This document analyzes a haze event over Singapore in October 2010 caused by smoke from fires in Sumatra, Indonesia. Photometric and LIDAR data were used to characterize the aerosols. Analysis showed high concentrations of fine particles less than 0.4 microns from the fires, with air quality degraded and visibility reduced. Trajectory and inversion modeling indicated both fresh and aged smoke was transported to Singapore from the Sumatra fires.

1. Photometric analysis of the October 2010 haze
event over Singapore
by
S.V. Salinas, B.N. Chew and S.C. Liew
IGARSS 2011, 24th - 29th Vancouver, Canada
Centre for Remote Imaging, Sensing and Processing

2. October 2010 Haze over Singapore
Centre for Remote Imaging, Sensing and Processing

3. October 2010 Haze over Singapore
Centre for Remote Imaging, Sensing and Processing

4. Overview
Vegetation fires are a normal phenomenon in the
South–East–Asia region.
Tropical wild fires can occur thought the year
and they are specially common during the dry
season months (June–November).
The severity of vegetation fires (bio–mass
burning) can be greatly by human intervention.
According the the Global Fire Emissions Database
(GFED), during the period 1997–2006, there were
two major fire episodes in Indonesia (1997,
2006) and two minor episodes (2002, 2004).
During the disastrous 1997 bio–mass burning
episode, the equivalent of 13–to–40 % of the
mean annual global carbon emissions from fossil
fuels were released into the atmosphere
Centre for Remote Imaging, Sensing and Processing

5. Overview
• On 15th October 2010,
persistent smoke–fire
activity over central
Sumatra, province of
Riau was detected.
• The prevailing south-
westerly to westerly
winds carried in smoke
haze from the fires in
Sumatra over Singapore
and peninsular
Malaysia.
• According to a press
release of NEA, on
19th October, the 24-
hr PSI 1 at 4pm was 56
and classified as a
moderate event. By
6pm, the 3-hr PSI has
increased to 78
approaching unhealthy
levels.
Centre for Remote Imaging, Sensing and Processing

6. Atmospheric Super-site in
Singapore
• Established under the cooperative framework of
the Seven South-East Asian Studies (7-SEAS)
program initiated by NASA and the Office of
Naval Research (ONR).
• Situated in National University of Singapore
(NUS).
• Main Site on Block E2 rooftop (1.3 N 103.7 E /
79 m).
• Secondary Site on Block S2S rooftop (~ 340 m
away from Main Site).
Centre for Remote Imaging, Sensing and Processing

7. The AERONET network
Perform observations of
direct and diffuse
transmitted radiation at
more than 180 locations
worldwide.
AERONET radiometer's
measure total columnar
optical depth and sky
radiance using 2 different
observation sequences:
almucantar and principal
plane scans.
Singapore's Sun-photometer
performs measurements at
six spectral bands i.e.
[0.340, 0.380, 0.440,
0.500, 0.675, 0.870, 1020]
nm.
Centre for Remote Imaging, Sensing and Processing

8. The MPLNET network
• Part of NASA’s MPLNET network.
• Compact and eye-safe LIDAR.
• Determines heights of aerosols and clouds by
measuring time-of-flight from transmission
of laser pulses to reception of returned
signals.
Optically Thin Cirrus
Local Aerosols
within
Boundary Transported Smoke Layer
Layer
Centre for Remote Imaging, Sensing and Processing

9. Methodology and data
processing
• The aerosol optical depth (AOD) at wavelength (λ) is one
of the standard parameters derived from Sun-photometers.
• AOD (τ_a) and its first (α) and second (α') spectral
derivatives respect to wavelength, are often used to
describe the interaction of aerosol particles present on
a given particle size distribution (PSD).
• The first derivative which is also known as the Angstrom
exponent (α), can provide a useful measure of the
average aerosol dimensions in the sub– and super–
micrometer particle size range.
• The Angstrom exponent itself is influenced by particle
number variations of the two fundamental modes (fine and
coarse).
• The second derivative (α') provides a useful means to
test the departure from linearity which is inherent from
the formulation of the Angstrom law, it also is a useful
indicator of particle size.
Centre for Remote Imaging, Sensing and Processing

10. Methodology and data
processing
• By starting from the basic assumption that the
PSD can be represented as a bi–modal
distribution, O’Neill and collaborators were
able to extract the fine (τ_f ) and coarse mode
(τ_c ) optical depth from the spectral shape of
the total AOD (τ_a = τ_f + τ_c ).
• Their scheme, known as the spectral
decomposition algorithm (SDA), was essentially
dependent on the fact that the coarse mode
spectral variation is approximately neutral.
• Once the fine mode fraction (η = τ_f /τ_a ) is
know, then fine mode equivalent of aerosol
optical depth and Angstrom exponent number can
be readily extracted.
Centre for Remote Imaging, Sensing and Processing

