Slide: 1Version 1.1, 30 June 2004
APPLICATIONS OF
METEOSAT SECOND
GENERATION (MSG)
METEOROLOGICAL USE OF
THE SEVIRI IR3.9 CHANNEL
Author: Jochen Kerkmann
kerkmann@eumetsat.de
Contributors: D. Rosenfeld (HUJ), H.J. Lutz (EUM)
J. Prieto (EUM), M. König (EUM)
Slide: 2Version 1.1, 30 June 2004
IR3.9: WEIGHTING FUNCTIONS
Slide: 3Version 1.1, 30 June 2004
IR3.9 Weighting Function
Max. signal from the
surface, but IR3.9
recieves substantial
contribution from
the upper
troposphere
Slide: 4Version 1.1, 30 June 2004
IR3.9 Weighting Function
The contribution
from the upper
troposphere is larger
in cold, winter
conditions
Slide: 5Version 1.1, 30 June 2004
IR3.9 Weighting Function
The weighting
function for tropical
areas (Nadir
viewing) is similar
to the one for Mid
Latitudes Summer
(60° viewing)
Slide: 6Version 1.1, 30 June 2004
IR3.9 Weighting Function
The contributions
from the mid and
upper troposphere
become larger with
increasing satellite
zenith angle ("limb
cooling")
Slide: 7Version 1.1, 30 June 2004
IR3.9: CO2 ABSORPTION
Slide: 8Version 1.1, 30 June 2004
The IR3.9 channel is a window channel
but close to the CO2 absorption band at 4-5 microns
IR3.9
Energy
spectrum
Slide: 9Version 1.1, 30 June 2004
Sun radiation Earth radiation
SEVIRI CHANNELS: IR3.9 µµµµm
Wavelength (µµµµm)
O3
CO2
H2O
Watt
per m2
and µµµµm
Slide: 10Version 1.1, 30 June 2004
I. Surface temperature and lapse rate in the lower troposphere (∆T_CO2 is
large for hot desert surfaces during daytime)
II. Height of the cloud (∆T_CO2 is small for high clouds)
III. Satellite viewing angle (so called "limb cooling" effect, ∆T_CO2 is large for
large satellite viewing angles)
CO2 Effect on Brightness
Temperature of Channel IR3.9
The "cooling" effect (∆T_CO2) of the CO2 absorption on channel IR3.9
depends on:
Slide: 11Version 1.1, 30 June 2004
IR3.9
IR10.8
MSG-1, 04 March 2004, 00:00 UTC
Brightness Temperatures
Differences (BTD) for Cloud-free
Ocean Targets
283 K
287 K
Sub-tropical, dry atmosphere,
Medium sat. viewing angle:
IR3.9 - IR10.8 = -4 K
294 K
296 K
Tropical, moist atmosphere,
Small sat. viewing angle:
IR3.9 - IR10.8 = -2K
285 K
Sub-tropical, dry atmosphere,
Large sat. viewing angle:
IR3.9 - IR10.8 = -7 K
292 K
Slide: 12Version 1.1, 30 June 2004
1 = Limb cooling in
the IR3.9 channel;
little cooling at
western limb because
of sat. position at
10°W.
MSG-1
24 April 2003
00:00 UTC
Channel 04
IR3.9
1
1
1
1
Toggle this
and the next
slide !
Slide: 13Version 1.1, 30 June 2004
MSG-1
24 April 2003
00:00 UTC
Channel 09
(IR10.8)
1
1
1
1
1 = Limb cooling in
the IR3.9 channel;
little cooling at
western limb because
of sat. position at
10°W.
Slide: 14Version 1.1, 30 June 2004
MSG-1
24 April 2003
00:00 UTC
Difference Image
IR10.8 - IR3.9
1
1 = Limb cooling in the
IR3.9 channel; little
cooling at western limb
because of sat. position
at 10°W.
