Air Pollution

iTransboundary Movement Of Airborne PollutantsA Methodology for Integrating Spaceborne Images and Ground Based Data

ii

Acknowledgments

We extend our sincere appreciation to all who have helped in the support and preparation of this report. This paper presents the

results of a joint project of the U.S. Environmental Protection Agency (EPA) and the United Nations Environment Programme

(UNEP) Global Resource Information Database (GRID) office in Sioux Falls. We particularly want to recognize the financial

and project support from the U.S. EPAs Office of International Activities. Technical and project support came from the

UNEP/GRID Center Sioux Falls, EPAs Office of Research and Development (ORD), National Oceanic and Atmospheric

Agency (NOAA), U.S. Geological Survey (USGS) Earth Resources Observation System (EROS) Data Center, National Aeronau-

tics and Space Administration (NASA), and Washington University at St. Louis. Additional resources came from the NOAA

Operational Significant Event Imagery Server project and the Naval Research Laboratory, who are to be commended for posting

their data on-line for research use. The main contact for the pilot project is Russ Bullock, who is a NOAA employee on detail to

EPA ORD. Their team consisted of Matt Landis and other ORD researchers. Finally, special thanks for their support on this

project to Amy Fraenkel of U.S. EPA, Ashbindu Singh of UNEP/GRID Sioux Falls, and Rudolf Husar of Washington University

in St. Louis. Ross Lunetta of U.S. EPA and Gene Fosnight of UNEP/GRID Sioux Falls kindly reviewed the document and

Kimberly Giese of UNEP/GRID Sioux Falls did an excellent job on the publication layout.

ISBN: 92-807-2081-3

For further copies of the report, please contact:

Ashbindu Singh Phone: 1-605-594-6107/6117Regional Coordinator Fax: 1-605-594-6119UNEP/Division of Early Warning & Assessment - North America Email: [email protected] EROS Data Center http://www.na.unep.net/Sioux Falls, SD 57198-0001 USA

This report was prepared by Jill Engel-Cox, Battelle Memorial Institute, Thomas P. DeFelice, Raytheon Company, USGS EROS

Data Center, and Stefan Falke, American Association for the Advancement of Science, Environmental Fellow.

The views expressed in this publication are not necessarily those of the agencies cooperating in this project. The designations

employed and the presentations do not imply the expression of any opinion whatsoever on the part of the cooperating agencies

concerning the legal status of any country, territory, city, or area or of its authorities, or of the delineation of its frontiers or

boundaries. Mention of a commercial company or product in this report does not imply endorsement by the United Nations

Environment Programme or the U.S. Environmental Protection Agency. The use of information from this publication concern-

ing proprietary products for publicity or advertising purposes is not permitted. Trademark names and symbols are used in an

editorial fashion with no intention of infringement on trademark or copyright laws. We regret any errors or omissions that may

have been unwittingly made. Although EPA provided input for the project, this report was not formally reviewed by EPA.

iii

EXECUTIVE SUMMARY

The availability of relevant and accurate envi-ronmental information is essential for environ-mental policy-makers. Recent improvements insatellite remote sensing technologies, ground-based monitors, and data access have resulted inthe ability to observe and assess major atmo-spheric and ecological events around the world

on a timely basis.

Each of these monitoringtechnologies reveals different and usefulinformation, yet rarely are the resulting data setsused together in an integrated manner. TheU.S. Environmental Protection Agency (EPA)and the United Nations Environment Pro-gramme (UNEP) Global Resource InformationDatabase (GRID) office in Sioux Falls identifiedan environmental issue of global interest as atest case for applying an integrated approach:the transboundary movement of atmosphericpollutants.

Transboundary movement of atmosphericpollutants has ramifications for human and

environmental health, as well as economicimpacts. As a result, it is the focus of manybilateral, regional, and international policyefforts. A central question with atmosphericpollutant transport is how to monitor pollutantmovement and how to merge different monitor-ing datasets into useful information. Highlyvisible regional plumes of dust, smoke, and

urban haze can be seen with satellite sensors,while ground-based monitoringof air pollutants such as fine

particulates, SO2, and toxics

occurs at the local level. Integra-tion of these two kinds of measure-

ments allows the user to remotelyobserve large environmental effects in

many areas of the world, whileobtaining more detailed information

from ground-based monitors. Hence,the combination of satellite-based sensor

data and ground-based monitoring datapromotes greater understanding of the

movement of pollutants than either dataset alone. Combined data sets are impor-

tant for use by both scientists and interna-tional policy-makers.

A standard methodology did not exist toguide and encourage integrated use of satellite

images and ground-based data to monitor andunderstand major pollution events, such as airpollution. Thus, a small team was assembled todevelop a methodology for the integration ofsatellite images and ground-based data. First, weconducted a literature and project reviewcovering past and current integrated remote andground-based data projects, a literature search ofpublished work, and a search of data sets andtechnologies that could be used in a combinedform. Second, based on this search and docu-mentation, a general methodology was devel-oped for using integrated spaceborne and

iv

ground-based data sets, intended as a guide forgeneral scientists and policy-makers. Third, wefound an existing project that was willing to be apilot for testing the methodology: a U.S. EPA-NOAA project that was using aerial and ground-based sampling to learn more about theairborne sources of mercury deposition in theFlorida Everglades.

This document presents the results of theliterature and project review, the completemethodology, and the outcome of the FloridaEverglades pilot project.

Review Of Prior Work

A review of the literature, existing projects, andexisting satellite sensors and ground-basedmonitors was conducted. Several projectsintegrated satellite imagery and ground-basedmonitoring data, primarily in the area of trans-ocean dust storms, forest fires, and urban haze.All of these projects were conducted in the last 5years and were of limited scope. The data,techniques, and projects identified through thereview confirm that improved satellite andground-based data are becoming available andcan be integrated effectively; however, this dataintegration has not been done extensively to-date. Additionally, for global assessment andmonitoring, many regions of the world do nothave adequate ground sampling, and where dataare available, the data are often not readilyavailable for incorporation in integrated applica-tions. Satellite monitoring in conjunction withlimited ground-based monitoring would be veryuseful in these regions.

Methodology

The methodology described herein is designedto overcome both technical and institutionalbarriers to integrating disparate information

from multiple agencies in multiple countries. Toachieve this task: 1) the project must be welldefined, articulated and constructed on a soundpractical and theoretical foundation; 2) appro-priate partners who are committed to the projectmust be identified to ensure that critical tech-nologies and policy concerns are addressed; 3)critical data sources must be identified and madeavailable through cooperating partners; 4) theknowledge of the partners and the data must beshared through common standards andelectronic communication; and 5) the projectmust be implemented to fulfill the needs ofthe partners.

Pilot Project

Although the methodology is applicable to awide range of pollutants, a single pilot projectwas needed to test the methodology. Thechosen pilot application was an environmentalissue of current international concern:transboundary air pollution and mercurydeposition. This pilot project supplemented anexisting study of the airborne sources of mercuryfound in fish living in the Florida Everglades.Possible airborne sources of mercury includedlocal sources, non-local U.S. sources, long-distance sources from other countries, orcombination of these. This pilot project supple-mented the existing project by providing satelliteinformation on general air pollution movementand sources to be combined with the groundand aircraft measurements of the mercury thatwere collected in the Everglades and offshore.

The methodology proved an effectivemechanism for integrating satellite informationinto ground and aircraft mercury monitoring,for identifying the relevant data sources, and forbuilding the necessary partnerships to helpidentify mercury sources.

vConclusions And Future Direction

Our findings include:

Integrating satellite images and ground-based data can be beneficial for under-standing environmental issues.

Recent technological advances, includinglaunch of new satellite technologies,growth of ground-based air monitoringnetworks, and increased on-line accessibil-ity of satellite sensor images and surfacebased observations, make the integrateduse of satellite images and ground-baseddata possible.

The general methodology for integratingsatellite images and ground-based data,including defining a project, findingpartners and resources, selecting datasources, communicating electronically, andconducting a project, is valid and has beenconfirmed through a pilot project.

Satellite images integrated with ground-based data provide more informationabout an environmental phenomenon thaneither dataset alone.

The combined use of satellite sensors withground sampling systems can be an effective tool

to help policy-makers with decisions concerningthe protection of human health and our envi-ronment. The spatial resolution and temporalfrequency coverage from satellite sensors willonly improve over time. Many countries areimproving their ground sampling capacity. Theusefulness and success of an integrated dataapproach will necessarily depend on the avail-ability of local ground-based data that can becombined with satellite imagery.