11. Methodology and data
processing
• For the October 2010 haze event we have extracted one month
non–cloud screened AERONET level 1.0 data.
• Since the SDA algorithm can be considered as a partial cloud
screening technique, no further cloud screening protocols
were applied; instead restrictions based on the Angstrom
number and its derivative (α > 0.75 and −1.1 < α' < 2.0) was
employed.
• However, the entire data set was quality assured according to
AERONET-SDA level 2.0 standards in which five of the seven
available photometer channels were included (bounded by the
380–870 nm channel range).
• As a requirement for SDA, measured AOD was fitted to a 2nd–
degree polynomial in log–log space [ln τ_a = P^(2) (ln λ)].
• Subsequently, parameters such as α and α' and its fine/coarse
mode counterparts were computed at a reference wavelength of
500 nm.
Centre for Remote Imaging, Sensing and Processing

12. Hot spot fire detection and in-
situ PM2.5 measurements for
October 2010
Centre for Remote Imaging, Sensing and Processing

13. Trans-boundary smoke fires and
PM2.5 measurements
FIG(1) : Fire detection by MODIS Rapid Response System. Most smoke fire hot spots were
located at the region of Sumatra, province of Riau, Indonesia (Courtesy of 7-SEAS data
repository).
Centre for Remote Imaging, Sensing and Processing

14. Trans-boundary smoke fires and
PM2.5 measurements
FIG(2) : Fire detection by MODIS Rapid Response System. Most smoke fire hot spots were
located at the region of Sumatra, province of Riau, Indonesia (Courtesy of 7-SEAS data
repository).
Centre for Remote Imaging, Sensing and Processing

15. Trans-boundary smoke fires and
PM2.5 measurements
FIG(3) : PM2.5 measurements at Singapore super-site (07th-July to 30th-July).
Concurrent MODIS detected fire counts for the same period.
Centre for Remote Imaging, Sensing and Processing

16. Trans-boundary smoke fires and
PM2.5 measurements
FIG(4) : 7-day Back trajectory computations for day 21st. Thanks to Tom L. Kucsera
(GESTAR/USRA) at NASA/Goddard.
Centre for Remote Imaging, Sensing and Processing

17. Trans-boundary smoke fires and
PM2.5 measurements
FIG(5) : 7-day Back trajectory computations for day 24th. Thanks to Tom L. Kucsera
(GESTAR/USRA) at NASA/Goddard.
Centre for Remote Imaging, Sensing and Processing

18. Bio–mass burning smoke over
Singapore: Photometric and Lidar
data description
Centre for Remote Imaging, Sensing and Processing

19. AOD and Angstrom exponent
distributions for Oct. 2010
FIG(6) : Combined Angstrom exponent and aerosol optical depth statistics and
concentration for Oct. haze event.
Centre for Remote Imaging, Sensing and Processing

20. Temporal evolution of fine mode
event: 16 Oct. 2010
FIG(7) : Fine and coarse mode AOD and Angstrom number retrievals (left), fine mode
fraction ratios. LIDAR times shown as vertical lines.
Centre for Remote Imaging, Sensing and Processing

21. Temporal evolution of fine mode
event: 16 Oct. 2010
FIG(8) : LIDAR NRB vertical profile. Three AOD and aerosol extinction profiles are
shown.
Centre for Remote Imaging, Sensing and Processing

22. Temporal evolution of fine mode
event: 20 Oct. 2010
FIG(9) : Fine and coarse mode AOD and Angstrom number retrievals (left), fine mode
fraction ratios. For this case no LIDAR times were available.
Centre for Remote Imaging, Sensing and Processing

23. Temporal evolution of fine mode
event: 24 Oct. 2010
FIG(10) : Fine and coarse mode AOD and Angstrom number retrievals (left), fine mode
fraction ratios. LIDAR times shown as vertical lines.
Centre for Remote Imaging, Sensing and Processing

24. Temporal evolution of fine mode
event: 24 Oct. 2010
FIG(11) : LIDAR NRB vertical profile. A single AOD and aerosol extinction profile is
shown.
Centre for Remote Imaging, Sensing and Processing

25. Aerosol classification for smoke
event of Oct. 2010
FIG(12) : Aerosol classification chart shows elevated fine mode fractions for days
16th , 20th and 24th.
Centre for Remote Imaging, Sensing and Processing

26. Aerosol climatology for smoke
event of Oct. 2010
FIG(13) : AERONET inversions : Aerosol size distribution for selected dates.
Centre for Remote Imaging, Sensing and Processing

27. Aerosol climatology for smoke
event of Oct. 2010
FIG(14) : AERONET inversions : Single scattering albedo.
Centre for Remote Imaging, Sensing and Processing

28. Summary
• During October 2010, • Trajectory analysis indicated
persistent smoke fire activity the presence of both, fresh
over central Sumatra, was and aging smoke.
detected. • LIDAR retrievals showed
• There was a substantial profiles consistent with highly
degradation of air quality and absorbing particles such as
reduced visibility. those from bio-mass burning.
• The greatest impact of the • Model inversions showed high
October 2010 smoke event concentration of very fine
was between days 14th to particles of the order of 0.4
24th (PM2.5 > 40.5μgr./m3). micron or less.
• Critical parameters such as • Particulate single scattering
the classical Angstrom albedo show the presence of
number, its fine mode version highly absorbing particles.
together with the fine mode • Large difference in SSA
fraction consistently indicate showed different fuel sources,
the presence of fine sub- combustion phases and
micron particles. aerosol aging.
Centre for Remote Imaging, Sensing and Processing