United States                                EPA/600/R-03/018
          Environmental Protection                       December, 2005
          Agency
                         Final report on:

MAPPING THE SPATIAL EXTENT OF GROUND DUST AND DEBRIS FROM
     THE COLLAPSE OF THE WORLD TRADE CENTER BUILDINGS
         David B. Jennings, David J. Williams and Donald Garofalo
             United States Environmental Protection Agency
                 Office of Research and Development
                National Exposure Research Laboratory
                   Environmental Sciences Division
                     Landscape Ecology Branch
            Environmental Photographic Interpretation Center
                       Reston, Virginia 20192

-------
                                  NOTICE
This document is a Final report. It has been peer-reviewed and comments reconciled.

Mention of trade names and/or commercial products does not constitute endorsement
or recommendation for use.

-------
ABSTRACT



      This report presents the results of an analysis into the geographical extent of the



ground dust/debris field produced by the collapse of the buildings at the World Trade



Center (WTC) on September 11, 2001.  The study focused on an area within an



approximate eight mile radius from the World Trade Center.  The study period ranged



from September 11, 2001 through September 13, 2001. This temporal limit is due to



an approximate two inch precipitation event on the  morning of September 14, 2001.



Various remote sensing imaging sources (aerial photographic and satellite image data)



and analytical techniques (qualitative interpretive analysis and quantitative image



processing analysis) were utilized in this study.



       Results include:



            High-spatial-resolution aerial photographs from September 11, 2001 and



September 13, 2001 show distinct primary and secondary deposition along roadways,



parking lots and other ground areas in lower Manhattan that extend as far north as



Canal Street on the September 13 photographs.  In the small area of Brooklyn covered



by the September 13 photographs, possible dust is observed on pier areas adjacent the



East River and directly south-southeast of the WTC area.  An excavated area and



multiple mounds of material are also observed on the piers in the general vicinity of the



possible dust.  The area of coverage for the high-spatial-resolution aerial photographs



was generally limited to the lower Manhattan  area during the study period. This



limitation  prevented a determination of ground dust/debris boundaries over the wider

-------
study area



      •     On one-meter IKONOS-1 satellite imagery from September 12, 2001



 ground dust/debris was observed in lower Manhattan.  However, the extreme oblique



perspective provided by the image precluded a boundary determination for the lower



Manhattan area.  For those areas outside of lower Manhattan, ground dust/debris could



not be ascertained.



      •     Multi-temporal Landsat 7 ETM+ multispectral images (September 12,



2001 and August 27, 2001) and EO-1 Hyperion satellite hyperspectral data were used



to assess ground dust/debris over a wide geographic area.



      A qualitative assessment of spatial patterns of high reflectance change, derived



from Landsat 7 temporal image ratio data, provided three distinct spatial patterns.  One



generally coincided with the ground boundaries derived from September 11, 2001 aerial



photographs in lower Manhattan,  south of Chambers Street. A second was related to



the WTC plume and a third was related to the eastern edges of shorelines (land/water



interface).  Lower reflectance change, although prevalent throughout the study area,



showed no coherent spatial pattern for ground dust/debris.



      The spatial patterns derived from spectral signature mapping of EO-1 Hyperion



data also showed a spatial coincidence with the ground dust/debris boundaries derived



from September 11, 2001 aerial photographs in lower Manhattan, south of Chambers



Street.  An additional pattern was noted along the west side of Manhattan proceeding



from the WTC area in the south towards Central Park to the north.



                                      iv

-------
      The maps produced from the Landsat and Hyperion data do not assure the



presence or absence of ground dust/debris for any given pixel, but rather illustrate



spatial relationships from which ground dust/debris, per pixel,  may be qualitatively



assessed. The spatial resolution provided by these sensors (30 meters) was not



sufficient for interpretive mapping of ground dust/debris boundaries, such as those



delineated using the high resolution aerial photographs.

-------
CONTENTS                                                          Page
NOTICE	ii
Abstract	iii
Introduction	1
General	4
Analysis	 9
  September 11, 2001                                                     9
      NYPD Aerial Photographs
           Data	 9
           Methods	10
           Results	11
  September 12, 2001	13
      Landsat 7 ETM+
           Data	13
           Methods	14
           Results	16
      EO-1 Hyperion
           Data	18
           Methods	18
           Results	19
      IKONOS-1
           Data	20
           Methods	20
           Results	20
  September 13, 2001                                                     21
      KAS Vertical Aerial Photographs
           Data	21
           Methods	21
           Results	22
REFERENCES	25

TABLES
Table 1     Summary Statistics, per band,  for Image Ratio brightness change values
           derived from multi-date Landsat 7 ETM+ imagery	17
Table 2     Listing of Digital Imagery and Aerial Photographs used for Analysis 	27

APPENDICES
Appendix A, Figures. Mapping the Spatial Extent of Ground Debris/Dust from the
           Collapse of the World Trade Center Buildings      	A-1
Appendix B, Local Climatological Data, New York C. Park, NY (NYC), SEP/2001  ....B-1
Appendix C, Listing of Standard Operating Procedures (SOPs) 	C-1

                                    vi

-------
                               INTRODUCTION








      This report presents the results of an analysis into the geographical extent of the



ground dust/debris deposition produced by the collapse of the buildings at the World



Trade Center (WTC) on September 11, 2001.  Specifically, the United States



Environmental Protection Agency's (USEPA) Office of Research and Development



(ORD) was tasked by the USEPA Region 2 Response and Recovery Operations Team



to evaluate remote sensing data to determine the geographical ground distribution of



the dust/debris produced by the collapse of the buildings at the WTC in New York City,



New York. The purpose of this mapping effort was to provide geographical boundaries



in support of USEPA Region 2 response and recovery operations (Final WTC Test and



Clean Program Plan, 2005).



      The analysis encompassed an approximate 256 square-mile area centered on



the WTC site (Figure 1).  Remote sensing data collected September 11, 2001 through



September 13, 2001 were utilized to investigate the spatial extent of the ground



dust/debris field. This temporal limit is due to an approximate two inch precipitation



event  on the morning of September 14, 2001 (Appendix B; NCDC, 2001) that altered



the extent of the original ground dust/debris field.