In the future, the methodology developed inthis study may be applied to other regionsaround the globe and to a wider range ofpollutants and media. Projects could includewater pollution monitoring, local air pollutionanalysis, or analysis of specific global policyissues. The benefits include a greater under-standing of important environmental issues andan increasing ability to clearly visualize theimpact. Ultimately, we hope to encourage amore collaborative relationship between thesatellite and ground-based monitoring scientificand policy communities.

vii

TABLE OF CONTENTS

Executive Summary............................................................................................................................................... iii

1. Objectives and Approach.............................................................................................................................. 1

2. Project and Literature Search ...................................................................................................................... 3

2.1 Background ..................................................................................................................... 3

2.2 Search Results .................................................................................................................. 5

2.3 Search Conclusions .......................................................................................................... 15

3. Methodology ................................................................................................................................................... 17

3.1 Step A. Define Project ...................................................................................................... 18

3.2 Step B. Find Appropriate Partners And Resources ............................................................ 19

3.3 Step C. Select Data Sources ............................................................................................. 21

3.4 Step D. Apply Techniques For Electronic Communications ............................................. 23

3.5 Step E. Conduct Project ................................................................................................... 24

4. Pilot Project ..................................................................................................................................................... 25

4.1 Project Description .......................................................................................................... 26

4.2 Partners And Resources .................................................................................................... 27

4.3 Data Sources .................................................................................................................... 28

4.4 Communication .............................................................................................................. 28

4.5 Project Implementation ................................................................................................... 29

4.6 Pilot Project Results ......................................................................................................... 29

4.7 Pilot Project Conclusions ................................................................................................. 42

5. Conclusions and Future Directions ............................................................................................................. 43

5.1 Conclusions ..................................................................................................................... 43

5.2 Future Directions ............................................................................................................. 43

Annotated References....................................................................................................................................... 45

Appendix A: Potential Relevant Satellites ..................................................................................................... 59

Appendix B: Methods For Remote Sampling Of Aerosols .......................................................................... 63

1Transboundary movement of atmosphericpollutants has international policy, economic,human health, and environmental ramifications.Atmospheric pollutants, such as aerosols, persis-tent bioaccumulative toxics, and gaseous pollut-ants, have significant impact on human andenvironmental health. A new generation ofground monitoring systems in connection withnew satellite imaging systems provides an oppor-tunity to investigate, design and implementeffective monitoring strategies for these atmo-spheric pollutants.

Atmospheric pollutants are of particularconcern since air masses flow freely acrossborders, leaving the geographic and politicaljurisdiction of the originating country andbecoming the responsibility of another. Forexample, sulfur dioxide emissions from oneindustrial region may be transported hundreds ofmiles and ultimately deposited as acidic com-pounds into a neighbors ecosystem. Wind blowndesert dust and forest fire smoke cross interna-tional borders and increase particulate matterconcentrations to levels that may exceed regula-tory standards and harm human health. Thestable chemical properties of persistent organicpollutants (POPs) promote their long rangetransport and their ability to bioaccumulate,which may increase toxicity in environmentswhere they have never been used or produced.

A fundamental question associated withtransboundary pollutant transport is how toeffectively monitor pollutant movement. Ground-based sensors can monitor conditions at specificgeographic points and times on either or bothsides of a political border but they provide alimited picture of pollutant sources, receptors,and the path they took to get from one to theother. They provide a particularly limited view,especially when large water masses separate the

countries involved in the source-receptor relation-ship. Integrating satellite images with pointmonitoring can fill in the spatial and temporalgaps. An integrated monitoring effort can aid thetracking of pollutant plumes, early detection andadvance warning systems, identification ofpollutant sources, and the general knowledge baseof pollutant physical and chemical characteristics all of which can be translated into informationuseful for negotiating international policies.

The U.S. Environmental Protection Agency(EPA) and the United Nations EnvironmentProgramme (UNEP) Global Resource Informa-tion Database (GRID) office in Sioux Fallsformed a small team to implement a joint projectrelated to transboundary movement of pollutants.Our main objective was to develop and verify amethodology to assess and monitor the movementof pollutants across international boundariesusing a combination of ground-based monitoringand space imaging data. The implementation ofthis project involved three general tasks:

1. Reviewed the science and current activitiesin the combined use of remote satelliteimages and ground-based monitoring datafor transboundary pollutant movement

2. Developed a general methodology touse integrated spaceborne andground-based datasets

3. Demonstrated the methodology through apilot project.

Due to the expertise and interests of theagencies and staff involved, the project focused onair pollutant transport, while attempting toremain general enough to be applicable to a widerrange of pollutants and regions.

This document represents the achievement ofabove objective and presents the results of thethree tasks.

1. OBJECTIVES AND APPROACH

3The initial task was to document past and currentprojects, datasets, and technologies that inte-grated data from some combination of satellite,aircraft and ground sensors. The immediategoal of the review was to guide the selection ofa pilot project that could be used to test therobustness of our general methodology forusing integrated datasets. The task was con-ducted within the context of how these kindsof projects could be used to assess and monitortransboundary movement of pollutants.

2.1 BACKGROUND

The most obvious model of integrated use ofspaceborne and in situ ground-based data wasthe National Weather Service. The NationalWeather Service predicts and reports weatherinformation consisting of a blend of weathersatellite images with ground-based data such aswind speed, precipitation, barometric pressure,and temperature. In a few other cases, ground-based data has been collected specifically toverify and refine satellite data models. How-ever, besides these notable exceptions, theexperts in remote sensing and the experts inground level monitoring have not consistentlycommunicated or worked closely with eachother, especially with time-relevant environ-mental or human health data.

Based on preliminary discussions with theexperts in these fields, we determined thatthese collaborations would be useful and wouldbe facilitated by three recent developments:

Rapid advances in the quality and availabil-ity of satellite images from government andprivate space organizations;

Expansion of ground-based monitoringnetworks by government environmentalagencies as well as development of better

on-line monitoring devices for a widervariety of constituents; and

Increased ability for rapid electroniccommunication of data and images andthe expansion of monitoring informationavailable on the Internet.

Recent developments in satellite remotesensing technologies and improved accessibilityto satellite data have resulted in the ability toobserve major ecological events around theworld on a daily basis. Scientists and regula-tors have observed very diverse events, fromdust storms in the Sahara and China, to denseindustrial and urban haze in the United States, toforest fires in Mexico, to algae blooms frompollution in the Mediterranean. Additionally,ground-based monitors have detected potentialcontaminants from cross-boundary events,indicating the existence of long-range atmospherictransport of various pollutants. These monitoringdata indicate that global events may be importantcontributors to the total environmental concen-trations of these pollutants in many countriesaround the world.

The technological developments to monitorthese events are so recent that, when combinedwith the isolation of the respective communities,they have not translated into routine use ofintegrated datasets. Based on this preliminaryliterature search, the advances in technology, andthe interest expressed by colleagues, there exists aclear multi-community interest in further investi-gation into the use of integrated datasets thatcontain both spaceborne remote sensing imagesand ground-based data.

Since the integrated use of spaceborne andground-based datasets was too large a subject toapproach with available resources, we sought tofocus our study on a single subject area that could

2. PROJECT AND LITERATURE SEARCH

4serve as an example. Some subjects, such asweather, natural disasters, and landuse change,were eliminated since they are already beingintensely monitored and studied at a number oforganizations internationally. Two other areasemerged as possibilities: water quality monitoringin a watershed (in particular algae blooms andsediment plumes) and air quality monitoring ofthe transport of dust and other particulates. Wechose to focus our study on the transboundarymotion of air pollution, in particular, aerosols/particulates and toxics.

Air pollution monitoring was selected forseveral reasons. Air pollutants, generated locallyand transported long distances, have a significanteffect on human health, particularly those with

asthma, the elderly, and children. Additionally,the transport of air pollutants is a significantcontributor to acid rain, poor visibility, climatechange, and bioaccumulation of toxics in remoteareas. Thus, air pollutant monitoring is animportant issue for both human and environmen-tal health on a global scale. UNEP is involved inforty one air related treaties as documented inECOLEX (http://www.ecolex.org), and many ofthese are concerned with the long distancetransport of airborne pollutants. The U.S. EPA isa party to many of these treaties. On a practicallevel, the transboundary movement of pollutantsmatched the interests and the skills of the staffinvolved with the project.

Figure 2-1. Landsat 7 Thematic Mapper Image of Vancouver Island, Canada, showing how aerosols can be enhanced or removedfrom the image. Image courtesy of Robert Crippen, NASA, 2000.

52.2 Search Results

The search focused on available spaceborneimages, available ground-based data includingaircraft-based data, modeling information, andprevious projects that integrated data from thesesources. This information was used to develop ageneral methodology for using integrated datasetsand to conduct a pilot project in the area of airquality (aerosols and toxics).

2.2.1 Available Spaceborne Images and Data

Hundreds of organizations use remote sensingdata that are provided by a short list of satelliteoperators, such as the U.S. National Aeronauticsand Space Administration (NASA), the EuropeanSpace Agency (ESA), the National SpaceDevelopment Agency of Japan, and the IndianSpace Research Organization (ISRO). Theorganizations and projects in the Referencessection represent a relevant sample, although notan exhaustive list. Similarly, Appendix A (page59) lists a few of the satellite sensors includingdetails about their technical abilities that arepotentially relevant to this work. Appendix B(page 63) lists some of the methods usedto retrieve aerosol measurements fromsatellite observations.

