Modeling and forecasting the distribution of volcanic ash...

Post on 22-Jul-2020

0 views 0 download

transcript

Modeling and forecasting the distribution of volcanic ash from the Eyjafjallajökull

eruption

A. Stohl, N. I. Kristiansen, S. Eckhardt, P. Seibert, F. Prata, J. F. Burkhart, K. Tørseth

Picture courtesy: Magnús Tumi Guðmundsson

This

study is partly supported

by ESA within

the Support to Aviation

for Volcanic Ash Avoidance

(SAVAA) project

15 April 0 UTC 16 April 0 UTC

17 April 0 UTC 18 April 0 UTC

The synoptic situation: Geopotential

@ 500 hPa

The FLEXPART modelLagrangian

particle

dispersion

model, similar

to the one

used

at the London VAAC

Meteorological

input data:Forecasts

use

NCEP GFS data, Analyses ECMWF data

Simulation

of volcanic

ash

in 31 size

bins-

gravitational

settling

-

dry deposition-

wet

deposition

Transport of 15 million ash

particles

by mean

winds, turbulence, convection, sedimentation

FLEXPART simulations of the eruptionSecond forecast ready: 15 April 16:00 UTC

http://transport.nilu.no/products/eyjafjallajokull

Forecasting System

Interactive

Tool

The source term – the first great unknown• Total ash

emission highly

uncertain-

Assume

10% of reported

tephra

production

in modeled

size

range (0.5-290 μm)

• Time variation

of emission rate not well

known-

Take what’s

available, anecdotal

evidence

• Eruption

column

height

is variable-

Reports, satellite

data (max. height

11 km)

• Height

emission profile

unknown- 4-layer C-shape

• Ash

size

distribution

at point

of emission not well

known-

Taken

from ground

samples around

Eyjafjallajökull

but

they

are biased

towards

larger

sizes

All these

uncertainties

will

affect

the final model

result!

Ash plume transport to Europe

Comparison with SEVIRI ash retrievals

15 April, 10 UTC

FLEXPART total ash15 April, 9-12 UTC

Comparison with SEVIRI ash retrievals (credit: Mike Pavolonis)

17 April, 20 UTC

FLEXPART total ash17 April, 18-21 UTC

Comparison with SEVIRI ash retrievals (credit: Mike Pavolonis)

18 April, 4:20 UTC

FLEXPART total ash18 April, 3-6 UTC

CloudsClouds

Comparison with AURA/OMI aerosol index (credit: AURA/OMI team)

FLEXPART total ash15 April 9-12

AURA/OMI aerosol index15 April 12 UTC

?

Increased

emission rate too

late?

Comparison with SEVIRI ash retrievals

16 April, 6 UTC

FLEXPART total ash16 April, 3-6 UTC

Comparison with SEVIRI ash retrievals

16 April, 18:30 UTC

FLEXPART total ash16 April, 18-21 UTC

Good agreement with London VAAC:  16 April 00:00

Notice: qualitative

comparison;isolines

not strictly

comparable

16 April 18:00

17 April 12:00

Howto

improve

the source termUse

observation

at the eruption

Site

(a priori) andUse

inverse modelling

approach

Effect

of wind

shear4-Nov-2002 eruption of Reventador, Ecuadorare

SO2

and Ash

Eruption

Sources x

(1..n) xa

a priory profile

Satellite observation y0

(1..m)

M

Emission sensitivity Matrix (m

n), as obtained from FLEXPART

σ

standard error of observation

Source-receptor matrix calculation with a Lagrangian particle dispersion model in backward mode, P. Seibert and A. Frank, ACP, 4, 51-63, 2004.

Inversion

Method

Emission Profile

(Jebel, 2007)

Challenges

of the EYJA* eruptionSource TermEruption

takes place

over a long

time periode, emissions

are constantly

added

up (started

runs every

3 hours

over 2 weeks, satellite

images available

every

15 mins -

SEVIRI)3 unknown

variables: timing, emission profile, size

distribution

TransportMainly

Ash

is emitted

shows many small‐scale features: 

ground‐bases lidar

of limited representativity, aircraft in‐situ data may 

also be difficult to interpret 

satellite data probably best data source 

observations alone far from delivering information need by air traffic 

authorities 

Why

is EYJA* difficult

to model

Different grainsize

settling

different transport direction

Conclusions•

Comparison with

measurement

data encouraging, FLEXPART and operational

VAAC model

(NAME) in good qualitative agreement

Quantitative dispersion simulations require quantitative source term (function of height, time, particle size range) -

SO2

source terms for explosive 

eruptions can be derived well with our inversion method ‐

ash

source term for 

continuous eruption much more complicated, research and development needed ‐

important role for inverse modelling

of satellite data is likely, but kind and way of 

usage of additional (in‐situ, ground‐based and airborne) measurements needs to 

be worked out 

Long‐lasting eruption of medium strength, emitting into the middle and upper 

troposphere, have the strong impacts on air traffic as large regions are 

contaminated (due to synoptic variability) and starting/landing

inhibited (due to 

low elevation of ash) ‐

the cloud shape tends to be quite complicated after some 

days of transport

Thank

you

for your

attention, ESA for funding.