Transcript Fire Detection by satellite for fire control in - GOFC/GOLD-Fire

FIRE DETECTION BY
SATELLITE FOR FIRE
CONTROL IN MONGOLIA
Global Geostationary Fire Monitoring Workshop on 23-25 March, 2004
Darmstadt Germany
S.Tuya, K.Kajiwara, Y.Honda
CEReS of Chiba University & JAXA
1
Presentation outline
1.
2.
3.
4.
5.
Introduction
Goal & Objective
Study area & Data
Methodology
Results & Conclusion
2
INTRODUCTION
Forests and grasslands play an important role in the economy development
of the country. Forest cover is 8.1% and grassland cover is 70% of all
territory. In an average year occur the 50-60 forest fires and 80-100 steppe
fires. Since 1987 the Information and Computer Center of Ministry for
Nature and the Environment daily receives the AVHRR (Advances Very
High Resolution Radiometer) data from NOAA meteorological satellite,
which can be used to detect and monitor the forest and steppe fire over
whole territory of Mongolia. Fire monitoring in Mongolia is essential for
all kind of land-use planning and forest management. To detect and
monitor wildfires and to support fire management activities with real time
information on fire events is of high priority. To meet this objective, a fire
detection methodology based an NOAA AVHRR data has been developed at
the Information Computer Center. To improve the fire monitoring, a
second processing chain was set up using the WFW software in 2000.
3
GOAL
We need an ability to real time quickly
detect, locate and respond fires using
satellite data.
To reducing their ecological and
economical damages in the country.
4
OBJECTIVE



Determine the location of active fires using
satellite data
Determine the total burned area
Compare the suitability of different satellite
data for fire monitoring and assessment
5
Study area
GEOGRAPHICAL LOCATION. (41O35'N - 52O09'N and 87O44'E
- 119O56'E) and bounded by Russia and China.
TOTAL TERRITORY: 1,566,500 sq. km.
POPULATION: more than 2.7 million persons.
CAPITAL: Ulaanbaatar. Its population is more than 650,000
persons.
BASIC OF MONGOLIAN ECONOMY: livestock farming.
CLIMATE: continental
6
Mean annual NDVI 1982-2000
Mongolia
Russia
Mongolia
China
7
Forest
8
Grassland
9
DATA USE

Satellite data
NOAA-AVHRR 1km ( 4,7 April 2000, 5.May, 2003)
LANDSAT-TM
( 7. April, 2000)
MODIS-TERRA 1km, 500m, 250 m
ADEOS-II, GLI
(5.May, 2003 26. May, 2003)
1km (5.May, 2003 26. May, 2003)
Ancillary data
* Rivers, lakes, road and political boundaries
10
1. METHODOLOGY at ICC

Fire detection methodology in practice at ICC
I. Active fire:
a) T3 > 45oC
b) R1(or R2) = 6 – 12
II. Burnt area:
a) T3 > 35 - 45oC
b) R1(or R2) = 3 – 6
CH3- Temperature of NOAA-AVHRR channel 3.
CH1 or CH2 – reflectance of NOAA-AVHRR channel 1 or 2
CH4 or 5-Temperature of NOAA-AVHRR channel 4 or 5 are used for cloud masking.
In the final image product, active fires are identified by visual
interpretation and plausibility check.
ICC -Information Computer Centre in Mongolia
11
Daily Fire Map and Hot Spots From NOAAAVHRR Data Using Traditional method at
ICC
night
afternoon
Trends of steppe fire over Dornod and Khentii aimags (North Eastern part of Mongolia). 07.April.2000
12
Total burned area map of
Mongolia 2000
13
Fire Frequency Map of Mongolia
1996-2001
14
2. METHODOLOGY at JRC