Many of the satellite images are increasinglyavailable internationally (with the exception ofmilitary information). Some agencies make dataaccess easier and more cost effective than others.Also, obtaining data easily usually requires high-speed access to the Internet. The most relevantsatellite sensors and images are discussed in moredetail below.

2.2.1.1 Landsat

Landsat is one of the longest running series ofearth observation satellites. The first Landsatsatellite was launched in 1972 and the mostrecent, Landsat 7, was launched in April 1999.The strength of Landsat is its data quality andexcellent spatial resolution, while its mainweakness is that an image for any particularregion is available only every 16 days.

Typically, Landsat data are used to monitorland use changes and to make land maps(Tmmervik et al 1998; De la Sierra et al 1995).However, Landsat data has been used for morediverse functions such as monitoring algae blooms(Rud and Gade 1999). Landsat images have beenused for a variety of air monitoring projects bymeasuring the differences or changes inreflectively (Otterman et al 1982). This includesstudies of air quality, in particular SO

2 and

particulates (Retalis et al 1999; Sifakis et al 1999;Deschamps and Sifakis 1992).

An example of Landsat data and how it can beused to find and visualize haze is shown in Figure2-1. This image from Vancouver Island usesLandsat Thematic Mapper Band 6 data to isolateareas of haze (Crippen and Blom 2000; Crippen1999). This image shows how haze can beremoved from the Landsat image for a better viewof the land, or how it may be enhanced in orderto better view the aerosols. This technique canbe used as a qualitative view of aerosols in theregion or used to validate ground-basedmeasurements of visibility or particulate matter.

2.2.1.2 Advanced Very High ResolutionRadiometer (AVHRR)

One of the most commonly used sensors foraerosol retrieval thus far is AVHRR. AVHRRtechnology has flown on various NOAA polarorbiting satellites since 1978, with NOAA-14 andNOAA-15 currently in orbit. Polar orbitingsatellite systems offer the advantage of daily globalcoverage, by making polar orbits roughly 14.1times daily, with the local solar time of eachsatellites passage essentially unchanged for anylatitude.

AVHRR collects data in the visible, near-infrared, and thermal infrared portions of thespectrum. Thus it is often effective to use severalbands of AVHRR data to analyze the same image.AVHRR data has also been used for a variety ofprojects, including monitoring of algae blooms(Chavez et al 1999; Rud and Gade 1999) andsediment levels in watersheds (Walker 1996;Woodruff et al 1999). AVHRR data have been

6Figure 2-2. Dust Storm in Texas, 14 December1999, with the second image enhanced by subtract-ing Channel 5 AVHRR data from Channel 4 toenhance silicates in the air. Image courtesy of theNOAA OSEI team.

used for aerosol monitoring; the OperationalSignificant Event Imagery group at NOAA usesAVHRR data for monitoring significant events ona daily basis. Figure 2-2 is an example of AVHRRimagery showing a dust storm in Texas. It is alsoan example of using different bands of data fromthe same satellite to enhance an image. Thesecond image was created by subtracting oneAVHRR data band from another in order toenhance the brightness of the silicates in the air.

An interesting use of AVHRR data is theAVHRR Pathfinder Atmosphere (PATMOS)project, a joint NOAA/NASA project to develop aclimatology of aerosol optical thickness (AOT)using AVHRR data from polar satellites from1981 to the present. AOT is a measure of theeffect of aerosol particles on the transmission of

solar radiation to the Earths surface; thus, thehigher the AOT, the less solar radiation reachesthe ground. The NOAA PATMOS team hasalready developed a climatology over oceans,where the background reflectance is much lowerand more stable than that of land, making theretrieval of AOT more reliable. They are nowbeginning a similar project, working in conjunc-tion with NASAs Global Aerosol ClimatologyProgram, to use AVHRR data to develop aclimatology over land.

2.2.1.3 Geostationary Operational EnvironmentalSatellites (GOES)

NOAA operates GOES primarily as weathersatellites. They are in a geosynchronous orbit onEarths equatorial plane, matching exactly the

7Earths rotation about its axis. This configurationallows each satellite to view the same areas of theEarth at all times from about 35,800 km(22,300 miles) above the Earths surface. Unlikethe polar orbiting satellites, the GOES satellitescan provide continuous monitoring of the Earthsatmosphere and surface over a large region of theWestern Hemisphere.

Besides weather prediction, GOES data hasbeen used for air quality monitoring, includingdetermining aerosol optical thickness (Fraser et al1984) and fire and smoke detection (Hotz 1998).The NOAA Operational Significant EventImagery group and others use GOES data forboth dust and fire monitoring.

A similar geostationary satellite sensorlaunched by the European Space Agency andoperated by Eumetsat is Meteosat, which collectsimages in both visible and infrared wavelengths.Meteosat is designed for weather observations but

has been used for aerosol optical thicknessretrieval. Additional satellites and sensors arebeing planned for Meteosat Second Generationthat will include expanded spectral coverage.

2.2.1.4 Total Ozone MappingSpectrometer (TOMS)

TOMS has been in use since 1978 on the Nim-bus-7 platform, scanning at ultraviolet wave-lengths. TOMS is most well known for mappingozone, including monitoring the Antarctic ozonehole and tropospheric ozone (Fishman and Balok1999; Fishman and Brackett 1997). However,TOMS is sensitive to absorbing aerosols and canbe used to monitor the motion of large aerosolplumes. TOMS has a very low resolution (50km), so it is most useful for monitoring events ona global or regional scale, rather than local. Figure2-3 is an example of global TOMS imageryshowing dust from the Sahara blowing across theAtlantic Ocean.

Figure 2-3. TOMS images showing dust blowing from the Sahara across the Atlantic. Image courtesy of the TOMSwebsite and NASA Goddard Space Flight Center.

82.2.1.5 Sea-viewing Wide Field-of-viewSensor (SeaWiFS)

SeaWiFS was designed as an ocean color sensor tocollect sea surface color and other ocean bio-optical properties. It is used extensively forobserving algae blooms, tracking oil spills,monitoring water pollution, among many otheruses (Chavez et al 1999; Woodruff et al, 1999; seealso the SeaWiFS website). However, its dailyvisible color images can provide some strikingimages of other non-ocean events, such as duststorms and smoke (Hotz 1999; see also theSeaWiFS website). Figure 2-4 is an exampleof a smoke image that was captured bySeaWiFS. In Figure 2-5, SeaWiFS reveals denseindustrial pollution.

2.2.1.6 New Satellites

The space agencies have launched or plan tolaunch several new earth observing satellites thatwill greatly increase the types of data available formonitoring the earths environment. Of particu-lar note are the instruments on the Terra satellitelaunched in December 1999, which have data

Figure 2-4. SeaWiFS image showing smoke from fires in Mexico, 5 June 1998. Image courtesy of theSeaWiFS website, NASA Goddard Space Flight Center.

Figure 2-5. SeaWiFS image showing industrial pollutionin eastern China, 2 January 2000. Image courtesy ofthe Visible Earth website, NASA Goddard SpaceFlight Center.

UrbanHaze

TianjinBeijing

Clouds

CHINA

Shanghai

Smoke

MEXICO

Clouds

9available for use in late 2000 and 2001. TheNASA Goddard Space Flight Center manages theTerra program.

One instrument launched on the Terra satellite,Multi-angle Imaging Spectro-Radiometer(MISR), uses 9 cameras to image each piece of theearth from nine angles, providing the ability to seein 3-dimensions, as well as distinguish andhighlight haze, dust, plumes, clouds, and otherevents. Figure 2-6 is a MISR image on March 6,2000, showing images from two angles andaerosol data derived from the oblique angle data.

A second instrument on Terra, Measurementsof Pollutants in the Troposphere (MOPITT),measures carbon monoxide and methane in thetroposphere, thus will directly measure ground-level pollutants. Figure 2-7 shows an early imagefrom MOPITT, of carbon monoxide over theUnited States, including the beginnings of aplume moving over the Atlantic. The NASAEarth Observing System (EOS) AURA satellitemission planned for launch in June 2003 willenhance the capabilities of MOPITT.

Several other instruments on Terra will providesignificant imagery and data as NASA and its

Figure 2-6. MISR Images and Data of Appalachians,March 6, 2000. The first panel is the downward-looking(nadir) view camera. The middle panel is forward-viewingat 70.5-degree camera. At this increased slant angle, theline-of-sight through the atmosphere is three times longer,and a thin haze over the Appalachians is significantly moreapparent. The third panel shows the airborne aerosolamount, derived using new methods that take advantage ofMISRs moderately high spatial resolution at very obliqueangles; gradations from blue to red indicate increasingaerosol abundance. Image and explanatory text courtesy ofNASA Jet Propulsion Laboratory,http://visibleearth.nasa.gov/cgi-bin/viewrecord?5898.