1
1
1
Larger differences
in cloud-free limb
areas (lower IR3.9
brightness
temperatures)
Slide: 15Version 1.1, 30 June 2004
IR3.9: SOLAR AND THERMAL
CONTRIBUTION
Slide: 16Version 1.1, 30 June 2004
SEVIRI
CHANNELS:
IR3.9 µµµµm
Planck blackbody radiance curves
Signal in IR3.9 channel comes from
reflected solar and emitted thermal
radiation !
Consequence for Planck relation
between radiance and temperature:
during day-time, temperature is not
representative of any in situ
temperature (see next slide) !
Slide: 17Version 1.1, 30 June 2004
Schematic: Blackbody
Radiation for T=300K
(actual scene temperature)
Wavelength
3.9
IR3.9 Radiance
at 300K
IR3.9 Radiance
Measurement:
300K + reflected
sunlight
Schematic: Blackbody
Radiation for T=350K
(satellite measured scene
temperature)
Radiance
Intensity
Slide: 18Version 1.1, 30 June 2004
3.9 µµµµm Imagery Presentation
• In the GOES IR3.9 Channel Tutorial, the 3.9 um imagery is
presented in terms of energy vs grey scale, i.e. cold clouds
appear dark, warm surfaces appear light-to-bright
• In order to better compare with the other IR channels, in this
presentation the display as infrared image is preferred, i.e. cold
clouds appear bright, warm surfaces appear dark
Should IR3.9 be displayed as visible or infrared image ?
Slide: 19Version 1.1, 30 June 2004
MSG-1, 24 April 2003, 00:00 UTC
Channel 04 (3.9 µm)
During night-time
channel 04 has only
the emitted thermal
contribution
Animation 1/6 Solar & Thermal
Contributions of
Channel 04
Slide: 20Version 1.1, 30 June 2004
MSG-1, 24 April 2003, 03:00 UTC
Channel 04 (3.9 µm)
Animation 2/6 Solar & Thermal
Contributions of
Channel 04
Slide: 21Version 1.1, 30 June 2004
MSG-1, 24 April 2003, 06:00 UTC
Channel 04 (3.9 µm)
Animation 3/6 Solar & Thermal
Contributions of
Channel 04
Slide: 22Version 1.1, 30 June 2004
MSG-1, 24 April 2003, 09:00 UTC
Channel 04 (3.9 µm)
Animation 4/6 Solar & Thermal
Contributions of
Channel 04
During day-time this
channel has a thermal
and a solar contribution.
Therefore, applications
and algorithms are
different for night- and
day-time !
Slide: 23Version 1.1, 30 June 2004
MSG-1, 24 April 2003, 12:00 UTC
Channel 04 (3.9 µm)
Animation 5/6 Solar & Thermal
Contributions of
Channel 04
Slide: 24Version 1.1, 30 June 2004
MSG-1, 24 April 2003, 15:00 UTC
Channel 04 (3.9 µm)
Animation 6/6 Solar & Thermal
Contributions of
Channel 04
1
1 = sunglint
(see also 03:00 UTC)
Slide: 25Version 1.1, 30 June 2004
MSG-1, 24 April 2003, 00:00 UTC, Channel 04
Cold
high-level ice clouds
cold snow surfaces
mid-level clouds
low-level water clouds
land surfaces
ocean, sea, lakes
Warm
IR 3.9 µµµµm
Nighttime
Only thermal contribution:
clouds are brighter (colder)
than ocean surfaces
Slide: 26Version 1.1, 30 June 2004