      Remote sensing data analyzed for this study included:



      1) Scanned, natural color, oblique aerial photographs collected from hand-held



cameras on September 11, 2001 by New York Police Department (NYPD)



photographers on board NYPD helicopters (NYPD, 2002).

-------
       2) Digital imagery collected from various satellite sensor platforms on September



12, 2001 to include IKONOS-1, Landsat 7 ETM+ and EO-1  Hyperion.



       Additionally, Landsat 7 ETM+ image data collected on August 27, 2001, were



acquired as a change detection vehicle to provide a pre and post September 11, 2001



comparison of landscape condition.



      3) Analog, black and white (B+W) panchromatic, vertical aerial photographs



collected on September 13, 2001 by Keystone Aerial Survey (KAS) using a standard 9"



x 9" frame aerial mapping camera.



      The aerial photographic coverages from September 11, 2001 and September 13,



2001 provided the means for high-resolution interpretive mapping but were limited to the



general area of lower Manhattan. Satellite imagery from September 12, 2001 provided



a means of wide-area coverage of the area of interest for this report.  In particular,



Landsat 7 was included to meet the specific request of analysis over the entire two-



hundred and fifty-six (256) square miles.



      For this study, the dust/debris extent was defined by those boundaries produced



from the collapse of the WTC buildings and demarcated as light/dark boundaries on the



ground. For example, as determined from the aerial photographs, boundaries were



generally distinct along the road network due to the contrast between the dark-toned



road surfaces and the light-toned dust/debris. The dust/debris covered the road



surfaces to the point where road and cross-walk lines were obscured. On the opposite



side of the boundary some faint dust may have been present but not in a sufficient



quantity to obscure surface details such as road lines or cross-walks (Figure 2B).  In

-------
contrast, it proved difficult to distinguish dust/debris boundaries atop light-toned roofs



due to a lack of tonal difference. Given the possibility of gradations of dust/debris



deposition on the landscape, areas of light dust/debris deposition may not be



detectable from the remote sensing data included in this report and therefore it is



possible that dust/debris may extend beyond the boundaries as delineated in this report.



      In contrast, boundaries could not be interpretively mapped from medium spatial



resolution satellite data (Landsat 7, Hyperion).  Mapping results from these raster data



were derived from image processing  algorithms based on  1) pixel reflectance change



per multi-date Landsat images, and 2) mapping of Hyperion data to endmember spectra



from lower Manhattan.  While these maps  may have an association with dust/debris,



they do not assure the physical presence/absence of ground dust/debris for any  given



pixel.



      The report addresses the spatial extent of ground dust/debris deposition as an



aggregate (paper, pulverized concrete and wall board, larger building materials, etc.)



and is concerned with material that would change the tonal quality and spectral



signature of ground features  in the area-of-interest. It does not analyze the chemical or



molecular composition of the ground  dust/debris. Further, the  analysis assumes no



spatial correlation between ground dust/debris  boundaries, as  mapped in this report,



and the dispersal of dust/debris into those  areas above ground level.



      The U.S. Environmental Protection Agency (EPA), Office of Research and



Development,  Environmental Sciences Division, Landscape Ecology Branch,



Environmental Photographic Interpretation Center (EPIC) in Reston, Virginia prepared

-------
this report for the USEPA Region 2 Response and Recovery Operations Team, New



York, New York.  The Figures in this report are located in Appendix A. Sources for the



aerial photographs and digital imagery, evaluated for this report, are listed in Table 1.



SPOT-4 and NYWCAP imagery, collected on September 12, 2001, were reviewed for



this report, however, they did not provide additional detail on ground dust/debris and



therefore were not included in this analysis.



      Note: the New York State Office for Technology (NYSOFT) began  contract



 airborne overflight operations with the Keystone Aerial Survey overflight on September



13, 2001.  No known airborne overflight collection operations occurred on September



14, 2001. Airborne operations for daily collection of 1-foot digital camera  imagery as



well as Light Detection and Ranging (LIDAR) data of the WTC began on September 15,



2001.




                                  GENERAL



      This report was prepared using a standard methodology that included the



following procedures:



                  •     data identification and acquisition,



                  •     imagery analysis and mapping, and



                  •     graphics and text preparation.



      Data identification and acquisition included a search of government and



commercial sources of imagery encompassing the study area.  Publicly available



airborne and satellite imagery collected over the WTC for the time period  September 11,



2001 through September 30, 2001 were identified and data matching the  temporal

-------
period September 11, 2001 through September 13, 2001 were acquired. United States



Geological Survey (USGS) topographic maps as well as a September 15, 2001



IKONOS-1 one-meter rectified image were acquired and used as reference data to



provide geographic context for the mapping of the dust/debris boundaries.



      Analysis and mapping techniques applied to the various data were unique to



each imagery type and included, 1) monoscopic visual interpretation of digital images



viewed within a CIS environment, 2) stereoscopic visual interpretation of analog



diapositive photographs viewed on  a backlit light table, and 3) digital image processing



utilizing pixel reflectance values. The individual data, methods and results of the



mapping of the ground dust/debris are specifically detailed in the Analysis section by



date and separate data type.

-------
      The process of visual interpretive analysis involves the examination and



comparison of numerous components of the remotely sensed data.  These components



include shadow, tone, color, texture, shape, size, pattern, and context of individual



elements of a photograph.  The image interpreter identifies objects, features, and



"signatures" associated with specific environmental conditions or events.  In a visual



interpretive context, "signature" refers to a combination of components or characteristics



that indicate a specific object, condition, or pattern of environmental relevance. The



professional and academic training, image interpretation experience gained through



repetitive observations of similar features or activities, deductive logic of the analysts as



well as collateral information are important factors employed in interpretive analysis.



When utilizing analog overlapping photographs, the analyst may utilize a stereoscope to



view a three-dimensional representation of the study area, thus providing the ability to



view the three dimensional relationships of landscape features.