Figure 2-7.MOPITT CarbonDioxide Data,March 5-7, 2000.Note plumemoving off easterncoast. Imagecourtesy of theMOPITTInstrument Team,NASA VisibleEarth and theCanadian SpaceAgency, http://visibleearth.nasa.gov/cgi-bin/viewrecord?551

10

partners begin to refine analysis methods andmake data sites operational.

2.2.2 Available Ground-based Data

Many government environmental organizationscollect and use ground-based data. In the UnitedStates, the most important organizations for airpollutant related data are EPAs Office of Air andRadiation (OAR) and the OAR Office of AirQuality Planning and Standards (OAQPS).

OAQPS maintains databases of air monitoringdata collected by EPA and the state governmentsfor compliance purposes. The best single sourceof ambient data, updated on a monthly basis, isAerometric Information Retrieval System (AIRS).AIRS is an electronic database with air qualitydata from monitoring stations throughout theU.S. collecting data about the criteria pollutants(CO, NO

2, O

3, Pb, PM

10, PM

2.5, SO

2), which

have a large impact on human health. Daily datais not currently available from the website, butthey are stored in the main AIRS databases andare available on request. Expected upgrades to thewebsite will allow access to the full set of data viathe web. Currently, a number of individual stateshave air quality data available in real-time ontheir own websites.

There are some research databases that couldbe tapped depending on project requirements.One example is the Indian Ocean Experiment(INDOEX) database for 1995-1999. The limitedavailability of INDOEX data precludes theiroperational use, but may be helpful for validatingalgorithms employed in projects using themethodologies discussed in this report.

Regulatory agencies throughout the world aremaking their data available to researchers and tothe public. Ground-based data are increasinglyavailable from major government sources,although in some countries, access to data may beimpeded by administrative and national limits onthe distribution of information. While thefollowing discussion of pollutants is focused onU.S. EPA sources, similar programs exist at somelevel in most countries that regulate air pollutants.

2.2.2.1 Particulate Matter

Particulate matter has impacts on human healthby reducing lung function and is of particularconcern to those with asthma, children, and theelderly. Generally, ground-based monitorsmeasure particulate matter under 10 microns,which is considered inhalable, or fine particulatematter under 2.5 microns, which has the mostimpact on human health.

EPA and other regulatory agencies internation-ally have extensive existing network of PM

10

monitors, which collect information for urbancompliance purposes. For fine particulate matterdata, OAQPS is in the process of establishing anetwork of about 1,000 fine particulate monitors.These data will be available via the AIRS systemas the sites are established. Additionally, a set ofapproximately 54 supersites will also be estab-lished (Mobley and Shaver 1999). The supersiteswill collect very detailed data beyond regulatoryrequirements (such as metals speciation of theparticulate data) in order to gain a more completeview of the sources of air pollution and to advancemonitoring technologies and systems. Thesupersites will be launched in several phases, withthe first set located in Atlanta and Fresno, andPhase 2 in New York City, Baltimore, Pittsburgh,St. Louis, Houston, and Los Angeles.

2.2.2.2 Air Toxics

EPA typically considers toxic air emissions as the188 Hazardous Air Pollutants (HAPs) as definedby the Clean Air Act. A few examples of HAPsinclude mercury compounds, lead compounds,benzene, chlorine, DDE, and many pesticidessuch as chlordane. Human health effects arediverse, including possible links to cancer, chroniceye, lung, or skin irritation, and neurological andreproductive disorders. Environmental effects arealso considerable, for example, toxics can poisonwaterways or bio-accumulate to cause disease,neurological and reproductive disorders in fauna.

HAPs are well documented as point sourceemissions but unlike the criteria pollutants,ambient levels of the HAPs are not measured on

11

World Weather Watch Program. Data from over8000 stations are typically included each monthand are accessible through the NCDC web server.Figure 2-8 shows the locations and density ofthe stations.

The SOD data include visibility observations.The visual range, or visibility, is the maximumdistance at which an observer can discern theoutline of an object against a horizon sky. It canbe used as a surrogate for fine particulate matterconcentrations because of the strong relationshipbetween increased fine particulate concentrationsand decreased visibility.

2.2.2.4 AErosol RObotic NETwork (AERONET)

AERONET is an optical ground-based aerosolmonitoring network and data archive supportedby NASAs Earth Observing System and used bymany non-NASA institutions. The networkconsists of automatic sun-sky scanning spectralradiometers, or sun photometers, that measurethe direct sunlight in specific spectral bands, andfrom which the aerosol optical thickness can bedetermined. The sun photometers are operated by

widespread basis, either in the U.S. since there areno federal requirements for ambient attainment,or worldwide. However, the U.S. EPA is inprocess of developing a national air toxicsprogram, including increased ambient air toxicsmonitoring at sites across the U.S. (Mobley andShaver 1999). As part of this program, 33 urbanair toxic HAPS are identified as the initial prioritypollutants, including volible organic compounds(VOCs), mercury compounds, and other toxicsthat have the greatest impact on urban air quality(Mobley and Shaver 1999). This program buildson current monitoring that already occurs,including 8 VOCS in 20 cities that are alreadystored in AIRS and the near future speciationdata for 10 HAPS metals from over 50 citiesparticipating in the PM

2.5 network (Mobley and

Shaver 1999).

2.2.2.3 Visibility

The global Summary of Day (SOD) databasedistributed by the National Climatic Data Center(NCDC) contains data that are derived from theWorld Meteorological Organization (WMO)

Figure 2-8. NCDC Visibility measurement station location density (Husar et al 2000).

12

a variety of universities and organizations. Datafrom this collaboration provide globally distrib-uted near real time observations of aerosols. Thereal time data undergo only preliminary process-ing, while the final quality-controlled data arefurther reviewed and made available about 6months after data collection. The data providecharacterization of aerosol properties that areunavailable from satellite sensors and help tovalidate the satellite data. Figure 2-9 shows thelocations and density of the AERONET stations.

2.2.2.5 Special Research Studies

Special studies of air monitoring and transportoccur within the research organizations of govern-

ment environmental agencies, universities, andnon-profit institutes. These projects typicallycollect detailed information in a specific area for alimited amount of time. Often these studies canbe tapped as sources of additional information orsupported with satellite images. For example, theIndian Ocean Experiment (INDOEX) wasdesigned to study regional consequences of globalwarming due to the cooling effect of aerosols.These tiny particles, about a micron or smaller indiameter, scatter sunlight back to space and causea regional cooling effect. These aerosols consistingof sulfates, soot, organic carbon and mineral dustare produced both naturally and by humanactivities. Still, the complex influence of

Figure 2-10. Modeling study showing the concentration of a hypothetical tracer released in China as it moves across thePacific. (Hanna et al 1999; Keating 1999) Image courtesy of Joe Pinto, EPA ORD/NCEA.

Figure 2-9. AERONET sun photometer site locations (map courtesy of Aeronet NASA GSFC).

13

motion of global aerosols in near real-time(NRL 2000).

Some models are modified and applied tospecific subjects. A relevant example is a model-ing study conducted by the U.S. EPA and theNorth Carolina Supercomputing Center toanalyze the motion of air toxic pollutants fromChina to North America. Figure 2-10 shows themodeled concentrations of a hypothetical tracerreleased in China and stretching across the Pacificin varying colors according to a logarithmic scale(Hanna et al, 1999).

2.2.5 Previous Integrated Work

Projects in the published and Internet literaturehave used integrated datasets composed of bothsatellite remote images and ground-based data.One term applied to the integration of monitor-ing data (both spaceborne and ground-based) isdata fusion (Pohl et al 1998; Wald 1999). In fact,one website calls itself the Data Fusion Server,although most of its examples are fusing informa-tion from different satellites as opposed to fusionwith ground-based data. Several organizations arededicated to different types of data integrationalthough they also usually specialize in satellitedata integration, including Committee on EarthObservation Satellites (CEOS), an internationalcoordinating body for satellite providers withNASA and NOAA as the U.S. representatives,and the Centre for Earth Observation (CEO)Project, a similar European program.

Data integration has been applied inchlorophyll monitoring (Kester et al 1996; Rudand Gade 1999; Walker 1996; Woodruff et al1999), land use change (De la Sierra et al 1995;Tmmervik et al 1998) and ocean monitoring(Chavez et al, 1999), among other areas.However, the focus of this study is to reviewthe integrated use of data to monitor airpollution transport.

In the earliest work, EPA scientists usedsatellite imagery and airport visibility (real timeparticle monitoring) to track haze moving aroundthe U.S. and out to sea as early as the 1970s(William Wilson, 1999). While there has been

aerosol cooling on global warming is notclearly understood.