IR 3.9 µµµµm
Nighttime
Cold
high-level ice clouds
mid-level clouds
low-level water clouds
land surfaces
ocean, sea, lakes
Warm
MSG-1, 24 April 2003, 00:00 UTC, Channel 04
Only thermal contribution:
clouds are brighter (colder)
than ocean surfaces
Slide: 27Version 1.1, 30 June 2004
Low reflectance / Cold
high-level ice clouds
snow surfaces
ocean, sea
cold land surfaces
warm land surfaces
low-level water clouds
hot land surfaces
High reflectance / Warm
IR 3.9 µµµµm
Daytime
MSG-1, 24 Feb 2003, 11:00 UTC, Channel 04
Thermal and solar
contribution: low clouds are
darker than ocean surfaces
Slide: 28Version 1.1, 30 June 2004
Low reflectance / Cold
high-level ice clouds
ocean, lakes
low-level water clouds
hot land surfaces
fires, sunglint areas
High reflectance / Warm
IR 3.9 µµµµm
Daytime
MSG-1, 07 July 2003, 11:00 UTC, Channel 04
Thermal and solar
contribution: low clouds are
darker than ocean surfaces
Slide: 29Version 1.1, 30 June 2004
IR3.9: INFLUENCE OF SURFACE
EMISSIVITY
Slide: 30Version 1.1, 30 June 2004
SEVIRI
CHANNELS:
IR3.9 µµµµm
Emissivity as a function of wavelength
and surface type: figure 01
• Emissivity more variable
near 3.9 µm
• Sandy areas appear 5-10 K
cooler at IR3.9 than at
IR10.8 (at night, dry
atmosphere)
• Different appearance of land
surfaces during daytime,
depending on surface type
IR3.9 IR10.8
Dry sand: 0.8 0.95
Slide: 31Version 1.1, 30 June 2004
SEVIRI CHANNELS: IR3.9 µµµµm
Emissivity as a function of wavelength and surface type: figure 02
Slide: 32Version 1.1, 30 June 2004
MSG-1, 24 April 2003, 00:00 UTC, difference 3.9 µm - 10.8 µm [K]
Sandy areas appear 5 - 10 K cooler at IR3.9 than at IR10.8
SEVIRI CHANNELS: IR3.9 µµµµm
Slide: 33Version 1.1, 30 June 2004
IR3.9: APPLICATIONS
Slide: 34Version 1.1, 30 June 2004
Detection of low clouds and fog [day and night]
Detection of thin Cirrus [day and night] and multilayer
clouds [day]
Cloud phase & particle size [day and night]
Sea and land surface temperature [night]
Detection of forest fires [day and night]
Urban heat island [night]
Super-cooled clouds [day and night]
Cloud top structures (overshooting tops) [day]
Sunglint [day]
METEOROLOGICAL USE OF
THE SEVIRI IR3.9 CHANNEL
Similar channel on AVHRR, ATSR and MODIS
Slide: 35Version 1.1, 30 June 2004
IR3.9: DETECTION OF
FOG / LOW STRATUS
(DAY AND NIGHT-TIME)
Slide: 36Version 1.1, 30 June 2004
Fog and Low Stratus/Sc
• The identification of fog and stratus at night is the main
application of the IR3.9 channel
• The technique is based on the principle that the emissivity of
water cloud at 3.9 µm is less than at 10.8 µm: IR3.9 shows more
reflection of cold atmosphere above. This is not the case for
cloud free surfaces (except sandy desert surfaces).
