Operational

Integrated System for wild-land fires

IS4FIRES

M.Sofiev1, J.Soares1, J. Hakkarainen1, E.

Ermakova2, R. Vankevich2

1.Finnish Meteorological Institute

2.Russian State Hydrometeorological University

Russian State

Hydrometeorological University

Outline

• Components and features of IS4FIRES

• From emission to AQ forecasting

• Specifics of IS4FIRES

diurnal variation of fire emission intensity

injection height of fire plumes

• Example of applications: fires 2012 in Eurasia

• Summary

TA

FRP

AOD

FRP

FRP

FRP

Plume top

CTM

(SILAM)

Emission factors

Diurnal variation

Emission flux

Injection height profile

Land use

Plume rise

param.

Offline

calibration

(Sofiev et al,

2009)

FRPFRPFRPFRPFRP

Fire information to emission: IS4FIRES v1.3

MODIS

MISR

SEVIRI

NRT&

offline

Offline

(Sofiev et al,

2012)

OfflineECMWF

Meteo

parameters

OfflineUSGS

Features of IS4FIRES v.1.3• Domains: global, European; resolution: 10km, daily + diurnal variation SEVIRI & VIIRS

• Timeliness: MODIS Rapid Response, persistence-based forecast

NRT fire alert for Finland (receiving antenna in Sodankyla)

• Primary scaling: FRP / TA to PM10 (Sofiev et al., 2009)

Cross-scaling (Andreae & Merlet, 2001): PM2.5, SO2, NO2 CO, NH3, HCHO

• Injection height:

crude ~(11.5) HABL

average injection profiles (global 3D map, 10 resolution)

dynamic (Sofiev et al, 2012)

• AQ impact: chemistry transport model SILAM

Fire plume forecasting: Europe, resolution of 20km and 1hr, horizon 72 hours

persistence assumption on fire behavior is the main limitation

• Data are available at

Emission 10km: http://is4fires.fmi.fi

Emission 0.5 deg: http://www.geiacenter.org

Forecasted PM fields: http://silam.fmi.fi

Plume rise: a new approach

• Plume rise: Briggs and others

Hot plume rises until is bended

out:

horizontal motion >> vertical

motion

Empirical parameterizations,

rise is limited for all stratification

types

• CAPE

Overheated parcel rises until in

equilibrium with surrounding air:

energy excess=0.

Rising through unstably

stratified layer increases energy

excess

U(z)

w(Tp,Ta,w0,..)

Data•MODIS active-fires: FRP

•ECMWF NWP fields: Habl (atmospheric boundary layer height), NFT (Brunt-Vaisala

frequency for free troposfere)

•MISR: plume height (Kahn et al., 2008; Mazzoni et al., 2007) to obtain α, β, γ, δ

coefficients α - ABL fraction passed “freely”

β - weight the contribution of the fire energy

γ - determines the power-law dependence on FRP

δ - determines the dependence on stability in the FT

Coefficient identification• MISR plume-height archive (fires in US, Siberia, Borneo, Africa, ~3000 cases)

• split the archive to learning and control dataset, perform fitting

• ranking fitting criterion: plume top within 500m from observed (approximate

accuracy of MISR itself)

Independent evaluation• CALIOP observations

Plume-rise formula (Sofiev et al, ACP, 2012)

' 2

' FTN

p ablH H FRP e

Comparison with other methods

0

1000

2000

3000

4000

0 1000 2000 3000 4000

Observed height

Ca

lcu

late

d h

eig

ht

High: 14%

Good: 42%

Low: 36%

0

1000

2000

3000

4000

0 1000 2000 3000 4000

Observed height

Ca

lcu

late

d h

eig

ht

High: 12%

Good: 37%

Low: 43%

0

1000

2000

3000

4000

0 1000 2000 3000 4000

Observed height

Ca

lcu

late

d h

eig

ht

Low: 17%

Good: 51%

High: 14%

0

1000

2000

3000

4000

0 1000 2000 3000 4000

Observed height

Ca

lcu

late

d h

eig

ht

Low: 17%

Good: 65%

High: 18%

Briggs’69

1-D model

BUOYANT

Briggs’84

New

formula

Hit rate (within 500m) for full MISR set (~2000 fires)

0

10

20

30

40

50

60

70

Briggs'69 Briggs'84 OND-86 BUOYANT Const =

1289m

New formula

Fire emission vertical profile: climatology

• Goal: 3-D mean global distribution of fire emission

supplement for existing fire emission databases

• Input

MODIS active-fires: FRP

ECMWF NWP fields: Habl, NFT

estimated diurnal variation using SEVIRI FRP

• Processing

two years (2001, 2008) ready (Sofiev et al, ACPD, 2012), full MODIS archive (2000-2012) on-going

sum-up the fire plumes, then normalize

fill-in gaps (no-fires grid cells with burns nearby)

• Evaluation

CALIOP

Height of 90% of emission injection, Feb, Aug

Injection profile, diurnal variation, Aug

Night time

Daytime

Recent fires (2012, July-November)

• July: Siberia and far east of Russia

• July-August: Northern America

• August: southern and south-eastern Europe

• September - October: Amazon basin

• October: Australia

• … and Africa non-stop, as usual…

5 July 2012: hemispheric-scale impact

Compare with: August, 2010, Russia

Moscow, 6.8.2010

Predicted (+16hrs):

1.5 mg PM10 / m3

Observed:

1.3 mg PM10 / m3

Virolahti, 8.8.2010

Predicted (+24hrs):

120 mg PM2.5 / m3

Observed:

140 mg PM2.5 / m3

Summary

• IS4FIRES data- and knowledge-base: http://is4fires.fmi.fi

Emission database, 2000-c.m. (daily, 10km, global, NRT)

GEIA database: 2000-2011 (daily, 0.5º, global)

• New semi-empirical method for injection height from wild-land fires

outperforms existing methods by 15-30% (fraction of good predictions)

can be recommended for 3-D models that possess fire intensity information

• Complementing the existing emission databases, the global map of “climatologic” injection profile from wild-land fires

accounts for diurnal/seasonal variation

targets all fire-related applications