      Digital  image processing is the computerized application of algorithms to digital



imagery.  The algorithms utilize the reflectance value and/or the spatial context of a



pixel to automate the process of interpretation and classification. However, while the



maps produced from  these methods may have an association with dust/debris, there is



no assurance that the results are specifically due to the physical presence of ground



dust/debris. The ENVI (RSI, 2004) and ERDAS IMAGINE (Leica Geosystems, 2004)



image processing software packages as well as the ArcView CIS (ESRI,  2004)



software package were utilized for the processing and  analysis of digital imagery used



in this report.

-------
      Mapped boundaries are presented on transparent overlays, attached to the



figures, and discussed in the associated text. All mapped boundaries are approximate



and will be identified in the graphics and text according to the analyst's degree of



confidence in the evidence. A qualitative distinction is made between certain, probable,



and possible identifications. When the analyst believes the identification is



unmistakable (certain), no qualifier is used. Probable is used when a limited number of



discernible characteristics allow the analyst to be reasonably sure of a particular



identification.   Possible is used when only a few characteristics are discernible, and the



analyst can only infer an identification.



      The satellite images and the image mosaic produced from the September 13



aerial photographs were rectified, prior to acquisition for this report, to the Universal



Transverse Mercator (UTM) projection, Zone 18N, NAD 83  datum, units=meters.  When



necessary for overlay analysis and reporting, un-rectified data were rectified to the



above projection.  Rectification was completed via an image-to-image registration



technique utilizing previously rectified images for reference  control points. The rectified



images presented in this report have not been evaluated for national map accuracy



standards (NMAS).



      The image figures in this report have been reproduced from either the original



digital data (as acquired  for this report) or from digital scans of analog data produced



"in-house" on a UMAX Mirage II digital scanner at a scan resolution of approximately



800 dots per inch (dpi). All figures in the report have been printed at 720dpi. Although



the reproductions allow effective display of the imagery and interpretive annotations,

-------
they may have lower resolution than the original imagery.  Therefore, environmental



features identified from the original data and described in the text may not be as clearly



discernible on the prints in this report. The specific images included in this report were



chosen based on providing the reader with the best possible spatial context regarding



the in situ environmental conditions,  however boundaries were analyzed and mapped



from many more photographs/images than were illustrated in this report. See Table 1



for a listing of photographs and images used for this report.



      Remote sensing and geospatial processing Quality Assurance Standard



Operating Procedures (SOPs), as set forth in the Master Quality Assurance Project Plan



(MQAPP) under Remote Sensing Support Services Contract (RSSSC) 68-D-00-267,



were implemented for this report  (Appendix C). Additionally, SOPs as set forth in the



Image processing and CIS software  documentation were implemented in the



processing of the data.  Due to the lack of ground truth data for the study area, a



quantitative error assessment of the  mapping results could not be provided. However,



multiple remote sensing analysts have provided input into the analysis, results and



review of this study.



      Mention of trade names or commercial products does not constitute endorsement



or recommendation for use.
                                       8

-------
                                  Analysis



      This section is listed by date with data, methods and results reported for each



data type.



September 11, 2001



NYPD Aerial Photographs



DATA



      Data from September 11, 2001 were provided by USEPA Region 2 in two



      formats:



   1)  One-hundred and ninety-seven (197) digital aerial images in JPG compressed



      image format were provided by USEPA Region 2. These data were produced



      from hand-held photographic images collected by NYPD photographers onboard



      NYPD helicopters on September 11, 2001.  Metadata, such as camera type,



      photograph time-tags, sequence of collection, photographic scale, photographers



      name and scan resolution of the digital images were not provided, therefore the



      specific parameters of these data cannot be conclusively determined. A relative



      temporal reference frame was established based on the time of the collapse of



      WTC buildings 2, 1 and 7 (approximately 9:50 am, 10:30 a.m. and 5:30  p.m.



      respectively on September 11, 2001; FEMA, 2002).  The identification and



      condition of the WTC buildings, in the images, provided a general collection time



      sequence.  The images were collected at both low and high oblique perspectives



      with the image foreground generally focused on lower Manhattan.  This  collection



      characteristic generally limited the interpretable portion of the imagery (and the

-------
      ground dust/debris boundary determination) to the lower Manhattan area. A



      spatial index of the NYPD JPG images was not produced for this study.



2)    NYPD oblique aerial photographs reproduced in the publicly available book



      Above Hallowed Ground (NYPD, 2002).  These photographs were also collected



      by NYPD photographers onboard NYPD helicopters, however, these data were



      not available to the USEPA as individual photographs or scanned images for this



      report.  Therefore, photographs from the book that were deemed necessary for



      inclusion in this report were scanned at 800 dots per inch (dpi), converted to a



      Tagged  Image File Format (TIFF) format and utilized as figures in this report.



Despite the lack of attributable metadata, the NYPD photographs were the "best



available" aerial image data source collected on September 11, 2001. The image scale



of the data is variable due to the non-consistent flying height of the platform(s) as well



as the oblique nature of the photographs.



METHODS



      The ground dust/debris boundaries were mapped by employing the combined



means of visual interpretive mapping techniques (Avery and Berlin, 2002; ASPRS,



1997) and digital mapping methods within a CIS.  On the computer screen, a "dual-



view" mode was employed whereby the un-rectified aerial images were  visually



analyzed in one view to determine the general placement of the dust/debris boundaries



while, in a second view, "heads-up"  digital collection techniques were employed to map



the dust/debris vector boundary using the rectified mosaic in Figure 3 as a base map.



      Visible boundaries delineated from September 11, 2001 data were based
                                      10

-------
primarily on tonal variation (light/dark feature demarcation) along paved roadways and



parking lots as well as change analysis of pre and post collapse images of WTC



buildings 1 and 2 respectively.