• Evolution of night-time fog and low-level stratus clouds is
easily observed by viewing the animation
Slide: 37Version 1.1, 30 June 2004
Fog and Low Stratus/Sc
Emissivity of water cloud at 3.9 µm is less than at 10.8 µm:
IR3.9 shows more reflection of cold atmosphere above
Slide: 38Version 1.1, 30 June 2004
1= low-level fog or stratus
2= cold clear ground
3 = warm clear ground (mountains)
4 = thin, high-level clouds
MSG-1, 09 November 2003, 03:15 UTC
IR3.9
Fog at night visible in
IR3.9 - IR10.8
brightness temperature
difference images
IR10.8 IR3.9 - IR10.8
111
22 2
444
3 3
3
Slide: 39Version 1.1, 30 June 2004
MSG-1
14 July 2003
03:00 UTC
Difference Image
IR3.9 - IR10.8
2
1
3
1= low-level fog or stratus
2= clear ground
3= high-level clouds
Fog at night
visible in
IR3.9 - IR10.8
brightness temperature
difference images
Animation 1/3
Slide: 40Version 1.1, 30 June 2004
MSG-1
14 July 2003
05:00 UTC
Difference Image
IR3.9 - IR10.8
2
1= low-level fog or stratus
2= clear ground
3= high-level clouds
1
3
Fog at dawn/dusk
not visible in
IR3.9 - IR10.8
brightness temperature
difference images
Animation 2/3
Slide: 41Version 1.1, 30 June 2004
MSG-1
14 July 2003
06:00 UTC
Difference Image
IR3.9 - IR10.8
1
2
3
1= low-level fog or stratus
2= clear ground
3= high-level clouds
Fog at day
visible in
IR3.9 - IR10.8
brightness temperature
difference images
Animation 3/3
Slide: 42Version 1.1, 30 June 2004
IR3.9
IR10.8
note: IR10.8 imagery is already strongly
enhanced in this example
Fog at day visible in
IR3.9 images
MSG-1
24 February 2003
12:00 UTC
1= low-level fog or stratus
BT(IR3.9)=290.1 K
BT(IR10.8)=267.0 K
2= cloud-free ocean
BT(IR3.9)=267.3 K
BT(IR10.8)=270.6 K
1
2
1
2
Slide: 43Version 1.1, 30 June 2004
Threshold (in K) for the IR10.8IR3.9
brightness temperature
difference to discriminate between
fog/low stratus and cloud free
vegetated areas as a function of total
water vapour content for a satellite
zenith angle of 48 degrees.
(from Nowcasting SAF,
Météo-France)
Detection of Fog &
Low Stratus/Sc at
Night-timeFog/low stratus
Cloud free
vegetated areas
Slide: 44Version 1.1, 30 June 2004
Detection of Fog &
Low Stratus/Sc at
Night-time
Fog/low
stratus
Cloud free
arid areas
Threshold (in K) for the IR10.8IR3.9
brightness temperature
difference to discriminate between
fog/low stratus and cloud free arid
areas as a function of total water
vapour content for a satellite zenith
angle of 48 degrees.
(from Nowcasting SAF,
Météo-France)
Slide: 45Version 1.1, 30 June 2004
Detection of Fog &
Low Stratus/Sc at
Night-time
Fog/low stratus
Cloud free
oceanic areas
Threshold (in K) for the IR10.8IR3.9
brightness temperature
difference to discriminate between
fog/low stratus and cloud free
oceanic areas as a function of total
water vapour content for a satellite
zenith angle of 48 degrees.
(from Nowcasting SAF,
Météo-France)
Slide: 46Version 1.1, 30 June 2004
Same as previous figure, but for the
IR12.0-IR3.9 brightness temperature
difference.
This test is more efficient over ocean
due to a higher contrast between
cloud free areas and low clouds (see
next slide).
(from Nowcasting SAF,
Météo-France)
Detection of Fog &
Low Stratus/Sc at
Night-time
Fog/low stratus
Cloud free
oceanic areas
Slide: 47Version 1.1, 30 June 2004
Night-time IR10.8-IR3.9 (left) and IR12.0-IR3.9 (right) brightness temperature
difference for cloud free oceanic targets (blue +) and low clouds (black diamond).
(from Nowcasting SAF, Météo-France)
Slide: 48Version 1.1, 30 June 2004
Night-time IR10.8-IR3.9
brightness temperature
difference for cloud free
oceanic targets (blue +)
and low clouds (black
diamond) as a function of
satellite zenith angle.