INDOEX field studies occurred over thetropical Indian Ocean where pristine air massesfrom the southern Indian Ocean includingAntarctica and urban pollution from the Indiansubcontinent meet. The data collections involvedmultiple aircraft, ships and island stations over theArabian Sea and the Indian Ocean. TheINDOEX dataset spans from 1995-1998, andincludes data from an intense field campaignundertaken during January to April 1999. Furtherdetails about INDOEX may be found athttp;//www-indoex.ucsd.edu/index.html

2.2.3 Aircraft-Based Information

Aircraft and balloons provide the opportunity forobtaining aerosol characteristics as a function ofaltitude. These data are typically available on alimited basis as part of special research projects.Aircraft data can often be a link between ground-based and satellite remote sensing information.For the purpose of this paper, aircraft data areconsidered a type of ground-based data.

2.2.4 Models

Models use ground, aircraft, and satellite data astheir inputs, forming a new dataset that is appli-cable to data integration for understandingaerosol transport. Two examples are:

Hybrid Single-Particle Lagrangian Inte-grated Trajectory (HYSPLIT) 4 Model.HYSPLIT is the newest version of a com-plete system for computing simple air parceltrajectories to complex dispersion anddeposition simulations. It is a result of ajoint effort between NOAA and AustraliasBureau of Meteorology. HYSPLIT modelsthe dispersion of a pollutant both horizon-tally and vertically (NOAA 1999).

Navy Aerosol Analysis and PredictionSystem (NAAPS). NAAPS is a global,multi-component aerosol analysis modelthat combines the current and expectedsatellite data streams with other availabledata. NAAPS predicts and simulates the

14

significant EPA interest since these early studies,very little published work exists. Non-EPApublished work indicated that scientists have usedLandsat data and ground-based data for a broadrange of purposes, for example, to characterizetropospheric desert aerosols over both land andocean (Otterman et al 1982). Work was alsodone to estimate the transport of SO

2 by combin-

ing aerosol optical thickness over land fromGOES satellite measurements with wind vectors(Fraser et al 1984). They noted the need forvalidation through a independent well designedground-based sampling experiment that iscoordinated with satellite observations.

A substantial amount of work related to dataintegration has been done in the 1990s, which isto be expected since the advances in satellite andground-based monitoring have all been fairlyrecent. This work typically falls into three subjectcategories: the monitoring of an Asian andAfrican dust events; fire and smoke detectionespecially in Central America; and the monitoringand verification of urban air quality.

2.2.5.1 Asian and African Dust Storms

Massive dust storms originating in the Gobidesert region of western China have beenmonitored using SeaWiFS satellite images. TheCenter for Air Pollution Impact and TrendAnalysis (CAPITA) at Washington University inSt. Louis monitors and publicly reports whenthese dust storms reach the west coast of the U.S.The data are enhanced through ground-baseddata and other satellite images provided by expertswho accessed the CAPITA web site. TheCAPITA project was conducted using just intime scientific input. It made use of NASA andNOAA images placed regularly onto the Internet.They also used SeaWiFS, GOES, AVHRR, andTOMS data with ground-truthing and limitedparticulate ground monitoring (Husar et al 2001).State and regional EPA staff used these images tomonitor for the dust as it approached and arrivedon the U.S. West Coast.

Aerosol Characterization Experiment-Asia(ACE-Asia) is another project that plans to

monitor trans-Pacific motion of pollutants,although they take a more historical and long-term research based approach. ACE-Asia is aNOAA and National Science Foundation fundedprogram that plans to combine ground-basedmonitors, aircraft flights, and satellite remotesensing to monitor aerosols and their transport inand around Asia. This is a multi-year programthat is just beginning to fund research projectsaimed at a long-term understanding of trans-Pacific pollutant movement.

Scientists have also been interested in the dustmoving from the Sahara and northern Africaacross the Atlantic towards the Caribbean and theAmericas. Only a few studies of this phenom-enon have used satellite and ground-based data inan integrated fashion. One study used satelliteimagery in conjunction with soundings fromships positioned in the Atlantic, combining thosedatasets with wind and weather data through analgorithm (Ott et al 1991). Another studyconducted by the University of Miami usedground-based sampling data of airborne aerosolscollected on a small island 4 km east of mainlandFlorida (Prospero 1999). Their sampling dataindicated a peak of aerosols from non-mainlandsources during June, July, and August, consistentwith an African source. These data werecompared with qualitative studies from theliterature that used AVHRR and TOMS data.The University of Miami study discussed how therelationship between the images and the sampleddata could impact EPAs administration of thenew PM

2.5 regulation.

2.2.5.2 Fire and Smoke Detection

Fire and smoke detection are both possible withsatellite imagery. Satellite images from TOMS,GOES, and SeaWiFS have been combined withground data from forest fighters in order tounderstand the scope and the motion of fires andsmoke (Hotz 1998).

Major fires in Central America and Mexicowere monitored in 1998 by CAPITA as the smokemoved into the U.S. Using a website formatsimilar to the Asian dust event, ground-based

15

their discoveries was a large cloud of dense haze,consisting of soot, sulfates, nitrates, organicparticles, fly ash, and mineral dust, which formsover the Indian Ocean. They were able toidentify the haze with satellites and determine itsconstituents with ground-based sampling stationson ships and islands.

Significant interest has been expressed in thetransport of urban pollutants from the east coastof the U.S. to Canada and to Europe. Thesepollutants and their transport are just beginningto be understood, especially the long-rangetransport to Europe, although there is evidencethat they may contribute 30-50% of the urban airpollutants in the North Atlantic (Keating 1999).

The AVHRR Pathfinder Atmosphere(PATMOS) project is developing a climatology ofaerosols over land over the next 2 years. Thisinformation will likely be correlated as much aspossible with any existing ground-based data.

2.3 SEARCH CONCLUSIONS

The results of this literature and project searchconfirm that the integrated use of satellite imagesand ground-based data is possible and useful forenvironmental monitoring and assessment. Therecent advances in satellite technology, the growthof ground-based air monitoring networks, and theincreased on-line accessibility of satellite sensorimages and ground-based observations supportthe effective implementation of integratedprojects. Their newness combined with theinstitutional separation of scientists in their fieldsand their countries has prevented the extensiveintegration of datasets in both research and real-time monitoring. The dialog started by thisliterature and project search confirms the interestof many researchers and policy staff in theintegration of these fields.

Based on these positive results, the followinggeneral methodology for using integratedspaceborne and ground-based datasetswas developed.

information was submitted by experts. Thisincluded data from the PM

2.5 network from four

cities in Texas (see CAPITAs website on CentralAmerica http://capita.wustl.edu/Central-America). These images and PM data were laterused by EPA to evaluate the compliance status ofcities who were most affected by the smoke.

The NOAA National Weather Service Interna-tional Activities Office, in conjunction withNOAA NESDIS and other offices, has a projectcalled Program to Address ASEAN RegionalTransboundary Smoke (PARTS), which installedan integrated forest fire monitoring network insoutheast Asia. This network will include accessto polar orbiting satellite imagery, ground-basedatmospheric monitoring (meteorology, air sam-pling, optical depth), and a computer-basedmodel to provide the capability to 10 countries insoutheast Asia to model and monitor fires andother aerosol events in real-time. This project isbeing piloted in early 2000.

2.2.5.3 Urban Air Quality

Several studies have been conducted in Athens,Greece to determine urban air quality (Retalis etal 1999; Sifakis et al 1999; Deschamps and Sifakis1992). Typically, the studies used Landsat imagesfrom one clear and one polluted day to quantita-tively determine the aerosol distribution in thecity. They also compared the satellite derivedaerosol values to a network of ground-based airmonitors that measure CO, NO

2, O

3, SO

2, and

particulates. They found a correlation betweenthe satellite-measured optical density and both theSO

2 and particulate ground-based measurements.

The Ecole des Mines de Paris, Center for EnergyStudies, Remote Sensing and Modeling Grouphave proposed similar work for Nantes, France.

A large international research project calledIndian Ocean Experiment (INDOEX) monitoredaerosols and urban haze over the Indian Oceantropical region, primarily in relation to climatechange. They used integrated data from 4aircraft, 2 ships, 8 satellite platforms, and numer-ous ground stations (Nguyen et al 1998). One of

17

The methodology was developed in a manner thatwould provide any interested individual ororganization with a basic framework for develop-ing and implementing a project that integrates theuse of ground-based data and spaceborne images.The methodology is not intended to be compre-hensive; rather, it is designed to be flexible enoughto be used by general scientists and policy makers

3. METHODOLOGY

to measure many types of environmental phe-nomena with integrated datasets. It was tested andrefined through application to a pilot project andwe used the assessment of transboundary motionof air pollutants as a basis for its development.Table 3-1 outlines the methodology that is thendescribed in detail.