RESULTS



       Figures 2A through 2E are centered on the intersection of West street and



Warren street in lower Manhattan and present examples of clearly visible ground



boundaries produced by the WTC building collapse clouds.  The figures are presented



here in order to clarify the definition of a dust/debris boundary and to illustrate how the



analysis and mapping proceeded in this report.  Boundaries were generally distinct



along the road network due to the contrast between the dark-toned paved road surfaces



and the light-toned dust/debris. The dust/debris covered the road surfaces to the point



where road and cross-walk lines were obscured. On the opposite side  of the boundary



some faint dust may have been present but not in a sufficient quantity to obscure



surface details such as road lines or cross-walks (Figure 2B).  It proved difficult to



distinguish the dust/debris boundaries from light-toned rooftops due to a lack of tonal



difference.  As stated in the introduction, boundaries derived from visual interpretive



analysis were based on clearly demarcated areas of contrast. Further,  a distinction was



observed with respect to the ground dust/debris directly deposited by the collapse



clouds (WTC buildings 1 and 2 as shown on Figure 2A-2E) and secondary ground



dust/debris deposition that may have been due to post-collapse dispersal vectors such



as transportation and wind. This secondary condition is evident in Figure 3, along West



Street, where vehicular traffic has likely  redistributed dust north of the original boundary
                                       11

-------
(Stuyvesant  School Bridge, Figure 2E).



      Figure 3 is a rectified mosaic of the photograph displayed on pages 48 and 49 in



Above Hallowed Ground (NYPD, 2002). The photograph was collected at a low-oblique



perspective and provided a base map from which to present the boundary in lower



Manhattan on a single image. The ground dust/debris boundary was derived from the



analysis of multiple images, most of which were not presented in this report, and



reflected ground dust/debris boundary conditions between 10:30 am (post-collapse,



WTC buildings 1 and 2) and 5:30 pm (pre-collapse, WTC building 7) on September 11,



2001.  The  mosaic seam, observed in the center of Figure 3,  is due to a lack of overlap



between the scanned images (pp.  48-49 in Above Hallowed Ground) and the oblique



characteristic of the aerial photograph.



      No visible ground dust/debris boundaries could be determined for the  limited



areas of Brooklyn, adjacent the East River, where coverage was available.



      Results (Figure 3)  show the northern limit (between North End Avenue and



Centre Street) of the ground dust/debris boundary at approximately Chambers Street.



However, variations in the northern boundary are observed 1) between Reade and



Duane streets along Greenwich  Street, Hudson Street and West Broadway,  and 2)



between Duane and Thomas Streets along Church Street. To the northeast, the



boundary appears to extend to the east from the vicinity of City Hall along a line



approximately parallel and south of the Brooklyn Bridge  (Frankfort Street) as it meets



the Franklin Delano Roosevelt expressway (FDR). The  southern boundary of the



ground debris field appears to extend to Battery Park and to the surrounding southern
                                      12

-------
limits of Manhattan. The western boundary generally extends to the Hudson River.





      The mapping of the eastern and southeastern boundaries (south of the



Brooklyn Bridge and east of Pearl and Water Streets) was problematic due to the limited



coverage of the area provided by the September 11, 2001 aerial photographs and the



plume emanating from "Ground Zero" which obscured the area.  However,  a dust/debris



boundary was delineated, for this day, based on the analysis of September 13, 2001



aerial photographs, where a "blanket" of ground dust/debris deposition - characterized



by light toned dust that completely covered the roadways and obscured white road and



cross-walk lines - was observed in the eastern and southeastern portions of lower



Manhattan.  The deposition characteristics in this area were consistent with areas of



lower Manhattan that were impacted by primary deposition from the WTC building



collapses on September 11, 2001. Therefore, it is likely that the boundary, south of the



Brooklyn Bridge did extend up to the FDR and possibly to the raised FDR roadway as



well as the adjacent piers along the East River on September 11, 2001. However,



because of the limited coverage, no qualitative distinction of the ground dust/debris in



this area was attempted on this date. See the results for September 13, 2001 for a



more comprehensive assessment of ground dust/debris in this area.








September 12, 2001



Landsat 7 ETM+



DATA
                                      13

-------
Landsat 7 ETM+ satellite data (path14/row 32) of September 12, 2001 included six



bands of multi-spectral data (bands 1-5, 7) ranging from the visible to mid-infrared at



28.5-meter spatial resolution and one band of panchromatic B+Wdata (band 8) at 14.5-



meter spatial resolution (USGS, 2004a).  Additionally a spatially coincident Landsat  7



ETM+ image from August, 27, 2001 was acquired and the identical bands were



processed for analysis. These pre and post WTC building collapse datasets were



processed in order to determine patterns of reflectance change that would be consistent



with the deposition of ground  dust/debris. These data also provided the best means for



analyzing ground dust/debris  across the entire 256 mile2 study area (Figure 4, Figure 5).



METHODS



      ENVI software was used to process the Landsat 7 ETM+ data.  Steps involved in



the data processing chain for each date of data consisted of:



      1) A geometric and radiometric correction to Level 1G was applied, prior to



acquisition by the USEPA, by the source vendor EROS Data Center (USGS, 2004b).



      2) A top-of-atmosphere correction (TOA) was applied to transform the pixel



digital numbers (DN) to exoatmospheric reflectance values (Lillesand et al., 2004) and



to provide a scene-to-scene reflectance calibration.



      3) A 256 mile2  subset centered on the World Trade Center site was



clipped from the original scenes.



      4) A Neighborhood Analysis consisting of a 3x 3 pixel, mean spatial filter



(Lillesand et al., 2004) was applied to each pixel to compensate for an approximate  one



pixel registration offset between the two dates of imagery. A low-pass, 3x3 mean filter
                                      14

-------
was used in order to maximize large-area patterns in brightness and to aid in the



mapping of large area contiguous boundaries associated with the dispersal of the



ground dust/debris. All subsequent analysis was performed using the spatially filtered



pixel reflectance values.