(from Nowcasting SAF,
Météo-France)
Slide: 49Version 1.1, 30 June 2004
IR3.9: CLOUD PHASE
AND PARTICLE SIZE
(MAINLY DAY-TIME)
Slide: 50Version 1.1, 30 June 2004
Reflection of Solar Radiation at IR3.9
• Reflection at IR3.9 is sensitive to
cloud phase and very sensitive to
particle size
• Higher reflection from water
droplets than from ice particles
• During daytime, clouds with small
water droplets (St, Sc) are much
darker than ice clouds
• Marine Sc (large water droplets) is
darker than Sc over land
Slide: 51Version 1.1, 30 June 2004
MSG-1, 07 July 2003, 11:00 UTC, Channel 04
Due to the high reflection
from water droplets at
IR3.9, low-level water
clouds are much darker
than high-level ice clouds
(during day-time)
IR3.9: Cloud
Phase
Slide: 52Version 1.1, 30 June 2004
MSG-1, 20 May 2003, 13:30 UTC
IR3.9: Cloud Particle Size
Channel 04 (IR3.9) Channel 09 (IR10.8)
IR3.9 shows much more cloud top structures than
IR10.8 (very sensitive to particle size)
1 1
1= ice clouds with very small particles
2= ice clouds with small particles
3= ice clouds with large ice particles
31 1
1
3
3
1
1
1
3
3
3
2 2
Slide: 53Version 1.1, 30 June 2004
IR3.9: Cloud Particle Size
The IR3.9 - IR10.8 brightness temperature
difference is very large (around +50 to +60
K) for cold, high-level clouds with small ice
particles (see next slide). This can be
exploited in RGB composites to highlight the
most severe parts of thunderstorms (here
shown in yellow, see also RGB tutorial).
MSG-1
20 May 2003
13:30 UTC
RGB Composite
R = WV6.2 - WV7.3
G = IR3.9 - IR10.8
B = NIR1.6 - VIS0.6
Slide: 54Version 1.1, 30 June 2004
IR3.9 - IR10.8 Brightness Temperature
Differences for Opaque Clouds
IR3.9 cloud at cloud at cloud at
Albedo 200 K 250 K 300 K
5 90 43 12
4 85 38 10
3 78 33 8
2 70 26 5
1 57 16 3
0 0 0 0
IR3.9 - IR10.8 brightness temperature difference (in K) for
different temperatures of the cloud assuming little humidity above
the cloud.
For cold clouds, the IR3.9 - IR10.8 BTD is very sensitive to albedo
(i.e. cloud particle size)
Slide: 55Version 1.1, 30 June 2004
IR3.9: DETECTION OF THIN
CIRRUS CLOUDS
(DAY AND NIGHT-TIME)
Slide: 56Version 1.1, 30 June 2004
Thin Cirrus Clouds
• Cirrus clouds are more transparent at 3.9 µm than at 10.8 µm
because of the stronger response at 3.9 µm to the warm radiation
from below
• In addition, thin cirrus is often patchy and only partially fills a
FOV (Field of View), further enhancing response at 3.9 µm
• Therefore, in 3.9-10.8 difference images, thin Cirrus can be
easily detected (large positive difference)
Slide: 57Version 1.1, 30 June 2004
Ice clouds are more transparent
(less absorption) at IR3.9 than at
IR10.8.
==> IR3.9 much warmer than
IR10.8 for thin ice (Cirrus)
clouds.
