DEVELOPING A SATELLITE-BASED TOOL TO MONITOR DUST AND SAND STORMS IN

THE UAE

Ali Al Suwaidi1, Adnan Al Rais2 and Hosni Ghedira3

1
Associate Researcher, 2Associate Researcher -Team Leader, Space Program
Emirates Institution for Advanced Science & Technology (EIAST), Dubai, UAE.
[email protected], [email protected]
3
Associate Professor of Water and Environmental Engineering, MASDAR Institute/MIT
[email protected]

ABSTRACT such events. In addition to its direct effect on surrounding air

quality, excessive presence of airborne dust affects both
Due to their low precipitation rates, arid regions are the local and regional environments due to its biogeochemical
world’s major source of atmospheric dust that has an impact impact on the ecosystem and its radiative-forcing effect on
on local, regional and global climate. Dust and sand storms the climate system [1], [2].
create potentially hazardous air quality to humans, and Presently, most of the Earth satellite data is available to
adversely affecting climate on a regional and world-wide users free of charge through different governmental space
scale. In addition to its direct effect on surrounding air agencies or through their affiliated centers and programs.
quality, excessive presence of airborne dust affects both The acquisition frequency of this data varies from 96 images
local and regional environments due to its biogeochemical per day for the European METEOSAT (~ 1km footprint).
impact on the ecosystem and its radiative-forcing effect on Acquiring such data can benefit scientists and researchers by
the climate system. incorporating remote-sensing-derived information in their
The objective of this research is to develop a thermal- observation systems for environmental monitoring,
based technique to detect and monitor dust and sand storm evaluation and planning.
events in the UAE from space. The developed tool will use Application of geostationary and polar orbiting remote
the difference in particle size between airborne dust and sensing in dust and sand storms has been widely investigated
surface sand to detect airborne dust over desert. Previous in the past two decades [1-5]. In this project, a new
similar studies have observed strong differences in infrared technique will be developed to detect and mask pixels with
emissivity between airborne dust particles, with size less moving dust from SEVIRI HRV and the two other visible
than 5 μm, and desert sand particles, with diameter greater channels (R01 and R02). This tool will be helped with a
than 70 μm. This difference in thermal behavior will be used second neural network system that detects and extracts
in this study to detect the presence and map the extent of predefined features in the dust and sandstorm fields. The
airborne dust over the study area. Several well-documented obtained dust storm simulations will be then re-sampled and
dust storm events that occurred between 2008 and 2009 will compared to the ones obtained by the NCAR WRF regional
be used to calibrate and validate the new tool. Due to their prediction model at 16-km resolution [6-7] as well as daily
high temporal resolution, geostationary data from aerosol maps produced by NASA (Geovanni tool). The
METEOSAT SEVIRI-MSG was used and Preliminary data images shown in Figure 2 illustrate the temporal evolution of
assessment tests have shown a great potential of this one dust storm event detected by METEOSAT SEVIRI-
approach in detecting airborne dust and sand over the bright MSG.
underlying surfaces. Thermal channels have also been used in detecting and
mapping aerosols over desert. The thermal-based
Index Terms— Dust, Sand Storm, METEOSAT, approaches use the comparison between thermal properties
SEVIRI-MSG. of mineral aerosols and the background temperature and
water vapor signals. These techniques have shown some
1. INTRODUCTION limitations due to the difficulties in separating aerosols from
terrestrial environment, particularly over desert and semiarid
Dust and sand storms are creating potentially hazardous air regions [8]. The NASA daily aerosol-mapping tool,
quality to humans, and adversely affecting climate on a Geovanni, is also limited to aerosol retrieval over ocean and
regional and worldwide scale. Remote sensing has shown to dark land surfaces.
be a valuable tool in detecting, mapping and forecasting

978-1-4244-9566-5/10/$26.00 ©2010 IEEE 1434 IGARSS 2010

Authorized licensed use limited to: CURTIN UNIVERSITY. Downloaded on August 16,2021 at 08:39:08 UTC from IEEE Xplore. Restrictions apply.
 2. METHODOLOGY 3. RESULTS AND DISCUSSION

The graph presented in figure 1 shows the temporal The thermal-based approach was applied over our study
variation of AOD averaged over water pixels in the UAE area and has shown a great potential in detecting dust over
coastal area. The two images shown in this figure were bright surfaces (desert). Although detecting dust over dark
acquired by Meteosat satellite in one clear day (November surfaces, such as water, can be easily done with visible
2008) and one dusty day (February 2009). The selected channels, detecting dust over bright surfaces such as desert
coastal pixels are located in the rectangle shown on the top remain challenging at that wavelengths. The obtained results
image. This figure shows clearly the effect of dust presence have demonstrated that a thermal difference approach can
on satellite derived AOD. Usually, the AOD is saturated in complement visible-based-approach in detecting dust in their
the presence of dust with values exceeding 0.9 (or 90%). originating location (desert). The SEVIRI scenes and
The developed tool will use the difference in particle size thermal emissions shown in figures 4, 5 and 6 show the
between airborne dust and surface sand to detect airborne potential of thermal-difference-approach in monitoring dust
dust over desert. presence under clear, moderate and severe dust-storm
conditions respectively.