      5) Temporal image ratio change detection methods (Lillesand et al., 2004) were



implemented as a means for identifying those pixels that increased in reflectance



between the two dates of imagery. Temporal image ratioing is a digital pre-



classification change detection technique that compares coincident, per band pixel



reflectance values of the two rectified images (Jensen, 1996). In this study, the



temporal image ratio approach compared the after image (September 12, T2) with the



before image (August 27, T^ to compute a per-pixel reflectance change ratio (T2/Ti).



Ratio values >  1 represented positive reflectance change over time.



      The temporal image ratio algorithm was applied to each individual band to create



seven separate change images.  Although image ratio data for all bands were produced



and analyzed, for purposes of graphic reporting, only image ratio data from band 7 and



band 8 were illustrated for this report.  Band 7 was the only band to approach an



increase in mean reflectance change between the two dates (0.98, Table 1), and its



spectral wavelength range (2.09 - 2.35 microns) was shown to have a strong



relationship with the WTC dust material (Clarke et al., 2001). Band 8 was included due



to the increase in spatial resolution (14.5 meters) which provided additional spatial



detail.  Pixel values corresponding to no-change and decreases in reflectance were not



reported. While image ratio figures for bands 1-5 were not included here, their spatial
                                       15

-------
results were similar to the patterns observed in bands 7-8 and discussed below.



RESULTS



      The temporal image ratio results provided insight into the spatial pattern of



change over the whole study area and allowed general conclusions based on those



patterns.  The ratio data do not assure the presence or absence of ground dust/debris,




for any given pixel, but rather denote those pixels that have greatly increased in



reflectance and thus may have an association with dust/debris deposition.



      Figure 5 and Figure 6, based on results from band 8 and band 7, respectively,



utilized pixel values corresponding to a positive increase in image ratio data of > 2



standard deviations (SD). Data resident within this threshold (> 2  SD) represented



approximately 3% and 2% of the  image ratio data in band 7 and 8 respectively and were



utilized to detect those areas most related to increases in reflectance.  From Figures 5



and 6 the following spatial patterns were observed:



      1)  The area of lower Manhattan, bounded approximately by Chambers street to



the north, has substantially increased in reflectance between the two dates (Table 1).



The boundary is in general spatial agreement with the ground dust/debris boundary



mapped from the September 11,  2001  photographs (in orange on  overlays).  Table 1  is



a summary of image ratio statistics derived for bands 1-5, 7 and 8 and provides the



mean and SD for each band over the entire study area as well as a mean value, per



band, for the area of lower Manhattan encompassed by the September 11 dust/debris



boundary as determined from the NYPD photographs.  Noted are the approximate  50%



increases in the mean  ratio values in lower Manhattan as compared to the mean values
                                      16

-------
for the entire study area.

Table 1.  Summary statistics, per band, for Image Ratio brightness change values
derived from multi-date Landsat 7 ETM+ imagery.

* total area-of-interest = 256 miles2 centered on the WTC.
** 9/11 boundary as mapped from NYPD photos.
Landsat 7 ETM+:
Band Number
Band 1
Band 2
Band 3
Band 4
Band 5
Band 7
Band 8
Mean image ratio
value for total
area-of-interest *
0.73
0.75
0.76
0.87
0.9
0.98
0.81
St. Dev.of image
ratio value for
total area-of-
interest
0.13
0.17
0.21
0.26
0.32
0.33
0.23
Mean image ratio value
for area encompassed
within the 9/11
boundary (Lower
Manhattan) **
1.01
1.14
1.27
1.33
1.39
1.54
1.28
      2) high reflectance change is noted along the eastern shorelines of land/water

interfaces throughout the study area.  This pattern is possibly related to an east/west

directional mis-registration of the two images which would have a particular effect of

distorting reflectance ratios related to north/south trending linear features.

      3) a large area of increased reflectance is associated with the plume extending

from the WTC area to the southwest.  The increase is probably due to reflectance from

the plume and not necessarily associated with a change in ground reflectance due to

deposition of dust/debris.

      4) clouds and shadow, present on August 27 (Figure 4), dominate the northern
                                       17

-------
portions of the study area as well as portions of Queens, Brooklyn and New Jersey and



were a source of large reflectance change.  The reflectance areas related to the clouds



and shadows, in proximity to lower Manhattan and the WTC plume, are mapped on



Figure 5 and Figure 6 overlays.  Due to the presence of clouds/shadows on August 27,



no determination of ground reflectance change could be ascertained in these areas on



September 12.



EO-1  Hyperion




DATA



      The hyperspectral dataset from the EO-1 Hyperion sensor (USGS, 2004c)



consists of 220 separate spectral bands covering the wavelengths from 0.4 -2.5 microns



at a spatial-resolution of 30 meters. The Hyperion data covered an approximate 4.7



mile swath across the center of the study area (Figure 7).



METHODS



       ENVI software was used to process the Hyperion data. Steps involved in the



data processing chain consisted of:



      1) removing noisy and defective spectral bands, by inspection, which reduced the



dataset to 154 bands.



      2) correcting for atmospheric effects by radiative transfer modeling using the



software ACORN (Atmospheric CORrection Now; AIG, 2004).



       3) noise reduction using principle component analysis (PCA).



       4) identifying candidate regions in the data that contained materials of interest



(in this case, ground dust/debris from the WTC collapse). Ground based spectral
                                      18

-------
signatures obtained by the USGS (Clark et al., 2001) were used to create a library of



signatures (endmember spectra) for material identification and mapping purposes. This



library was compared to spectral signatures in the Hyperion dataset and areas (pixels)



matching entries in the library were used in the mapping algorithms.  Spectral



signatures were obtained just north of the WTC site and were used in the material



mapping algorithms.



      5) Areas matching these endmember spectra were mapped throughout the entire



image. The Mixture Tuned Matched Filter (MTMF; Boardman, 1998) was used for this



analysis due to the fact that the spectra of materials  in cluttered urban areas are a



mixture of many other materials at the pixel resolution of the Hyperion sensor (30



meters). The spectral features of the WTC dust were similar enough to common



building material found in urban areas to defeat the spectral feature algorithms. This



problem, combined with the noise inherent in data obtained from space, made it difficult



to spectrally differentiate ground dust/debris from the surrounding urban background



material.