Thin Cirrus Clouds
Slide: 58Version 1.1, 30 June 2004
IR3.9 IR10.8
IR3.9-IR10.8 Cirrus clouds
optically thin (transparent)
IR3.9 = 258 K; IR10.8 = 240 K
IR3.9 - IR10.8 = +18 K
MSG-1, 4 March 2004, 00:00 UTC
Cirrus clouds
optically thick
IR3.9 = 222 K; IR10.8 = 219 K
IR3.9 - IR10.8 = +3 K
Slide: 59Version 1.1, 30 June 2004
Thin Cirrus Clouds (Night-time)
Channel IR10.8 difference IR3.9 - IR10.8
MSG-1, 14 July 2003, 02:00 UTC
Slide: 60Version 1.1, 30 June 2004
Thin Cirrus Clouds (Day-time)
Channel VIS0.6 Channel IR10.8 diff. IR3.9 - IR10.8
MSG-1, 25 June 2003, 10:00 UTC
Slide: 61Version 1.1, 30 June 2004
IR3.9: DETECTION OF
MULTI-LAYER CLOUDS
(DAY-TIME)
Slide: 62Version 1.1, 30 June 2004
Multi-layer Clouds
• During daytime, low-level water clouds, with their higher
reflectivity, appear much darker ("warmer") than high-level
ice clouds
• Low-level water clouds can easily be detected at 3.9 um,
even below Cirrus clouds
Slide: 63Version 1.1, 30 June 2004
Multi-layer
Clouds
MSG-1
3 June 2003
12:00 UTC
Channel 04 (3.9 µm)
1= low-level water clouds
2= high-level ice clouds
1
2
2
2
1
1
Slide: 64Version 1.1, 30 June 2004
Multi-layer
Clouds
MSG-1
25 June 2003
10:00 UTC
Channel 04 (3.9 µm)
1= low-level water clouds
2= high-level ice clouds
1
2
1
2
21
Slide: 65Version 1.1, 30 June 2004
Multi-layer
Clouds
MSG-1
25 June 2003
10:00 UTC
Channel 04 (3.9 µm)
1= low-level water clouds
2= high-level ice clouds
2
2
2
1
1
1
Slide: 66Version 1.1, 30 June 2004
IR3.9: DETECTION OF
SUPERCOOLED CLOUDS
(DAY AND NIGHT-TIME)
Slide: 67Version 1.1, 30 June 2004
• Cloud tops consisting of supercooled water droplets may be
located by using:
– the 3.9 µm imagery to identify phase: supercooled water
clouds have high reflection and appear dark.
– the 10.8 µm imagery to determine cloud top temperature:
supercooled water clouds have a top temperature between
0°C and -40°C.
Supercooled Clouds (Day-time)
Slide: 68Version 1.1, 30 June 2004
MSG-1
24 February 2003
11:00 UTC
Channel 04 (3.9 µm)
Supercooled
Clouds
(Day-time)
1= low-level water clouds
(top temp. -5°C)
2= mid-level water clouds
(top temp. -20°C)
(suppercooled cloud)
3= snow
1
1
1
2
1
3
Slide: 69Version 1.1, 30 June 2004
MSG-1
24 February 2003
11:00 UTC
Channel 09 (10.8 µm)
Supercooled
Clouds
(Day-time)
1= low-level water clouds
(top temp. -5°C)
2= mid-level water clouds
(top temp. -20°C)
(suppercooled cloud)
3= snow
1
2
1
3
1
1
Slide: 70Version 1.1, 30 June 2004
MSG-1
24 February 2003
11:00 UTC
Channel 02 (0.8 µm)
Supercooled
Clouds
(Day-time)
1= low-level water clouds
(top temp. -5°C)
2= mid-level water clouds
(top temp. -20°C)
(suppercooled cloud)
3= snow
1
2
1
3
1
1
Slide: 71Version 1.1, 30 June 2004
MSG-1
24 February 2003
11:00 UTC
RGB Composite
R = VIS0.8
G = IR3.9r
B = IR10.8
Supercooled
Clouds
(Day-time)
In RGB 02-04r-09
composites, supercooled
clouds appear in greenishyellowish
colours
(greenish for thin clouds;
yellowish for thick
clouds)
1
2
1
3
1
1
Slide: 72Version 1.1, 30 June 2004
Supercooled Clouds (Nighttime)
• During the night-time hours, water clouds can also be
distinguished from ice clouds by using the "fog product"
(difference IR3.9 - IR10.8).
• Similar to the day-time application, the "fog product" and the
10.8 µm imagery can be used together to locate cloud tops
consisting of supercooled water at night.
Slide: 73Version 1.1, 30 June 2004
IR3.9: SUN GLINT
(DAY-TIME)
Slide: 74Version 1.1, 30 June 2004
Sun Glint
• There is very strong reflection of solar radiation at 3.9 µm
• This causes sun glint to be very bright in the 3.9 µm imagery and,
at low solar angles, the sensor (and thus the image) becomes
saturated
• Possible application: wind speed and direction, oil spills ...
