Скачать 186.51 Kb.
CHAPTER 15 |
Measurement of Aerosol Properties over Urban Environments from Satellite Remote Sensing
Min M Oo, Ph.D., E.E.
Research Associate, Space Science and Engineering Center
University of Wisconsin-Madison
Lakshmi Madhavan Bomidi, Ph.D., Physics
Research Associate Optical Remote Sensing Laboratory
Electrical Engineering Department
City College of the City University of New York
New York, USA
Barry M Gross, Ph.D., E.E.
Professor, Optical Remote Sensing Laboratory
Electrical Engineering Department
City College of the City University of New York
New York, USA
ABSTRACT: Aerosols have a significant effect on the Earth’s radiation balance but unfortunately, aerosol climate forcings are extremely difficult to quantify. This results from difficulties in retrieving aerosol loadings on a global scale due to the high variability of aerosol optical properties on one hand and their indirect effect on cloud processes and precipitation on the other. Aerosols also have a significant health impact with fine mode aerosols being strongly regulated by the EPA and therefore, aerosol observations that can assist in both air-quality monitoring and prediction are in demand. However, due to the variability of their optical properties, aerosol retrieval is inherently uncertain and these difficulties are further magnified when aerosol retrieval over land is required. This is due to the difficulty in separating the aerosol signature from the ground reflection signature. Another layer of complexity arises when retrieval of aerosols in urban areas is considered, due to the brightness and angular dependence of the surface reflection in comparison to dark vegetation or water surfaces. In addition, the need to retrieve aerosol properties beyond the column aerosol optical depth results in further difficulties that have not been completely overcome. The purpose of this chapter is to first present the satellite sensors and operational techniques most commonly used for aerosol retrieval and the underlying strengths and weaknesses in these approaches. We then discuss the next generation of sensors and algorithms and how they can be expected to help overcome present day difficulties.
It is well known that accurate global characterization of Aerosol Optical Depth (AOD) is essential in accurately determining the energy balance for climate change studies (Charlson et al., 1992) as well as quantifying fine particle pollutants and subsequent health risks (Wilson and Suh, 1997). Regarding climate forcings, aerosol interactions can affect the climate both directly and indirectly. For example, direct forcing impacts occur through modification of the Earth’s albedo due to highly scattering aerosols which result in a net cooling by scattering solar radiation back into space (Twomey 1974, 1977). At the same time, aerosols can modify cloud processes through so-called indirect mechanisms (Twomey 1980; Ackerman et al., 2004). Unfortunately, since observations of clouds and aerosols cannot be done at the same time from space, the main validation of this effect relies on statistical metrics (Nakajima et al., 2001) where individual aerosol and cloud properties are averaged in time and space.
Figure 15.1. Climate Forcings due to atmospheric constituents (from IPCC assessment report, 2007)
All of these forcing mechanisms should be contrasted against green house gas emissions (GHG) and associated warming mechanisms. In particular, GHG emissions are based on accurate determination of these gases. Fortunately, such measurements are easier from space since they rely mainly on infrared absorption which can be well measured by polar orbiting spectrometers such as NASA AIRS (Atmospheric InfraRed Sounder) and Near Infrared (NIR) spectrometers such as SCanning Imaging Absorption SpectroMeter for Atmospheric CHartographY (SCIAMACHY) on board the European Space Agency (ESA) Environmental Satellite ENVISAT (Rublev and Uspenskii 2006; Crevoisier et al., 2003) For these reasons, the uncertainty of climate forcing due to GHG is much less than those due to direct and indirect aerosol mechanisms. This is clearly illustrated in the Radiative Forcing error bars depicted in Figure 15.1 taken from Intergovernmental Panel on Climate Change’s Fourth Assessment Report (2007) (http://www.ipcc.ch). Complicating issues even more, is that trace gases are rather homogenous in the atmosphere whereas aerosols are much more variable in space and time.
On the other hand, aerosols can have significant effects on human health with exposure linked to various health risks including heart and lung related distress such as asthma. In particular, fine mode particulates (PM2.5) are most important since these small particles penetrate deep into the lung tissue and can result in significant lung and cardiac distress (EPA 1996). In fact, the Environmental Protection Agency (EPA) has strict PM National Ambient Air Quality Standards (NAAQS) setting 24-hour exposure at 35 µg/m3 (http://www.dec.ny.gov/).