RESULTS



      As with the Landsat 7 reflectance change results, we found a spatial coincidence



between the Hyperion results and the results derived from the September 11, 2001



aerial photographs in lower Manhattan south of Chambers Street (Figure 7, Figure 8).



An additional pattern was noted along the west side  of Manhattan proceeding from the



WTC area in the south towards Central  Park to the north.



      The mapped areas do not assure the presence or absence of ground dust/debris
                                      19

-------
for any given pixel, but rather indicate the likelihood that a Hyperion pixel is spectrally



similar in the signature library - here, ground dust/debris from the WTC building



collapses in lower Manhattan.



IKONOS-1



DATA



      Rectified digital imagery data were collected by Space Imaging's IKONOS-1



satellite (Space Imaging Inc., 2004) at nominal four-meter multispectral spatial



resolution and nominal one-meter panchromatic spatial resolution. These data were



subsequently merged to produce a multi-spectral dataset at nominal one-meter



resolution. The imagery covered approximately 50% of the study area (Figure 9 ) and



included portions of Manhattan, Brooklyn, Staten Island and New Jersey.  However, the



low collection angle of acquisition of the sensor (approximately 32 degrees) produced a



low-oblique image, rather than  the standard near-vertical image, and served to degrade



the spatial resolution and spectral integrity of the data. In the lower Manhattan area this



caused extreme building layover/shadow on the imagery and obstructed



ground/roadway visibility.



METHODS



      The imagery was analyzed, using visual interpretive methods, from within the



ArcView CIS environment.



RESULTS



       Although a boundary map was not produced from the IKONOS-1 data, some



areas of dust/debris were visible in lower Manhattan.  Probable light-toned dust/debris
                                      20

-------
was visible along West Street extending north from the WTC area to approximately



Canal street.  Additionally, ground dust/debris was visible at discrete points (Figure 10)



south of Canal Street along Church Street, along West Broadway, east of Centre Street



in the vicinity of the approaches to the Brooklyn Bridge and in a parking lot located



between Pearl and Water Streets to the south of the Brooklyn Bridge.



      The distinctive light-toned demarcation of dust/debris, evident in lower



Manhattan, was not observed in  areas outside of lower Manhattan.








September 13, 2001



KAS Vertical Aerial Photographs



DATA



      Data from September 13,  2001 consisted of 78 B+Wfilm diapositives



(transparencies) of 9" x 9" vertical and oblique aerial photographs of the lower



Manhattan collected at a scale of 1:6,000 (KAS, 2004). Time of collection was



approximately 9:00a.m. eastern daylight time (edt). The areal coverage of the



photographs (Figure 11) was limited to lower Manhattan and a small portion of Brooklyn



west of Hicks Street and adjacent the East River.



METHODS



      Visual stereoscopic interpretive techniques were employed using a Richards 500



zoom stereoscope backlit light table to determine the dust/debris  boundaries.



Stereoscopic viewing provided the ability to visualize vertical as well as horizontal



spatial relationships of landscape features. As with the September  11, 2001 data,
                                       21

-------
"heads-up" digital collection methods were implemented to map the boundaries.



Determination of boundaries generally relied upon the analysis of tonal and texture



variation, along with stereoscopic visualization, to delineate clearly defined light/dark



edges associated with light toned dust/debris on dark toned road pavements.








RESULTS



      A probable change in the northern ground dust/debris  boundary, in lower



Manhattan, occurred between September 11, 2001 and September 13, 2001.  The



probable northern boundary on September 13 extended approximately to Canal Street



with other possible areas of dust extending beyond Canal Street (Figure 12).  Multiple



possibilities exist for the apparent change in the northern dust/debris boundary.  First,



WTC building 7 collapsed in the intervening time period which may have caused a wider



distribution of dust than that delineated for the September 11, 2001  dust/debris map.



Second, a transportation vector may have brought about a redistribution of dust.  This



scenario  is most evident along West Street which was apparently an ingress/egress



route for WTC operations. And third, an atmospheric vector due to a shift in wind



direction to the north-northwest on September 13, 2001 (NCDC, 2003).  The new lower



Manhattan boundary area, existing between Chambers and Canal Streets, is the outer



limit of an irregular and patchy distribution of dust/debris that appears to have



accumulated along roadways and curbs in high traffic zones.   Throughout this area,



vehicle tracks are visible in the dust on the roads. This is particularly evident along



Canal Street in proximity to a probable heavy equipment staging area (Figure  13).
                                      22

-------
      A possible area of ground dust/debris is observed in the Brooklyn pier areas,



adjacent the East River, directly east of Governors Island and south-southeast of the



WTC (Figures 12,  14, 15).  The dust appears to be limited to pier areas along the East



River, west of Columbia Street, and south of Congress Street.  The southeastern limits



could not be determined due to a lack of photographic coverage south of Sedgewick



Street. The dust appears to lightly coat the paved areas of the wharfs (light-toned lines



are visible) and distinct vehicle track patterns are observed in the dust (Figures 14, 15).



 The vehicle track patterns, however, are not observed on the roadways in the



surrounding area, such as Columbia Street or the Brooklyn-Queens Expressway.



Additionally, an excavation area with adjacent crane and multiple mounds of material



are observed in the general vicinity of the possible dust (Figure 15). A specific



association between the observed dust, the excavation activity and the mounded



material could not  be ascertained.





      In addition,  due to improved photographic quality and coverage of the September



13 photographs in comparison the September 11, 2001 photographs, a qualitative



distinction was made regarding the ground dust/debris in the southeastern portions of



lower Manhattan along the FDR and areas east along the river.  Possible and probable



ground dust/debris was now noted along the raised portions of the FDR and on the



piers extending into the East River (Figure 12).  These areas were obscured on the



September 11, 2001 NYPD photographs and no qualitative distinction of the ground



dust/debris was attempted form the September 11, 2001 photographs.
                                      23

-------
      A more comprehensive ground dust/debris mapping effort in Brooklyn could not



be accomplished due to a lack of photographic coverage east of Hicks Street.
                                      24

-------
REFERENCES

AIG (Analytical Imaging and Geophysics LLC). 2004. ACORN, version 4.  44509
      Arapahoe Ave., suite 100, Boulder CO.  URL: http://www.imspec.com/ (last
      accessed 29 November, 2005).