Slide: 75Version 1.1, 30 June 2004
Sun Glint
Channel VIS0.6 (inverted) Channel IR3.9
Early morning sun glint over the Gulf of Arabia
MSG-1, 14 July 2003, 04:00 UTC
Slide: 76Version 1.1, 30 June 2004
Early morning sun glint over the Arabian & Red Sea
MSG-1, 24 April 2003, 4:15 UTC, Channel 04 (3.9 µm)
Sun Glint
Slide: 77Version 1.1, 30 June 2004
Midday sun glint over the Kongo river
MSG-1, 24 March 2004, 09:00 UTC, Channel 04 (3.9 µm)
Sun Glint
Slide: 78Version 1.1, 30 June 2004
Afternoon sun glint over the Atlantic
MSG-1, 04 March 2004, 12:00 UTC, Channel 04 (3.9 µm)
Sun Glint
Slide: 79Version 1.1, 30 June 2004
SUB-PIXEL RESPONSE
OF IR3.9 CHANNEL
Slide: 80Version 1.1, 30 June 2004
IR3.9 Channel: Sub-pixel response
• Radiance is not linear with temperature: B = Tα/λ
• The response to changes in scene temperature is much larger at
shorter wavelengths
Strong non-linear increase of radiance with
increasing temperature
More "linear" increase of radiance with
increasing temperature
Slide: 81Version 1.1, 30 June 2004
IR3.9 Channel: Sub-pixel response
If there is variability within a field-of-view (FOV), then the radiance
for that FOV is a linear combination of the separate radiances (not
their temperatures)
300 K 500 K
300 K 300 K
IR3.9: B = T13.6
B=(B1+B2+B3+B4)/4
T=B1/13.6
T = 451 K
IR10.8.9: B = T4.8
B=(B1+B2+B3+B4)/4
T = B1/4.8
T = 392 K
example: “hot” spot within one
pixel observation
Slide: 82Version 1.1, 30 June 2004
IR3.9: DETECTION OF
(FOREST) FIRES
(DAY AND NIGHT-TIME)
Slide: 83Version 1.1, 30 June 2004
IR3.9 Channel: Sub-pixel response
• Its strong sensitivity to subpixel
"hot areas" makes the
IR3.9 channel very useful in
fire detection.
TB (3.9 µµµµm) - TB (10.8 µµµµm)
• If only 5% of the pixel is at
500 K, the IR3.9 channel
measures 360 K, while the
IR10.8 measures less than
320 K.
• If large fractions of the pixels
are covered by fire, both
channles can easily detect the
fire.
Slide: 84Version 1.1, 30 June 2004
Fires over Angola and Kongo
MSG imagery on 25 June 2003 at 10:00 UTC
Channel 04 (3.9 µµµµm) Channel 09 (10.8 µµµµm)
Angola
Kongo
Channel 04 detects the
fires (black spots), while
Channel 09 does not
detect the fires.
Slide: 85Version 1.1, 30 June 2004
Fires over Portugal and Spain
Channel 04 (3.9 µµµµm) Channel 09 (10.8 µµµµm)
MSG imagery on 3 August 2003 at 12:00 UTC
In case of very large
fires, also Channel 09
detects some fires.