One major reason that aerosol measurements over urban areas are important is that extended urban centers such as NYC, Mexico City, Shanghai etc. can have significant aerosol loadings with air quality levels that are above EPA standards. Monitoring air pollution levels from space is based on the observation that the column AOD under certain conditions is strongly related to PM2.5 (Al-Saadi et al., 2005; Engel-Cox et al., 2006). Such correlations are strongest in the Eastern US where dark vegetative surfaces make the retrieval of aerosols more accurate. Unfortunately, other mechanisms make it very difficult to connect satellite derived AOD to surface particle mass. These include the variability of the Planetary Boundary Layer Height (PBL Height), the existence of aloft aerosol plumes as well as the natural variability of aerosols as manifested in the ratio of scattering to particle volume. These difficulties in connecting air quality parameters to satellite derived AOD and possible strategies in accounting for these effects based on climatologies and/or other ancillary data is presented in detail in the previous book chapter.
Based on the above discussion, it is important that remote sensing methods be developed which measure aerosol optical depth over urban areas both for air quality as well as providing unique opportunities to address scientific issues such as aerosol-cloud interactions where heavy urban aerosol loadings are expected to provide the best opportunities in observing these hard to quantify climate modifying mechanisms (Wu et al., 2009). In addition, the need for some level of aerosol speciation is clearly of great importance but, unfortunately, can only be partially assisted by satellite observations. In fact, the best that can be expected for aerosol observations beyond column optical depth is some degree of separating out fine and coarse modes which is of obvious help both for air quality as well as identifying anthropogenic aerosols (fine mode dominated) in relation to natural aerosols as well as quantifying absorbing aerosols as measured by their single scattering albedo (SSA). Therefore, in this chapter, we will focus on the present and future observational capabilities for urban aerosol monitoring and, when possible, assess the capabilities of retrieving not only AOD but fine-coarse mode separation and SSA. The chapter will therefore focus on those satellite observations that provide operational and validated retrievals most crucial for climate and air-quality applications.
The structure of the chapter is as follows. In section 1, a brief description is given of those aerosol physical properties which are needed for aerosol retrieval. Since all satellite aerosol retrieval methods are based on an aerosol’s ability to scatter solar radiation, the microphysical properties of aerosols will be discussed with emphasis on those needed to quantify optical properties (size distributions and complex refractive index). In section 2, remote sensing algorithms based on single observation approaches are given with special focus on providing a natural hierarchy of algorithms which go from the simplest (single channel) systems to the most sophisticated systems in which polarized multispectral multiangle capabilities are utilized to their limit. In section 3, we look at methods which use ancillary data to refine global based algorithms. This approach shows significant promise, not only in removing biases in the AOD retrieval, but also in increasing the spatial resolution by allowing retrieval over brighter urban pixels. In section 4, a brief survey of next generation algorithms will be presented and an assessment of future capabilities given, and overall conclusions and future outlook is given in section 5.
15.1.1 Optical properties of aerosols
The optical properties of atmospheric aerosols are mainly determined by their size distribution and chemical composition as well as overall particle concentration. Unfortunately, unlike trace gases whose optical properties are well characterized and have stable spectral features suitable for identification and quantification, retrieving a complete description of aerosol microphysical and chemical properties from optical data is not feasible due to the lack of sharp identifying spectral and/or angular features in the aerosol scattering process. Historically, with single channel observations providing a single piece of information (i.e., Top of Atmosphere (TOA) reflection at a single wavelength), only the most basic aerosol quantity could be retrieved (i.e., AOD). However, with the advent of multispectral and/or multiangle capacity in the latest observing sensors, decomposition of aerosols into fine and coarse mode (fine/coarse mode fraction) and identification of aerosol absorption properties (i.e., SSA) has made its way into current retrieval algorithms. To extract this additional information, however, requires that aerosol models be developed which capture the essential properties of aerosols as they exist naturally in ambient conditions.
15.1.2 Aerosol Models
The simplest assumption used for most aerosol models is that aerosols are spherical and are externally mixed (Hess et al., 1998, Seinfeld and Pandis 1998). In this important case, aerosols are described by a small number of parameters that define the size distribution or modes and the complex refractive index. Most commonly, the size distribution is assumed to be of log-normal type with both a fine and coarse mode included. In fact, the bimodal log-normal distribution parameterization is used in almost all satellite remote sensing efforts and is given by Equation (1);which is consistent with measurement retrievals taken from AERONET (AErosol RObotic NETwork) (Holben et al., 2001).