Fire detection methodology WFW in JRC
I.Threshold Fire Test:
a selection of pixels that could potentially contain fires, and thus be called "fire
pixels".
A pixel is selected as a potential fire if:
Tb(3) > 311K and Tb(3) - Tb(4) > 8K
II. Contextual Fire Test: a confirmation of the fire pixel classification by comparing the pixel with its
immediate neighborhood.
A potential fire is then confirmed if:
[Tb(3) - Tb(4)] > Tb(34)bg + 2 s(34)bg and Tb(3) > Tb(3)bg + 2 s(3)bg + 3K.
Tb(i) represents the brightness temperature of channel i (i = 3, 4, 5).
Tb(3)bg = Mean T b(3) in the background.
s(3)bg = Standard deviation of T b(3) in the background.
Tb(34)bg = Mean value of [T b(3) - Tb(4)] of pixels in the background.
s(34) bg = Standard deviation of [Tb(3) - Tb(4)] of pixels in the background
15
Daily Fire Map and Hot Spots From NOAAAVHRR Data Using WFW system at JRC
Daily, global fire maps are built up at the JRC in Italy from this regional
data by automatically sharing regional fire maps over the internet. Global
fire information is then available on-line, in near real-time.
16
Daily Fire Map and Hot Spots From NOAAAVHRR Data Using WFW system at JRC
17
Fire Frequency Map of Mongolia
for the period of March-May 2000 using
WFW and Arc View
18
Comparison of NOAA-AVHRR data and
Landsat-TM data for fire monitoring
Burned Area Map of Dornod Region ( 2000.04.10 )
Example of Burned area
LANDSAT-TM
Example of Burned area
NOAA-14
19
Comparison of NOAA-AVHRR data and
Landsat-TM data for fire monitoring
Active Fire of Dornod Region ( 2000.04.10 )
Landsat-TM Active fire
NOAA-14 Active Fire
20
3. METHODOLOGY USING
THRESHOLD VALUE
Fire detection threshold for potential fire pixels
1.For NOAA-AVHRR

CH (3) > 311K and CH (3) - CH (4) > 8K
CH (2) < 0.20
2. For MODIS-TERRA
CH21>360K
CH31>320K and CH21- CH31>20K
3. For ADEOS-II, GLI
CH30>330K
21
Fire map using NOAA-AVHRR 1km
Steppe fire in 05.May 2003,
Northern Mongolia
- Red points is Hot spots
- Dark blue is burned area ( 7953.sq.km2 )
22
Fire map using MODIS-TERRA 1km
Steppe fire in 05.May 2003,
Northern Mongolia
- Red points is Hot spots
- Dark green is burned area (7521.sq.km2 )
23
Fire map using ADEOS-II, GLI 1km
Steppe fire in 05.May 2003,
Northern Mongolia
- Red points is Hot spots
- Dark brown is burned area (7838.sq.km2 )
24
Burned area map using MODISTERRA- 1km, 500m,250m
I
I
I
II
II
II
1km
I- 3633.sq.km2
II- 3888.sq.km2
500m
I- 3603.sq.km2
II- 3752.sq.km2
250m
I- 3626.sq.km2
II- 3734.sq.km2
Burned area of steppe fire on 05.May 2003
Dornod region in the Northern Mongolia
25
Burnt area maps of Mongolia for the
spring period of 2003
ADEOS-II GLI 26.05.2003
ADEOS-II GLI 05.05.2003
26
Operative service for end users
Government
State Emergency
Committee
ICC
Internet
X25 protocol
Inter
Organization
Network
Other organizations
Ministry
Nature and
Environment
Civil Defence
Office
Fire Fighting
Office
Fire Fighting
Office in
aimags
Aimag
Administrative
Staff
Hydrometeorological
Center in Aimags
27
Result

The new processing chain can detect fires and burnt area
automatically.

To a small extend, both methodologies confuses active fires
with very hot land surfaces.

The major disadvantage of the WFW system compared to the
local method is, that real time observation is not possible.
Necessary ephemeris data for the fire processing is available at
the earliest one day after the image reception.
28
Result

AVHRR has two major advantages for fire monitoring. First,
its observation covers the entire region everyday at a
moderate resolution 1.1 km, which is critical for operational
fire monitoring. Second, it has wide spectral coverage. But
AVHRR images give the general locations and size of
burned area of current fires.

Used ADEOS-II GLI and MODIS-TERRA images can
progressed the accuracy for calculating the burned area and
hotspots. The estimation of burned area using new sensors
gives details information on burnt areas for the
environmental assessments of damage. A totally 4,946,99
thousand ha grassland was burnt on 26.May, 2003
29
Conclusion



Fire monitoring methodology improvement in
Mongolia is essential for all kind of land-use
planning and forest management.
A large data base can be achieved over the entire fire
season for further evaluations and research activities.
The WFW approach is able to cover large areas (e.g.
entire NOAA scene), where as the traditional
method concentrates on specific regions of interests.
30
Conclusion


To a small extend, both methodologies confuses
active fires with very hot land surfaces.
Understand the impacts of global environmental
change related on the major individual influences
of local and regional climate change. Therefore
wild fires in the Mongolia are one of factors of
local area individuals to great global change.
Consequently, I think fire monitoring in Mongolia
is one part of local activities contribution to global
change research.
31
Global
Regional Asia
Local
32