Cross-Correlation between World Fire Atlas and Environmental Classifications

Diane Defrenne, SERCOOlivier Arino, ESA

Objective of the Cross Correlation with Environmental Classifications:

• Develop a set of Fire Behaviour Rules.• Develop a set of Emission prediction factor.• Increasing the WFA confidence.

CLASSIFICATIONS USED TO CROSS CORRELATE WITH THE WFA.

CROSS-CORRELATION WITH METEOROLOGICAL DATA.

CROSS-CORRELATION WITH ATHMOSPHERIC DATA.

FURTHER DEVELOPMENTS.

Classification available:• Meteorological data ECMWF 40 Years Re-Analysis, monthly fields (resolution

2.5°).• Vegetation Classification GLC2000 (30 sec°).• Atmospheric chemistry observation TEMIS, monthly fields (0.25°).

ENVIRONMENTAL CLASSIFICATIONS

Classification Criteria:• Temporal coverage about 5 years between November 1995 and May 2005.• Global Geographical coverage almost complete.

System used to generate the correlations:• Use of the World Fire Atlas Tool that permits the discrimination of the WFA

data in time and space.• Microsoft Visual Basic 6.0• Free GIS object library (www.inovagis.org).

Preliminary processing on the input files:• Generate from each classification file in various format (netcdf, grd, …) a raster file on

Plate Carré.• Manage the various resolution of each classification.

OBJECTIVE Fire Behaviour Rules.

Input: WFA

ECMWF Temperature from ECMWF ERA 40

ECMWF Precipitation from ECMWF ERA 40

Vegetation classification from GLC2000

Output: Hot spots detected from WFA.

Mean of monthly temperature over the region.

Mean of monthly cumulated precipitation over the region.

Classification of Vegetation index from GLC2000 (% of surface use by each vegetation).

Region: South America, Africa, Siberia.

WFA and METEOROLOGICAL CLASSIFICATION

Methodology

•Difference between local climate (Temperature and precipitation).

•Vegetation classification for each tile.

•Correlation between number of fire by month and meteorological parameters (the precipitation and the temperature).

•Extract some fire behaviour rules.

AMAZON

F(time): Time rule, F(T): Temperature rule , F(P): Precipitation rule

AFRICA

Fires depends of Vegetation and Precipitation.

Fires depends of temperature.

Cross-correlation between NO2 concentration and hot spots detected First Results:

Comparison Between NO2 Monthly mean Tropospheric column and the number of fires over Zambia

0

100

200

300

400

500

600

Apr-96

Jul-96

Oct-96

Jan-97

Apr-97

Jul-97

Oct-97

Jan-98

Apr-98

Jul-98

Oct-98

Jan-99

Apr-99

Jul-99

Oct-99

Jan-00

Apr-00

Jul-00

Oct-00

Jan-01

Apr-01

Jul-01

Oct-01

Jan-02

Apr-02

Jul-02

Oct-02

Jan-03

Apr-03

Jul-03

Oct-03

Jan-04

Apr-04

Jul-04

Oct-04

Jan-05

Apr-05

number of fires

NO2 Tropospheric columnmean (10e15 molec/cm²)

Comparison Between NO2 Monthly mean Tropospheric column and the number of fires over MOZAMBIQUE

0

100

200

300

400

500

600

700

Apr-96

Jul-96O

ct-96

Jan-97

Apr-97

Jul-97

Oct-97

Jan-98

Apr-98

Jul-98

Oct-98

Jan-99A

pr-99

Jul-99

Oct-99

Jan-00

Apr-00

Jul-00

Oct-00

Jan-01

Apr-01

Jul-01O

ct-01

Jan-02

Apr-02

Jul-02

Oct-02

Jan-03

Apr-03

Jul-03

Oct-03

Jan-04A

pr-04

Jul-04

Oct-04

Jan-05

Apr-05

number of fires

NO2 Tropospheric columnmean (10e15 molec/cm²)

Comparison Between NO2 Monthly mean Tropospheric column and the number of fires over ALGERIA