Table 3-1. Outline of Methodology

Step A. Define projectA.1 Define project goal and objectivesA.2 Define the information of interestA.3 Conduct a literature searchA.4 Determine if the information of interest is observable with satellite imageryA.5 Determine if the information of interest can be monitored at ground levelA.6 Determine temporal resolution requirementsA.7 Select a specific geographic region of interestA.8 Determine if qualitative or quantitative data is needed or bothA.9 State your expected outcomes

Step B. Find appropriate partners and resourcesB.1 Find appropriate remote sensing partnersB.2 Find appropriate ground-based data partnersB.3 Address resource availability issues

Step C. Select data sourcesC.1 Select satellite(s) and the appropriate channelsC.2 Select ground-based monitoring measurement systemsC.3 Ensure compatibility of datasetsC.4 Develop methods for quantitative data analysis

Step D. Apply techniques for electronic communicationD.1 Collect data and images from existing on-line databasesD.2 Use servers and websites to share large imagesD.3 Use e-mail to notify others of new developments or data availability

Step E. Conduct projectE.1 Secure resourcesE.2 Finalize roles and responsibilitiesE.3 Develop project planE.4 Launch projectE.5 Produce deliverables and monitor the projectE.6 Complete the project and celebrate with your team

18

A.4 Determine if the information of interest isobservable with satellite imagery.

Generally, a phenomenon can be monitoredwith satellite instruments if it creates a plume orobscures visibility (in air or water), changes thethermal reflectance at specified spectral wave-lengths, alters the reflectively of the atmosphere(by changing its constituents), or is a visible land-based event. Often, images and data need toundergo significant analysis in order to observethe event clearly and correctly.

At this early point in the project, conduct aninitial review of the literature and availablesatellite sensors, especially how investigators haveused them in other projects. Many phenomenathat at first glance seem to be invisible tosatellite imagery can be monitored now or in thenear future using new sensors, combinations ofsatellite channels, or other spectral analysistechniques. Also, some events can be monitoredby satellite sensors that, when combined withground-based data, can provide qualitativeindicators for the information of interest. Forexample, monitoring the transport of particulates,combined with ground-based analysis of theparticulate constituents, can provide informationon the source of air toxics that cannot be seendirectly with satellites alone.

A.5 Determine if the information of interest can bemonitored at ground level.

Ground level data can be collected by your owninstruments or can come from an existing net-work. Weather related information is monitoredworldwide by organizations like World Meteoro-logical Organization. Related meteorologicaldata, such as visibility, are also available at arelatively high spatial resolution. Environmentalagencies in most countries and cities monitorground level ambient data, typically includingparticulates, SO

2, NO

X, lead, ozone, and volatile

organics, and sometimes soot and a variety oftoxics. These same agencies also usually monitorfor water quality in certain water systems, includ-ing temperature, turbidity, dissolved oxygen, algae

3.1 STEP A. DEFINE PROJECT

Integrated use of satellite images and ground-based data can be a valid and productive methodto study many environmental phenomena. Whilenot utilized extensively to-date, researchers havesuccessfully used joint datasets to study chloro-phyll and algae blooms, land use change, volca-noes, weather conditions, and general oceanmonitoring, among many other environmentalphenomena. Joint datasets have also been used tostudy urban air pollution (SO

2 and particulates)

for specific cities, movement of forest firesplumes, and trans-Pacific and trans-Atlantic duststorms. Current research efforts are underway toincorporate satellite sensors and ground monitor-ing stations to study other air constituentsdirectly, such as toxics, ground level ozone, andozone precursors.

Therefore, when developing a new monitoringproject, it is important to review how satelliteimagery and ground-based data can be usedtogether and to define the project boundaries tobe able to successfully use both. The followingsteps will help define the scope of the project.

A.1 Define project goal and objectives.

Define the purpose of conducting the project,what the project will accomplish, and what isoutside of the project scope.

A.2 Define the information of interest.

Examples include water turbidity, algae blooms,dust storms, forest fire plumes, criteria air pollut-ants, aerosols, toxic air pollutants, and manyothers. To the extent possible, define specificmeasurement and precision requirements.

A.3 Conduct a literature search.

Review the literature, the Internet, and othersources of information about projects that may besimilar to yours. This will provide information tohelp you: define your project, determine feasibil-ity, find partners, find data sources, understandexisting methods of integrating those data sources,and avoid reinventing a project or activity.

19

blooms, and sometimes toxics. All of these dataare typically available from national organizations,as well as state and local organizations. Finally,government, academic, and private organizationsconduct a wide variety of research projects thatcollect specific data for certain areas or events.

A.6 Determine temporal resolution requirements.

How rapidly the information you are collectingchanges over time will directly influence thesources and quality of data. Depending if condi-tions change hourly, daily, monthly, seasonally, orannually, different satellites and ground sensorswould be used to characterize the environmentalparameter you are interested in. Geostationary(geosynchronous) satellites often measure a largeregion several times an hour while some polarorbiters measure a region twice daily. High-spatial resolution land monitoring satellites maypass over an area on the order of only every twoweeks. Ground-based data are also collected overa wide variety of time periods, from hourly todaily to every few weeks.

Also important at this stage is to determine ifyou are interested in near real-time data in orderto make immediate decisions, or if long-termseries (historical) data are desired to analyze longertrends or past events, or both. How often yourdata are collected will also influence the ways youshare and archive this data. Determine the rate ofchange and timing you are interested in order toselect the appropriate data sources later.

A.7 Select a specific geographic region of interest.

The geographic region of interest should bedefined using process-based criteria as opposed toonly political or economic criteria. In order toremain manageable, most projects focus on aspecific area, such as a certain city, a watershed, aborder region between two countries, an ecosys-tem, or a trans-ocean region. Exceptions to thisare when pollutants have an essentially globalimpact, for example chlorofluorocarbons orgreenhouse gases. Others, such as persistentbioaccumulative toxics, may be emitted fromspecific locations, but transported long distances

and concentrated in areas far from their source. Ifyou are interested in the influx of pollutants to acertain region, then the surrounding areas at aconsiderable distance may also be of interest.Choose a region of a size that is most informativeto the event or pollutant you are interested in, yetmanageable with available monitoring. Keep inmind that once a project is completed in a specificlocation, the methods used there can often beexpanded to similar regions elsewhere.

A.8 Determine whether qualitative or quantitativedata are needed or both.

For some projects, general satellite images are allthat are needed; in other cases, specific processeddata such as the aerosol optical thickness arerequired. Qualitative data are easier to obtain,although it still requires enhancement and skilledinterpretation of the satellite images. Mostground-based data are more quantitative,although simple visual observations on theground can be useful, such as observations of fogor haze. Initial consideration should be giventhe comparability of data, covered in more detailin section 3.3.

A.9 State your expected outcomes.

Based on all this input, clearly define thedeliverables and the outcomes you hope toachieve by the end of the project. Relate thoseoutcomes to a human or environmental impact.

3.2 STEP B. FIND APPROPRIATE PARTNERSAND RESOURCES

By its very nature, using integrated spaceborneand ground-based datasets is a multi-disciplinaryactivity. Therefore, it is important to find theappropriate people and partners to conductvarious elements of the work. One difficulty inachieving this is overcoming the barriers withinand between technical fields and organizations.These barriers include lack of knowledge aboutthe abilities of other fields, how to find the rightpeople in the right organizations, issues related tofunding and resource availability, and otherphysical and institutional barriers.

20

Aerospace Center, Deutsches Zentrumfr Luft- und Raumfahrt (DLR), whichmanages the German satellites anddata collection.

Canadian Space Agency (CSA) and theCanada Centre for Remote Sensing (CCRS)in Natural Resources Canada CSA hasseveral satellites of interest and remotesensing data are collected and used bythe CCRS.

Other organizations that should be consideredas potential partners are international monitoringgroups. Examples include the UNEP/GRIDCenters, the Center for Earth Observations, andmany academic institutions and universities. Thebest way to find these organizations is to reviewthe literature and search the Internet to findothers that have published work similar to yourarea of interest.

Many of the government and internationalorganizations that provide satellite data aredescribed in more detail in the Research Resultssection. Private companies are another resourcefor satellite remotely sensed data. Althoughengaging private companies can be a costlyoption, if high spatial resolution images (1 km orless) are necessary for your project, they shouldbe considered.

When looking for partners from governmentorganizations and monitoring groups, start withthe organizations in your country since they willlikely be the easiest partners at first. Next, choosethose that have placed their information for easyuse by researchers on the Internet. Finally, giventhe ability to share data and information via emailand the Internet, there is no need to limit yourproject to domestic partners, so internationalorganizations, scientists, and partners canbe sought.

B.2 Find appropriate ground-based data partners.

Unlike satellite data, ground-based informationcan be available from all levels of government(federal, state, local), from organizations withdiverse missions, and from many university and

Finding and recruiting partners will be linkedto the next step of selecting data sources, sinceyou want partners who possess the experiencewith and the access to the required data. How-ever, finding available people with the rightexpertise is of primary importance, since theycan lead you to the data themselves, no mattertheir source.

B.1 Find appropriate remote sensing partners.

A few private companies and several nationalspace organizations, such as U.S. NationalAeronautics and Space Administration (NASA),the European Space Agency (ESA), and theNational Space Development Agency of Japan,conduct the actual launch of the majority of earthmonitoring satellites. However, the operation andcollection of data from these satellites is done bymany organizations. In the U.S. government,major satellite information sources include:

NASA Coordinates the overall earthobservation programs and almost all types ofsatellite data of various kinds.