Figure 1: Temporal variation of Aerosol Optical Depth

(AOD) in UAE coastal area.

Other thermal-based techniques used the difference in

particle size between airborne dust and surface sand to
detect airborne dust over desert as shown in Figure 3. A
study performed by Wald and al. has shown strong
differences in infrared emissivity between airborne dust
particles, with size less than 5 μm, and desert sand particles,
with diameter greater than 70 μm. This difference in thermal
behavior was detected by MODIS thermal channels and was
used to detect the presence and map the extent of airborne
dust [9]. Figure 4: Clear Day December22, 2008

Figure 3: Differences in thermal emission in clear and

dusty days.

1435

Authorized licensed use limited to: CURTIN UNIVERSITY. Downloaded on August 16,2021 at 08:39:08 UTC from IEEE Xplore. Restrictions apply.
 4. CONCLUSION

Retrieving dust and sand storm properties over their

originating location (i.e., desert, arid and semiarid regions)
using conventional visible and near-infrared wavelengths is
a difficult task because of the bright underlying surfaces
over such regions.
To overcome this limitation, several approaches have
been developed to retrieve aerosol optical properties over
bright land surfaces such as desert. Most of these techniques
use the sensor multi spectral properties by selecting highly
contrasted areas as reference targets. This approach has
shown low efficiency when applied over large and
homogenous bright areas like desert with limited land cover
variations.
To overcome this limitation, thermal channels have
been tested in detecting and mapping aerosols over desert.
The thermal-based approaches use the comparison between
thermal properties of mineral aerosols and the background
temperature and water vapor signals. These techniques have
shown some limitations due to the difficulties in separating
aerosols from terrestrial environment, particularly over
desert and semiarid regions. In this research, MSG/SEVIRI
Figure 5: Moderate Sand Storm February 13, 2009 Thermal Channels (T04-T09) have shown great potential in
detecting and classifying Dust and Sand Storms over bright
surfaces (i.e. Desert). Additionally the difference between
these two channels can be used to classify dusty conditions
into different intensity levels. The three charts presented in
figure 7 show the magnitude of thermal differences under
clear (a), moderate (b) and heavy (c) storm conditions.
These results have shown that the thermal difference is
positively correlated with dust intensity. A multiple
regression model using thermal emissions, solar and satellite
angles and land cover parameters is currently being
developed to improve the dust classification tool.

Figure 6: Severe Sand Storm February 12, 2009

(a)

1436

Authorized licensed use limited to: CURTIN UNIVERSITY. Downloaded on August 16,2021 at 08:39:08 UTC from IEEE Xplore. Restrictions apply.
 5. REFERENCES

[1] J. Haywood and O. Bougher, “Estimates of the direct and

indirect radiative forcing due to tropospheric aerosols: A review,”
Rev. Geophys., vol. 38, no. 4, pp. 513–543, 2000.
[2] N. C. Hsu, J. R. Herman, and C.Weaver, “Determination of
radiative forcing of Saharan dust using combined TOMS and
ERBE data,” J. Geophys. Res., vol. 105, no. D16, pp. 20 649–20
661, 2000.
[3] M. D. King, Y. J. Kaufman, D. Tanré, and T. Nakajima,
“Remote sensing of tropospheric aerosols from space: Past, present
and future,” Bull. Amer. Meteorol. Soc., vol. 80, no. 11, pp. 2229–
2259, Nov. 1999.

(b) [4] D. Tanré and M. Legrand, “On the satellite retrieval of Saharan
dust optical thickness over land: Two different approaches,” J.
Geophys. Res.,vol. 96, no. D3, pp. 5221–5227, Mar. 1991.
[5] D. Tanré, E. F. Vermote, B. N. Holben, and Y. J. Kaufman,
“Satellite aerosol retrieval over land surfaces using the structure
functions,” in Proc .IGARSS, 1992, vol. 2, pp. 1474–1477.
[6] S. Caquineau, A. Gaudichet, L. Gomes, and M. Legrand
(2002), Mineralogy of Saharan dust transported over northwestern
tropical Atlantic Ocean in relation to source regions, J. Geophys.
Res., 107(D15), 4251, doi:10.1029/2000JD000247.
[7] A. T. Evan, A. K. Heidinger, and M. J. Pavolonis (2006),
Development of a new over–water advanced very high resolution
radiometer dust detection algorithm, International Journal of
Remote Sensing, 27(18), pp. 3903-3924.
[8] Hsu, N.C. Si-Chee Tsay King, M.D. Herman, J.R. (2004).
Aerosol Properties over Bright-Reflecting Source Regions IEEE
Transactions on Geoscience and Remote Sensing, vol. 42, no. 3,
(c)
march 2004
Figure 7: Relationship between thermal difference (T04-T09) and [9] Wald, A. E., Y. J. Kaufman, D. Tanré, and B.-C. Gao (1998),
dust intensity Daytime and nighttime detection of mineral dust over desert using
infrared spectral contrast, J. Geophys. Res., 103(D24), 32,307–
32,313.

1437

Authorized licensed use limited to: CURTIN UNIVERSITY. Downloaded on August 16,2021 at 08:39:08 UTC from IEEE Xplore. Restrictions apply.