0

50

100

150

200

250

Jul-96

Oct-96

Jan-97

Apr-97

Jul-97

Oct-97

Jan-98

Apr-98

Jul-98

Oct-98

Jan-99

Apr-99

Jul-99

Oct-99

Jan-00

Apr-00

Jul-00

Oct-00

Jan-01

Apr-01

Jul-01

Oct-01

Jan-02

Apr-02

Jul-02

Oct-02

Jan-03

Apr-03

Jul-03

Oct-03

Jan-04

Apr-04

Jul-04

Oct-04

Jan-05

Apr-05

number of fires

NO2 Troposphericcolumn mean(10e15 molec/cm²)

Comparison Between NO2 Monthly mean Tropospheric column and the number of fires over PORTUGAL

0

20

40

60

80

100

120

140

160

180

200

Ap

r-96

Jul-9

6

Oct-9

6

Jan

-97

Ap

r-97

Jul-9

7

Oct-9

7

Jan

-98

Ap

r-98

Jul-9

8

Oct-9

8

Jan

-99

Ap

r-99

Jul-9

9

Oct-9

9

Jan

-00

Ap

r-00

Jul-0

0

Oct-0

0

Jan

-01

Ap

r-01

Jul-0

1

Oct-0

1

Jan

-02

Ap

r-02

Jul-0

2

Oct-0

2

Jan

-03

Ap

r-03

Jul-0

3

Oct-0

3

Jan

-04

Ap

r-04

Jul-0

4

Oct-0

4

Jan

-05

Ap

r-05

number of fires

NO2 Tropospheric columnmean (10e15 molec/cm²)

Predicting the No2 Concentration from the hot spots detected by the satellite:

METHODOLOGY:• Divide the region of interest in small tile (5°x5°).• Group the tiles having a similar vegetation following the GLC2000

index.• Select inside each zone a central region of study with enough fires.• Perform a cross-correlation between the number of fires by month

and the No2 mean concentration in a month from march 1996 to may 2005 (without Jan 1998 and November 2003 because data from TEMIS are not available or complete) for each region selected.

• Perform a linear regression over each region.• Try to generate an emission prediction factor for each region and to

find the natural NO2 emission cycle.

ASSUMPTION:• GLC2000 is a fixed vegetation classification the vegetation has

globally changed between 1996 and 2005.• TEMIS NO2 tropospheric concentration data available from April

1996 to date and WFA data available from July 1996 to date.• The emission predictor factor is considering linear and doesn’t take

in consideration the natural NO2 emission cycle.

Using satellite data to better understand Ozone budget. N.Savage

REPARTITION OVER AFRICA:

RegionType of vegetationGLC2000 index

% of the surface covered by this vegetation

Dimension of each area

Total number of fires

Region_1 1 67.66 % 8 x (5°x5°) 9389

Region_22 37.37 %

6 x (5°x5°) 25579 3 33.99 %

Region_33 33.51 %

7 x (5°x5°) 34793 12 29.26 %

Region_412 45.86 %

5 x (5°x5°) 7303 13 35.08 %

GLC2000 Index Vegetation:12 Shrub Cover, closed-open, deciduous (with or without sparse tree layer)13 Herbaceous Cover, closed-open16 Cultivated and managed areas

1 Tree Cover, broadleaved, evergreen2 Tree Cover, broadleaved, deciduous, closed3 Tree Cover, broadleaved, deciduous, open

Discriminate region by GLC2000 Classification

Emission prediction factor

1 C=152,09+31.49(HS)

2 C=88.07+26.46(HS)

NOT ENOUGH HOT SPOTS !!

3 C=?

4 C=163.36+24.90(HS)

5 C=159.84+65.046(HS)

FURTHER DEVELOPMENTS

Behaviour Model:• Fully define F(time), F(precipitation) and F(temperature) from

Cross-correlation over some significant regions• Use ERA 40 4 time daily data set to refine in time the Behaviour

model.

Emission prediction factor:• Results are very encouraging.• Complete the study to define the NO2 emission prediction

factor and seasonal cycle.

Vegetation Classification:• Cross-correlate the WFA data with the classification from

GLOBCOVER for 2005 (available by the end of 2007).

Thank You!