields horizontally averaged over the full LES simulation domain
(1
In this section, results obtained from the 1-D SCM and 3-D LES of the case study are presented and analysed.
A first most obvious question is whether we need to parameterise volcanic
updraft. Figure
In simulations with no volcanic heat source,
Although both Fig.
Hereafter, the height at which there is a maximum detrainment of the tracer will be referred to as the “maximum injection height”. The sensitivity of the latter against entrainment and detrainment at the base of the updraft will be investigated, with the aim at obtaining better agreement between the reference LES and M.EDMF simulations.
It is well known that both entrainment and detrainment have an impact on the
updraft development because they affect buoyancy at all updraft levels
Figure
Updraft temperature (
The question of fresh air entrainment at the base of highly buoyant plumes is
actually relevant for all types of high-temperature surface sources inducing
convection in the atmosphere, i.e volcanoes but also combustions and in
particular biomass fires
In the present work we keep it as simple as possible. We started from the
simple observation that a dominant part of fresh air has been already
entrained into the plume within few tens of metres above the ground
(Fig.
Sensitivity of plume characteristics to various entrainment
(
To compare the
The sensitivity of our model to a range of prescribed values of
Figure
Figure
Results from the best-fitted SCM simulation with respect to the
reference LES simulation (Fig.
Up to here,
In our simulation, ad hoc fresh air entrainment is prescribed only in the
first model layer. The question that arises is whether the fraction
In order to represent deep convective injections of volcanic emissions
into the low to mid troposphere in case of effusive eruptions, the
EDMF parameterisation by
We have shown the need to input the specific heat source in order to generate
deep plumes using the Méso-NH model by adapting the EDMF scheme. LES
simulations were also initialised using water vapour mass flux, sensible heat
flux and
Entrainment of ambient air in a volcanic plume is largely known to be one of
the key parameters affecting its buoyancy. Since the first experiments by
As this parameterisation has been used in an idealised and controlled set-up for one particular case study (January 2010 summit eruption), further work needs to be undertaken whereby the parameterisation is tested for different configurations (i.e. changes in volcanic heat sources; idealised and real case simulations). Furthermore, further investigation is needed on how entrainment and detrainment should be formulated, not only at the base but also at all levels of the updraft. Ideally, a formulation valid at all levels and for a large variety of eruption cases should be sought.
In the case of the LES, the surface fluxes corresponding to those from the
volcanic updraft at surface level in the single column simulation, occur over
one whole grid cell and hence the surface
To evaluate the mass flux of e.g. water vapour, the question to answer is
what mass
The surface sensible heat flux is basically the energy quantity per
unit of time and surface which is efficient in causing a temperature change at
constant pressure in the lowest atmospheric layer. Therefore, the enthalpy
change must be considered. The enthalpy change
We want to know what enthalpy change is caused in the atmosphere by injection
of a mass
This finally yields
Méso-NH model documentation and the model itself are available from the
website
We greatly acknowledge the Piton de la Fournaise Volcanological Observatory (OVPF/IPGP) for providing pictures of the January and October 2010 summit eruptions, along with information relating to the eruptions themselves. We also thank the Méso-NH assistance team for continuous support and C. Barthe and Meteo-France for kindly providing us with ALADIN-Réunion atmospheric files as well as wind observation data. This work was performed using HPC resources from GENCI-IDRIS (grant x2014010005) and CALMIP (grant P12171). We wish to acknowledge the use of the NCAR Command Language (NCL, Boulder, Colorado) version 6.0.0 software for analysis and graphics in this paper. We thank the Observatoire des Milieux Naturels et des Changements globaux (OMNCG) and Observatoires des Sciences de l'Univers (OSU), Réunion along with the MoPAV project of the LEFE – CHAT program by INSU – CNRS (Institut National des Sciences de l'Univers – Centre National de la Recherche Scientifique) for their financial support and interest in this project. We finally thank P. Bechtold and the other anonymous referee for their careful reviews and constructive comments, as well as our colleage J.-P. Pinty for helpful discussions. Edited by: R. Sander