| Atmospheric aerosols play an important role in climate
change. In fact, much research supports the theory that atmospheric aerosols could partly
neutralize the heating effect of greenhouse gases. Aerosols are typically studied over
water, due to the relatively constant reflectance of water and the ability to easily
separate surface and atmospheric contributions on the satellite signal. The remote sensing
of atmospheric aerosols over land, however, is still a challenging issue. The influence of
surface reflectance over land is much more complex on a satellite signal, which makes it
difficult to separate the surface and atmospheric contributions. Dr. Fang Li, senior research associate, and Professor Rudolf Husar, at
Washington University in St. Louis, Missouri, are taking a novel approach to aerosol
pollution research. Using ENVI, they have proven that SeaWiFS (Sea-viewing Wide Field-of
view Sensor) satellite data, typically used for the study of the Earth's waters, can be
successfully used in cloud-free areas over both water and land in the study of aerosol
pollution. Their 30-day pilot study has shown the existence of large, 1km x 1km scale hazy
air masses with strongly varying spectral extinction - a measure for aerosol attenuation
to sunlight - which could be a sign of aerosol pollution, and has demonstrated that the
methodology is potentially applicable for daily, global monitoring of aerosols and surface
color over cloud-free areas.
Co-Retrieval of Aerosol and Surface Reflectances Gives
More Accurate Picture
Dr. Li and Professor Husar's primary goal is to extract
atmospheric aerosol information - particulate matter air pollution - in the Northeastern
region of the United States in an effort to identify the existence of aerosol pollution in
that area. "Standard remote sensing techniques provide only the combined surface and
aerosol reflectance, but not their separate contributions," explains Dr. Li. "We
are focusing on a unique technique called co-retrieval. The challenge of co-retrieval is
to separate the aerosol and surface reflectances in order to have 'truer' measurements of
both."
Their project is based on a radiative transfer theory that
states that sensed radiation is decomposed into scattering and absorption by gases and
aerosols, as well as reflection from surfaces and clouds, and that air scattering and
surface/aerosol reflectance are assumed to be additive.
The co-retrieval process entails several steps. First, one
removes air scattering and absorption from the satellite images. Second, long-term data
are used to generate aerosol and cloud-free surface images. Third, the initial aerosol
properties are estimated over non-cloudy areas. Fourth, the aerosol is derived for each
pixel based on an excess reflectance. Lastly, daily aerosol-free surface reflectances are
reconstructed, and the aerosol properties are updated.
Although the retrievals of aerosol and surface reflectance
are interdependent, and provide a mutual quality control through iteration, successful
aerosol retrieval depends on high quality surface reflectance data, and the surface
reflectance can only be retrieved for known aerosol-optical properties.
For Dr. Li and Professor Husar's research, the retrieved
aerosol parameters are the columnar aerosol optical thickness at several wavelengths, such
as .416 and .67 um, and the spectral aerosol reflectance function. Both the aerosol
parameters and the surface reflectance are retrieved over all cloud-free areas at all
wavelengths.
Dr Li and Professor Husar use eight wavelength (0.4-0.9 m)
SeaWiFS LAC satellite data to co-retrieve both the columnar aerosol optical properties and
the aerosol-free surface reflectance. Columnar aerosol optical properties are aerosol
optical properties integrated with respect to the height from the ground surface to the
top of the atmosphere. The properties indicate the quantity of aerosol load that exists in
the entire atmospheric layer. Then, they separate the aerosol reflectance, the air
molecular reflectance, and the surface reflectance from the satellite-measured total
reflectance.
Dr. Li is primarily interested in the spectral aerosol
optical thickness, and spectral surface reflectance. "Spectral aerosol optical
thickness indicates how large the aerosol Angstrom exponent is," explains Dr. Li.
"Usually, the larger the Angstrom exponent is, the smaller the average aerosol
particle size is." Spectral surface reflectance allows him to measure ground biomass.
In addition to SeaWiFS data, Dr. Li and Professor Husar also
use WMO visibility data and GIS datasets. WMO visibility data are visibility records
collected by the World Meteorological Organization from the meteorological stations and
some airports throughout the world. They use the visibility data as a reference index of
air pollution to compare the atmospheric aerosol optical properties measured on ground and
from space. If there are unique properties, the visibility information is a critical
element of most radiative transfer models used in satellite atmospheric correction. The
GIS datasets used are ArcView Shape Files, which provide geophysical border information
that is overlaid on the ENVI image for a map-formatted output.
ENVI Provides Powerful Image Processing and Analysis
Dr. Li used IDL to write
the data processing routines for the analysis of the SeaWiFS data. He uses ENVI for the
data processing and image analysis. Although Dr. Li typically works with data sets that
range from 30 to 50 MB, it is not uncommon for him to work with SeaWiFS processed data as
large as 130 GB.
Using ENVI, Dr. Li was able to develop a technique called the
"cloud screen approach." "The cloud screen approach is based on the
at-sensor spectral reflectance," explained Dr. Li. "Cloud spectral reflectance
is in flat shape and has high absolute value for each band. We set both the spectral slope
threshold and band absolute value threshold to filter the cloud pixels in images."
Dr. Li considered other packages before choosing ENVI, but
none measured up to ENVI's power and flexibility. "We looked at SeaDAS, which is a
freeware package supported by the NASA ocean biochemistry program specifically designed
for the analysis of SeaWiFS data. However, it is only available on sgi and Sun platforms,
and we needed a more platform-independent solution."
"Two features convinced us to buy ENVI. One is that it
easily displays SeaWiFS data in gray scale as well as RGB. Another is that there is great
flexibility for users to manipulate image datasets and render them. I really appreciate
ENVI's flexibility in input and output files," said Dr. Li.
"Also, we could not display so many images in different
band combinations, nor show the x, y, z, profiles as easily, without ENVI."
Dr. Li uses the Building Geometry File routine and
Georeference HDF SeaWiFS Data routines the most. "The former is used to produce the
satellite viewing zenith and azimuth angles, and solar zenith and azimuth angles. The
latter allows us to register the image into the proper latitude and longitude
coordinates," he said.
"The ENVI z-profile feature is also very important to
us. We use it to recognize the spectral reflectance characteristics of vegetation,
seawater and cloud, so that we can create the filter of cloud. Gray and RGB arbitrary band
selections enable us to find that the bands seven and eight are only minutely influenced
by atmosphere, so their corresponding images can be used to distinguish the boundary of
the ground surface between sea and land," he said.
"ENVI has been very instrumental to us. It is a very
powerful and effective image-processing package in the field of remote sensing. It is so
good at rendering color images in a flexible way, and does so very quickly. It also allows
users to develop their own interfaces, which is a great benefit for someone like me. I
look forward to using ENVI in the next stage of this project, and in future remote sensing
projects."
*SeaWiFS is the NASA name for the OrbView-2 sensor owned
and operated by Orbital Imaging Corporation, Dulles, VA. |