ASPRS (American Society of Photogrammetry and Remote Sensing). 1997. Manual of
      Photographic Interpretation, 2nd edition.  Ed., Warren R. Philipson. ASPRS,
      Bethesda, MD., pp.689.

Avery I.E. and G. L. Berlin. 2002. Fundamentals of Remote Sensing and Airphoto
      Interpretation.  Prentice Hall, Upper Saddle River, NJ.,  pp. 472.

Boardman, J.W. 1998. Leveraging the High Dimensionality of AVIRIS Data for
      Improved Sub-Pixel Target Unmixing and Rejection of False Positives: Mixture
      Tuned Matched Filtering. AVIRIS Airborne Geosciences Workshop Proceedings
      (JPL Publication 97-12).

Clark, R., R. Green, G. Swayze, G. Meeker, S. Sutley, T. Hoefen, K.  Livo, G. Plumlee,
      B. Pavri, C. Sarture, S. Wilson, P. Hageman, P. Lamothe, J. Vance, J.
      Boardman I. Brownfield, C. Gent, L. Morath, J. Taggart, P. Theodorakos, and M.
      Adams.  2001.  Environmental Studies of the World Trade Center area after the
      September 11, 2001 attack. U. S. Geological Survey, Open File Report OFR-01-
      0429. URL: http://pubs.usgs.gov/of/2001/ofr-01-0429/ (last accessed 29
      November, 2005).

ESRI (Environmental Systems Research Institute). 2004. ArcView CIS, version 3.3.
      380 New York Street, Redlands, CA. URL: http://www.esri.com/ (last accessed
      29 November,  2005).

FEMA (Federal Emergency Management Agency). 2002. World Trade Center Building
      Performance Study, FEMA 403, URL:
      http://www.fema.gov/librarv/wtcstudv.shtm (last accessed 29 November, 2005).

Jensen, J. 1996. Introductory Digital Image Processing. Prentice Hall, Upper Saddle
      River, NJ.,  pp. 318.

KAS (Keystone Aerial Survey). 2004.  URL:  http://www.kevstoneaerialsurveys.com/
      (last accessed 29 November, 2005).
                                     25

-------
Leica Geosystems. 2004.  ERDAS IMAGINE, version 8.6. 2801 Buford Highway, Suite
      400, Atlanta, GA. URL: http://gis.leica-geosystems.com/ (last accessed 29
      November, 2005).

Lillesand, T., R. Keifer and J.  Chipman.  2004.  Remote Sensing and Image
      Interpretation.  John Wiley and Sons, Hoboken, N.J., pp. 763.

NCDC (National Climate Data Center). 2001. Daily Climate summary, Central Park,
      New York, N.Y.

NYPD (New York Police Department). 2002. Above Hallowed Ground. Viking press,
      New York, N.Y., pp.192.

RSI (Research Systems Institute). 2004. ENVI (The Environment for Visualizing
      Images), version 3.6.   4990 Pearl East Circle, Boulder, CO.  URL:
      http://www.rsinc.com/ (last accessed 29 November, 2005).

Space Imaging  Inc. 2004. 12076 Grant Street, Thornton, CO 80241. URL:
      http://www.spaceimaging.com (last accessed 29 November,  2005).

USGS (United States Geological Survey). 2004a. URL:
      http://landsat7.usgs.gov/index.php(last accessed 29 November, 2005).

USGS (United States Geological Survey). 2004b. URL:
      http://eosims.cr.usgs.gov:5725/DATASET  DOCS/landsat7 dataset.html#1 (last
      accessed 29 November, 2005).

USGS (United States Geological Survey). 2004c. URL:
      http://eo1.usgs.gov/hyperion.php  (last accessed 29 November, 2005).

Final WTC Test and Clean Program Plan. 2005. URL: http://www.epa.gov/wtc/panel
      (last accessed 29 November, 2005).
                                     26

-------
Table 2. Listing of Digital Imagery and Aerial Photographs used for Analysis
Image Associated
Source Fiaures
Landsat 7 ETM+
NYPD
Landsat 7 ETM+
EO-1 Hyperion
IKONOS-1
SPOT-4
NYWCAP
KAS
IKONOS-1
4
2-3
5-6
7-8
9-10
*
*
11-15
**
Date of
Acquisition
08-27-01
09-11-01
09-12-01
09-12-01
09-12-01
09-12-01
09-12-01
09-13-01
09-15-01
Original
Scale/GSD
28.5m
Variable
28.5m
30m
1m/4m
20m
Variable
1:6,000
1m/4m
Data type
as supplied to EPA
MS Imagery
Color Photographs
MS Imagery
HS Imagery
MS Imagery
MS Imagery
Color Imagery
B&W Photographs
MS Imagery
NYPD: New York Police Department
NYWCAP: New York Wing Civil Air Patrol
KAS: Keystone Aerial Survey
MS: Multi-spectral
HS: Hyper-spectral
GSD: Ground Sample Distance

*The SPOT-4 and NYWCAP imagery, collected on September 12, 2001, were reviewed
for this report, however, they did not provide additional detail on ground dust/debris and
therefore were not included in this analysis.