NOAA Collects and analyzes weather,ocean, and air related data.

USGS Collects and analyzes land pro-cesses including land cover, geology, land-forms, hydrology, biodiversiy and naturalhazards.

Other Government Organizations Processdata, produce data products, and dissemi-nate required products to internal andexternal clients.

Internationally, many satellites and organiza-tions exist and a significant amount of data fromthese satellites is available. Three of the manyorganizations include:

European Space Agency (ESA) The ESAlaunches and maintains satellites, collectsand stores data, and also uses other organi-zations to operate and collect the data fromits satellites, including Eumetsat, whichoperates Meteosat and other satellites.

German Remote Sensing Data Center(DFD) DFD is a part of the German

21

academic-type research institutions. For example,these data could be from national regulatory airmonitoring networks, airport visibility databases,local water quality monitors, or research projectson specific subjects. The source of data that youchoose depends on the size of the region and thetime period of interest. Often several ground-based datasets can be used, such as single data setscollected for a specific project (such as, an air orwater sampling campaign during a pollutionevent), combined with data from an on-goingexisting network (such as, a network of airmonitoring stations or ocean buoys).

Other examples of sources of ground-baseddata are described in the section 2.2. The impor-tance of searching the literature and the Internetcannot be overstated when seeking partnershipsand data sources. Data are scattered widelyamong diverse groups and organizations, thusthey must be actively sought out.

B.3 Address resource availability issues.

Once data sources have been found and potentialpartnerships identified, the availability of re-sources must be addressed. A significant amountof data is available on the Internet or fromgovernment service-type organizations at little orno cost. However, having the resources to analyzeand process that information is what makespartnerships essential. Some organizations arewilling and able to partner if your work is similaror additive to work they are conducting. Mostorganizations are always interested in well-defined, interesting projects that could use theirunique skills.

Another resource option is volunteers, who arepeople or organizations that have specific skills orinterests, but may lack the resources to dedicateinterested personnel to your project. Signedagreements (e.g., memorandum of understanding)or more informal written agreements are twoapproaches to overcome these resource barriers,especially if the agreements are specific about theroles and responsibilities of all parties, thedeliverables, and other details. Seeking jointfunding or sharing available funding is another

approach. Regardless, it is important that theissue of resources, including funding, be addressedwhen partnerships are being developed andmonitored carefully as the project progresses.

3.3 STEP C. SELECT DATA SOURCES

The next step for the team (the project managerand partners) is to more thoroughly explore thesources of available information and data, fromboth spaceborne satellites and ground-basedmonitors. It is also important to characterizerelevant datasets. A well-defined project scopewill make data selection easier.

There are two general kinds of data thatyou might consider adopting, at least initially.They are:

real-time data

time-series (historical) data

Real-time data are needed if decisions madewith that data would influence rapidly changingsituations or would assist in the timing of sam-pling, such as increasing ground-based particulatesampling during a forest fire or sending a ship tosample an algae bloom identified with a satellite.Time-series (historical) data are used for change ortrend analyses through a review of past conditionsor a past event. They can also be used to plan forfuture real-time monitoring.

C.1 Select satellite(s) and the appropriate channels.

Using the data sources that you investigated whenfinding partners and resources, select the specificsatellites and channels (wavelengths) that willwork for your project. Often, multiple satellitescan be used to monitor and analyze the sameevent, each providing different information andinsight, such as using AVHRR to visualize thetransport of a dust plume and TOMS to quantifythe aerosol optical thickness. Also, differentchannels of the same satellite can be used toenhance images; for example, the subtraction ofone channel from another can sometimes allowdust to be more easily distinguished from clouds.More information on these techniques is availablein section 2.2.

22

satellite data quantities exist within theatmosphere, i.e., are they gas concentrationswithin the stratosphere, the troposphere, onthe ground, or are they cumulative ofmultiple layers.

Temporal frequency: Ensure that yourimages and ground data are taken atcomparable times. It is important to under-stand the relationship between the time anddate of ground samples and features in theimages. For example, if you are usingLandsat data that takes a single image ofyour city once every 16 days, and yourparticulate monitor collects data every 6days, it is important that those monitoringdays correspond and their relationshipis understood.

Temporal duration: Duration is importantwhen the data collected from groundmonitors may be an 8-hour or multi-dayaverage, but a satellite image is a singlesnapshot. Understand and carefully con-sider how conditions change over time andhow they compare to a single or a series ofdiscreet images.

Application: Compare the qualities of whatis being measured. For example, somesatellites measure aerosols under 1 micron,but ground-based monitors may measureparticulates under 10 or 2.5 microns.

These data characteristics define why youcannot readily assume that all of your datasets willmatch in space and time. This does not mean thatdatasets have to be a perfect match to use them.In fact, it would be difficult to find perfectlymatched data given that the monitoring systemswere usually developed for very different purposes(for example, global warming versus humanhealth). However, as long as the differences arerecognized, then appropriate comparisons can bemade. This is particularly true when the param-eter you are interested in is not being directlymeasured, and properties of related measured dataare combined to derive information; for example,gaining an understanding of the movement of

Investigate all the satellites that you thinkmay have relevant data, since their names donot always accurately describe their completeabilities (such as TOMS, the Total OzoneMapping Spectrometer, which has also success-fully been used to map aerosols). Rely on yourliterature search of similar projects and workwith your expert partners.

C.2 Select ground based monitoringmeasurement systems.

Using the data sources that you investigatedwhen finding partners and resources, select thespecific ground-based networks that will workfor your project. If collecting your ownprimary ground-based data, consider ways thatexisting networks could enhance your dataset.

C.3 Ensure compatibility of datasets.

Once you have determined the datasets youwill be using, you need to compare theircharacteristics and valid applications. Integra-tion of ground and image data requires com-parison of detailed point data with larger, moregeneral image information. Some of the moreimportant characteristics that should beconsidered are:

Spatial coverage: Make sure that data areavailable in the region or area of interestusing both datasets, i.e., ground-baseddata are collected in that region and asatellite images of a comparable area.

Scale/Resolution: Data from satelliteshave a specific resolution ranging fromseveral hundred meters to 50 or morekilometers. Ground-based point sourcedata may be impacted on a very smalllocal scale, such as an air sampler next toa highway or refinery. Consider how theresolutions of the data sources comparewhen reviewing specific data.

Atmospheric vertical layer: Whenretrieving atmospheric constituents (i.e.,pollutants) from satellite data andcomparing them to ground-based data, itis important to know where the retrieved

23

toxics by monitoring particulate movementcombined with ground-based chemical analysis.

C.4 Develop methods for quantitative data analysis.

The transformation of data into informationtypically follows several steps 1) data processing,quality assurance, data reduction and cataloging;2) data modeling and analysis; 3) feature orinformation extraction from images; and 4)visualization and decision support. One type ofdata model is data integration. In order tocompare data in a quantitative manner, it isnecessary to combine, reconcile and transformthese into a form relevant to the specific effects ofconcern. Each sensor provides a unique view intothe multidimensional features of the effect. Themultiple features can support each other andwhen fused can identify the event being studiedwith more certainty. For example, multiplesatellite images can be fused by simplygeoreferencing and superimposing them or, morecomplexly, conducting mathematical operationsto combine them. These types of integration canalso incorporate point data such as surfaceobservations from AERONET. The resultingintegrated images are analyzed for trends, such asthe spatial extent of an aerosol plume, its sizedistribution, or chemical composition. Thesefeatures from the multiple data sources are thenassembled to formulate a more complete pictureof environmental conditions. Your project shouldinvestigate existing data integration techniquesand their applicability to your data.

3.4 STEP D. APPLY TECHNIQUES FORELECTRONIC COMMUNICATION

Often when working on multi-disciplinaryprojects, and especially if the data is real-time orcollected in the field, communication and datatransfer between those involved is critical. TheInternet and affordable high memory computershave made this process much easier, and they canbe configured to obtain essentially all satellite dataand some of the ground-based data automatically.

However, as with any automated system, well-documented and especially working backupplans are imperative. Internet and e-mail can beused to coordinate a small project involving 2 or3 people or a large program of early warning andimage processing. Occasional face-to-facemeetings can be important for making rapidprogress and building teams.

D.1 Collect data and images from existing on-linedatabases.

A significant amount of data and images arealready available on-line, including manysatellite images and ground-based datasets fromfederal agencies, such as air quality data andaerosol optical thickness from nationwidenetworks. Some of these data may be obtainedthrough web-based servers, although some datamust be ordered via Internet request forms.Using on-line databases is an efficient way tocollect data, both real-time and historical. Theyare also useful for identifying and browsingavailable data and determining the datasets thatare relevant to your project.