Slide: 86Version 1.1, 30 June 2004
IR3.9: DETECTION OF
BROKEN CLOUDS
(DAY AND NIGHT-TIME)
Slide: 87Version 1.1, 30 June 2004
IR3.9 Channel: Sub-pixel response
Its strong sensitivity to sub-pixel
variations makes the IR3.9
channel also useful for fractional
cloud cover (if the effects of
emissivity and atmospheric
moisture can be ignored)
TB (3.9 µµµµm) - TB (10.8 µµµµm)
Slide: 88Version 1.1, 30 June 2004
Detection of Broken Clouds
MSG-1, 14 July 2003, 2:00 UTC
a lot more cloud structure is visible in the difference image
Channel IR10.8 difference IR3.9 - IR10.8
1
1 = desert surface (bluish)
2 2
2 = non-broken clouds (bluish)
3 3
3
3 = broken clouds (reddish)
4
4
4
4 = Cirrus clouds (yellowish)
Slide: 89Version 1.1, 30 June 2004
IR3.9: DETECTION OF
URBAN HEAT "ISLANDS"
(DAY-TIME)
Slide: 90Version 1.1, 30 June 2004
Urban Heat "Islands"
• MSG IR imagery makes it possible to locate urban heat
"islands" under clear sky conditions
• The IR3.9 channel is better than IR10.8 better because of high
sensitivity to sub-pixel temperature variations (warm areas in
cities are like little fires)
• Stronger signal (temperature difference city - surroundings) in
IR3.9 than in the other IR channels
Slide: 91Version 1.1, 30 June 2004
Urban Heat
"Islands"
MSG-1
14 July 2003
02:00 UTC
BT Channel 04 (3.9 µm)
Paris: 287 K
Surrounding: 281 K
Paris
Slide: 92Version 1.1, 30 June 2004
Urban Heat
"Islands"
MSG-1
14 July 2003
02:00 UTC
BT Channel 09 (10.8 µm)
Paris: 291 K
Surrounding: 286 K
Paris
Slide: 93Version 1.1, 30 June 2004
MSG-1
14 July 2003
02:00 UTC
BT Channel 04 (3.9 µm)
Urban Heat
"Islands"
Cairo
Slide: 94Version 1.1, 30 June 2004
NOISE IN THE IR3.9 CHANNEL
Slide: 95Version 1.1, 30 June 2004
Noise in the IR3.9 Channel
Radiance
Wavelength (µµµµm)
• At IR10.8, equivalent brightness temperatures can be determined very accurately at
both warm and cold scene temperatues
• At IR3.9, the radiance increases rapidly with increasing temperature (see next slide)
• Since measurement accuracy is constant, the result is a much less accurate
temperature measurement at cold scene temperatures in the IR3.9 channel
Slide: 96Version 1.1, 30 June 2004
Noise in the IR3.9 Channel
At IR3.9, a small
change in radiance
corrisponds to a
large change in
temperature
Slide: 97Version 1.1, 30 June 2004
During the night, the IR3.9 channel
cannot be used for cold cloud tops.
Below BTs of 220 K the IR3.9
channel is very noisy (radiances
close to zero).
RAW RAD TEMP
[count] [mW/m2] [K]
54 0.01 218
53 0.01 213
52 0.00 205
51 0.00 131
Noise in the IR3.9 Channel: Example
MSG-1, 14 July 2003, 02:00 UTC, IR3.9
Interpretation: IR3.9 imagery does a
fine job for warm scene temperatures,
but at night it is not useful for cold
scenes like thunderstorm tops.
IR3.9: noisy picture
Slide: 98Version 1.1, 30 June 2004
MSG-1, 14 July 2003, 02:00 UTC, IR10.8
Noise in the IR3.9 Channel: Example
IR10.8: smooth picture IR3.9-IR10.8: also noisy picture
MSG-1, 14 July 2003, 02:00 UTC,
IR3.9 - IR10.8
Slide: 99Version 1.1, 30 June 2004
SUMMARY
Slide: 100Version 1.1, 30 June 2004
Comparison IR3.9 vs IR10.8
• IR3.9 has solar contribution [daytime]
• IR3.9 is not a pure window channel
(CO2 band) ➨ Limb cooling
• Emissivities in IR3.9 differ from IR10.8
• IR3.9 is very sensitive to sub-pixel temperature
variations
• Noise in IR3.9 makes it useless for T < 220 K
• Strong sun glint in IR3.9
Slide: 101Version 1.1, 30 June 2004
http://www.cira.colostate.edu/ramm/goes39/cover.htm