Here, the log-normal parameters describe the total volume in each mode, the mean mode radius and mode width. AERONET is an extensive global scanning sky radiometer network capable of retrieving not only AOD but bimodal size distribution parameters (Dubovik and King 2000; Dubovik and Holben 2002a,b) and complex refractive index (or SSA) and functions as the main validation network in which satellite remote sensing aerosol retrievals are compared.
The characteristics of aerosols from AERONET for different regions worldwide show significant variability not only in AOD but in the ratio of fine – coarse mode and SSA (Dubovik, and Holben 2002b). The main urban feature in contrast with other cases (desert, oceanic) is the higher fine mode aerosol contribution with angstrom coefficients ranging from 1.2440-8702.5 (Schuster et al., 2006). Aerosol absorption in urban areas such as the North East (NE) coast of the US have high SSA at 440nm 0.9200.98 but cases such as Mexico City with heavy biomass burning contributions have SSA much smaller with ranges 0.6000.92. This is consistent with the high degree of variability expected since the albedo strongly depends on the age of the burning aerosol from the combustion source.
In addition, it is possible that the effective mode radii can depend on optical depth. This dynamic feature occurs for hygroscopic aerosols which, by definition, undergo changes in both physical (effective radius) and optical (refractive index) when subjected to high relative humidity RH (Hanel 1976) conditions. Examples of such aerosols include ammonium sulfate (NH4)2SO4, sea salt, and ammonium nitrate NH4NO3 which, for example, are the dominant modes in many urban coastal areas (i.e., North East US). However, not all aerosols are hygroscopic (Vlasenko et al., 2005). Most prominent among these are desert (mineral) dusts and most organic aerosol species. Therefore, aerosol models must account for the high probability of these aerosol classes in different areas of the world. For example, aerosol models are best modeled as static over the West Coast, USA and in many other world regions where dust and organic aerosol components are significant while dynamic models are best suited to hygroscopic aerosols (NE). Further complications arise with mineral dust species since these particles are far from spherical. To model these aerosols, non-spherical models including the most popular model of spheroids (Dubovik et al., 2006) are used in operational models. These models may be considered within the general Particle Size Distribution framework since they are characterized by a size distribution and refractive index. However, realistic assumptions on the statistics of the aspect ratio (semi major / semi minor axis) of constituent ellipsoids must be assumed a-priori.
15.2 SATELLITE REMOTE SENSING OF AEROSOL OVER LAND
15.2.1 General Considerations of current Aerosol over Land Algorithms
The characteristics and retrieval properties of selected satellite instruments discussed in this chapter are shown below
The algorithms for aerosol retrieval depend strongly on the different characteristics of the sensor, but all algorithms must in some way quantify the land surface contribution as well as obtain some estimate of aerosol type in order to retrieve AOD. The reason for this is that any solar reflection signal over a land pixel will consist of photons that have either directly interacted with the ground or interact only with the atmosphere. These photons, upon collection, are in essence indistinguishable unless a more sophisticated approach that can estimate the surface contribution or reduce its contaminating effects is implemented. This is referred to as surface correction and is an integral part of any aerosol retrieval over land scheme.
In addition, all aerosol retrieval algorithms require accurate and robust cloud screening procedures and different algorithms and sensors with different spatial resolutions have different ways to deal with it. Therefore, a significant but hard to quantify uncertainty in retrieval algorithms could be due to partial cloud contaminations (Kaufman et al., 2005). The cloud screening procedure for each retrieval algorithm is not included since it is beyond the focus of this chapter.
In describing the different aerosol algorithms in what follows, the most useful means of partitioning these algorithms is to discuss how they handle the surface correction. Possible strategies include:
|Airborne Measurements of Spectral Shortwave Radiation in Cloud and Aerosol Remote Sensing and Energy Budget Studies||P2 Integration of Geographic Information Systems and Remote Sensing (Topics in Remote Sensing, 5)|
|The Nansen Environmental and Remote Sensing Center||Volcanology, environmental remote sensing, and atmospheric chemistry|
|Gis / Remote Sensing / Landscape Modeling / Hazards Mapping||Natural resource inventory and monitoring related publications on remote sensing/gis|
|The Corona Charged Aerosol Detector – a “Universal” Detector for the Measurement of Non-volatile Components in Food, Natural Products and Supplements||Geog 579 Remote Sensing – 16237 001 Syllabus Instructor: Dr. Ron Resmini|
|Optical Remote sensing for monitoring evolution of ablation season mountain snow cover||（１）”On the characterization and gas sensing properties of Cu(II) tetra(alkylamino carbonyl) phthalocyanine lb films”, S. Capone, R. Rella, P. Siciliano, L|