** The IKONOS-1 imagery, collected on September 15, 2001, was used only as base
data for "heads-up" mapping of the September 11, 2001 and September 13, 2001
photographs.
                                     27

-------
APPENDIX B: Local Climatological Data, New York C. Park, NY (NYC), SEP/2001
           §  I
           1  2
                  SEPTEMBER 2001
                  LOCAL CLIMATOLOGICAL DATA
                  NOAA, National Climatic Data Center
                   "•EMPERA'URE f
< o

 6
       I DEG DAYS
        BASE 65
    23  F   =
    gm  5   o
    LU h ' UJ   O
    3>  I   O
                                                WEATHER
                                 NEW YORK C.PARK, NY

                                 CENTRAL PARK OBSERVATORY   (NYC)
                                 Lat: 40°47'N  Long: 73°58'W  Elev (Ground): 158 Feet
                                 Time Zone:  EASTERN   WBAN: 94728 ISSN #:0198  3601
   PREC P"~K" ON PRESSURE
    IINCHPo.   IINUtri.UFHt,.
                                                             0700 1300  2400 2400
                                                             LSI  LSI  LSI LSI
5 S
& Uj
12
                                                                                   WIND

    SPEED - MPI!
    DP - ~6No OF DE9REE3
          MAXIMUM
        5 SEC 2  MIN
         Q   I Q
I£V, - -<_•»   MAXIMUM TEMP _ 11
| THUtdERS'ORMsj
ooo
'll*!
rnrnfrt
SUNSHINE, CLOUD. &
i t-EA LEVEL
MAX MUM
M N MUM
M N MUM -EMP ^ JJ
M'NMUM~EMP i 0 : •
I IEAVY FOG . 1
PREC
PREC
3NOV.
                                                            ON _ 0 01 NCI I
                                                       PREC'P'"ATGN £ 0 10 NCH .
                                                       SNOWFALL *: 1.0 NCH
                                                        B-1

-------
  HOURLY PRECIPITATION
    (WATER EQUIVALENT IN INCHES)
NEW YORK C.PARK, NY
SEPTEMBER 2001   NYC  WBAN # 94728

LJU
i

01
02
03
04
05
Qfi
07
08
09
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
! ' 2400
FOR HOUR (LST) ENDING AT j^ FOR HOUR (LST) ENDING AT uJ ^ g i? LST

1
0 10












0.38






0.03



0.11






2













0.05










0.12






3













0.30










0.07






4













0.13






0.16



0.03






5




















0.17



0.04





I
6 7













0.16 0,19






T |



0.01 j






8













0.35
















i ii- > - : i - i 	 < c i z ~ *
•< I I Q 5 £ t » -i?
9 10| 11 12 p 13 14 15| 16 17 18 1 19 20 21 22 23 24 » 5 1 | $
i ior II1 01 0.10
!02 I °2 °-°°
:03 ' °3 OOD
O.OSi 04 0.08 0.01 1 T 1 04 0 14
JQ5 I I °5 °'OQ
J06 06 0.00
I 1 07 I 07 0.00
:08 II 08 0.00
109 I i 09 0.00
J10 0.41] 0,17^ 0.01 1 0.57' 10 1.16
1 11 II 11 0.00
!12 ' 12 ooo
I 113 . I 13 0.00
0.20' 0.06! 0,02 0,03 i 14 0,02' 0,01 T I 14 1,90
I 115 ' 15 000
i
1 16 I 1S °'°°
117 ! 17 0.00
118 I 18 0-0°
I !19 ' 19 O'^O
0.04 0.05 i 20 016 003; 0.07J 0.12; T T 0.24; 0.02 0.01 008 001 20 0.63
1 21 II 21 0.36
!22 II 22 0.00
!23 II 23 0.00
i24 ! I 001 003 24 0.04
0.011 001 ! 0.01 J25 II 25 °-41
326 II 26 0.00
i27 ! 27 0.00
I '28 I T I 28 T
!29 I.I . 29 0.00
i30 T 0.03! 0,011 003 0.111 0,04, 010: 0.03' 001 T 30 0.36
MAXIMUM SHORT DURATION PRECIPITATION (See Note)
_. _ ' Note ; The sum of the hourly totals is qivsn when
, Time Period (Minutes) 5 10 15 20 30 45 60 80 100, 120 150 180 it differs from the daily tote' NWS does not edit
: Preaoitatar, (Inches) 26 37 .41 .44 57 .57 .57 .58 58 ' 60 .73 '' .84 ' ASOShouriy vaiuesoutmayeditdaiiyandmonthiy
1 ' .... .... i ; tola's. Hourly. dai:y, and monthly totais are printed
EndmgDate 14 10 10 10 10 10 10 10 10 14 14 14 as reported by the ASOS site

Ending Time (Hour/Win) LjOp^9_JI£[8_ 1421 i 1836 1843 1843! 1843 1 843 i Jt843J_j:)84|J 0252 |_0322J
   Date and time are not entered for TRACE amounts
NOAA, National Climatic Data Center
                                           PAGE 2
                                           B-2

-------

-------
APPENDIX C: Listing of Standard Operating Procedures.

      The following remote sensing and geospatial Quality Assurance Standard
Operating Procedures (SOPs), as set forth in the Master Quality Assurance Project Plan
(MQAPP) under Remote Sensing Support Services Contract (RSSSC) 68-D-00-267,
were implemented for this report.

      SOP Number             SOP Title

1.     RS-04-97-01-R11         Data Searches
2.     RS-04-98-08-R4          Digital Imagery Acquisition
3.     RS-05-98-08-R5          Maintaining Project Files
4.     RS-06-97-07-R7          Inventory Control
5.     RS-05-97-06-R9          Standard Category 3 Report Writing and Basic
                              Measurements
6.     RS-05-97-07-R9          Standard Category 3 Site Analysis Quality Control
7.     RS-06-97-03-R7          Producing Category 3 Reports
8.     RS-05-97-02-R9          Photointerpretation: Hazardous Waste Sites
9.     RS-05-97-04-R6          Photointerpretation: Land Use/Land Cover
10.   RS-05-98-11-R7          Production of Digitally Scanned Images
11.   RS-08-98-09-R4          Geometric Rectification of Scanned Aerial  Photos
12.   RS-09-99-01-R2          Mosaicking of Digital Imagery
13.   RS-08-98-08-R3          CIS Coverage Quality Control
14.   RS-08-98-05-R5          CIS Metadata Documentation
                                     C-1

-------