Many researchers are beginning to put imagesand data from their models on-line in anoperational mode. Typically, these model runsare available for use for research purposes ifdevelopers are given proper credit. These modelsoften include a summary of the data and imagesused to develop them on a daily basis. Theability to access data, images, and models on-linecan greatly reduce the amount of resourcesneeded by a small qualitative project.

D.2 Use servers and websites to share large images.

Large images that have been downloaded andprocessed can be placed on a server and accessedthrough the worldwide web. This avoids someof the problems of sending large amounts of datathrough e-mail, which is especially important ifany of your team is accessing the data remotely.The web can also serve as a platform for dialogor feedback.

24

E.3 Develop project plan.

Using the basic materials developed thus far aswell as the resource information and defined rolesand responsibilities, develop a project plan. Theproject plan will be specific to your informationneed, and should guide the project throughout itslife. It should include the following elements:scope, partners and their roles and responsibilities,proposed project description, schedule, mile-stones, quality requirements, deliverables, budget,project constraints, project resources, and report-ing requirements.

E.4 Launch project.

Initiate the project, follow the project plan, andcommunicate regularly with partners.

E.5 Produce deliverables and monitor the project.

Document your successes and review the areasthat were less successful. Always keep in mind thedevelopment of future work, partners, andresources while the project is on-going, especiallyif you had to reduce scope to fit available re-sources. Place your data and reports on theInternet, present at conferences, and publish, inorder to allow others use your data.

E.6 Complete the project and celebrate withyour team.

Broadcast your successes to others and congratu-late your team.

D.3 Use e-mail to notify others of new developmentsor data availability.

E-mail can be used to communicate changes anddevelopments in the project and to notify theteam when data or images are available on thewebsite. If you have a project and many peopleinterested in its activities a listserver can enableyou to send mass mailings to interested people.

3.5 STEP E. CONDUCT PROJECT

The previous four steps pave the way to imple-mentation of the project. The following steps arefinal preparation and implementation guidelines.Many of these are basic well-known concepts ofproject management and thus are presented inoutline form only.

E.1 Secure resources.

Ensure that you have sufficient resources, bothfinancial and labor, to implement the project.This should be done using standard projectmanagement and planning concepts. The morespecific you are about your project, the moreaccurate your resource estimates. If needed,reduce scope or plan for a phased implementationin order not to overestimate the amount of workthat can be accomplished.

E.2 Finalize roles and responsibilities.

Similar to ensuring resources, finalize the role thateach partner will play. Make sure they under-stand what is expected and their level of effort.

25

We used several criteria to select an existing studyas a pilot project to refine and demonstrate ourmethodology for monitoring transboundaryaerosol transport. First, the study had to berelevant to a U.S. environmental issue, as requiredfor work funded by the U.S. EPA. Second, it hadto be relevant for other countries, as required byUNEP/GRID Sioux Falls Center. Third, it hadto use the capabilities of the partners involved,

4. PILOT PROJECT

particularly U.S. EPA, UNEP/GRID Sioux Falls,Battelle Memorial Institute, U.S. GeologicalSurveys EROS Data Center, and WashingtonUniversity in St. Louis.

The selected pilot study tested themethodology on an existing EPA-NOAA FloridaEverglades aerial and ground-based mercurymonitoring project. Table 4-1 outlines theapplication of the methodology followed by a

Table 4-1. Methodology Used for Transboundary Aerosol Transport Pilot Study

A. Define Project. The selected study is an existing EPA-NOAA ground-based and aerial mercurymonitoring project conducted in southern Florida. It is investigating the airborne sources ofmercury found in fauna living in the Everglades. This study used an airplane equipped withinnovative equipment to measure various forms of mercury, as well as particulates, NO

X, CO

2, and

other factors that help identify the airborne sources of ambient mercury concentrations. Our roleis to provide satellite imagery that will qualitatively support their analytical data.

B. Gather partners and resources. Our partners are U.S. Environmental Protection Agency(Office of International Activities and other offices), UNEP/GRID Sioux Falls related staff (includ-ing Raytheon staff from the USGS EROS Center), National Oceanic and Atmospheric Agency(NOAA), Battelle Memorial Institute, and Washington University in St. Louis. Our resourcesinclude National Aeronautic and Space Administration (NASA), NOAA Operational SignificantEvent Imagery Server (OSEI) project, Naval Research Laboratory (NRL), and information fromthe main partner organizations.

C. Select data sources. Data sources for this project include: Images from TOMS, SeaWiFS, AVHRR, Landsat 7, and Meteosat 7 NOAA daily operational significant event imagery report data Aerial and ground-based data from the Florida sampling team (not yet available) EPA AIRS data for Monroe and Dade counties, Florida NRL NAAPS model and associated images Model results from HYSPLIT

D. Apply techniques for electronic communication. The team communicates through e-mail aswell as through a website provided by Washington University (http://capita.wustl.edu/Databases/UserDomains/EDISSM/).

E. Conduct project. Selected results from the pilot project are presented in this document.

26

more detailed description of the project andthe results.

4.1 PROJECT DESCRIPTION

The selected pilot study provided assistance to anexisting EPA-NOAA Florida-based aerial andground mercury monitoring project beingconducted because of high levels of mercuryfound in fish living in the Everglades. The sourceof the mercury is unknown and may be comingfrom several sources (Guentzel 1997). The EPA-NOAA project is examining the deposition ofmercury to the Everglades and trying to furtherdetermine its source. Current theories suggest

that airborne mercury may be coming from localsources, non-local U.S. sources, the AtlanticOcean, Africa, Central America, or some combi-nation of these. This study used an airplaneequipped with innovative equipment to measurevarious forms of mercury, as well as particulates,NO

X, CO

2, and other factors that will help

identify the sources of ambient mercury concen-trations. Ground level air sampling was also usedas a baseline and included measurements similarto those made on the aircraft. Intensive monitor-ing periods occurred during January and June2000. Table 4-2 shows the airborne samplinginformation; ground-based sampling continued

Table 4-2. Airborne Sampling Dates and Locations

January Sampling Runs June Sampling Runs

500010000

15008500150085008000

115001500750015007500150075008000

750500500

AtlanticAtlanticGulf of MexicoGulf of MexicoAtlanticAtlanticAtlanticAtlanticAtlanticAtlanticAtlanticAtlanticEvergladesEvergladesAtlanticAtlanticEvergladesEverglades

01/18/0001/18/0001/20/0001/20/0001/20/0001/20/0001/23/0001/23/0001/25/0001/25/0001/26/0001/26/0001/27/0001/27/0001/31/0001/31/0002/01/0002/01/00

06/03/0006/03/0006/03/0006/04/0006/04/0006/04/0006/06/0006/06/0006/09/0006/09/0006/12/0006/12/0006/14/0006/14/0006/15/0006/15/0006/18/0006/18/0006/21/0006/21/0006/22/0006/22/0006/22/0006/25/0006/25/0006/26/0006/26/00

100001500

50010000

1500200

10005000

115001150010000

450010000

5000100010001000

20010000

500055005500

1000010000

150010000

5000

AtlanticAtlanticAtlanticAtlanticAtlanticAtlanticEvergladesEvergladesAtlanticAtlanticAtlanticAtlanticAtlanticAtlanticEvergladesEvergladesAtlanticAtlanticAtlanticAtlanticAtlanticAtlanticAtlanticAtlanticAtlanticAtlanticAtlantic

Date Altitude (feet) Water Body Date Altitude (feet) Water Body

27

4.2.1 U.S. Environmental Protection Agency

This project originated from the EPA Office ofInternational Activities (OIA) due to theirconnection with UNEP/GRID Sioux Falls, theirinterest in transboundary pollution, their partici-pation in the persistent organic pollutant treaty,and their support of the Florida mercury project.They supported the contract with BattelleMemorial Institute to coordinate this work.

The EPAs Office of Research and Develop-ment (ORD) is one of the partners in the Floridamercury study. The ORD researchers helped thepilot project by identifying the most usefulimagery for their project.

4.2.2 UNEP/GRID Sioux Falls and USGSEROS Data Center

UNEP/GRID Sioux Falls is part of UNEPsDivision of Early Warning & Assessment. Amongthe missions of this division is the evaluation ofmethodologies that contribute to UNEPs missionof support for international treaties and conven-tions. The investigation of assessment andmonitoring strategies is crucial to this mission.

The UNEP/GRID Sioux Falls partnershipwith USGS EROS Center and NASA promotedaccess to expertise and data available throughthese partners. Atmospheric and radiometricscientists at the USGS EROS Data Center haveextensive experience in the analysis of the atmo-sphere component in the image signature. Thisexpertise translates into a detailed knowledge ofthe constituents of the atmosphere. This expertiseis backed up by years of direct observations,which are correlated to the image overpasses.

4.2.3 National Oceanic and AtmosphericAgency (NOAA)

NOAA is one of the major partners in the Floridaproject, providing the airplane for sampling aswell as research services and project management.

Air Pollution

Documents

theunepgrid center sioux

pilot project

joint project

financialand project

usathis report

epas office of research

eros data center http

research use