Haze and Visibility Research In the
Paso del Norte and West Texas
December 2002 – August 2003 FINAL REPORT FOR PROJECT 582-3-60797
FY 2003
To the Texas Commission on Environmental Quality
(Formerly Texas Natural Resource Conservation Commission) Technical Analysis Division
Submitted by N. J. Parks, R.W. Gray, W.W. Li, D. Raina,
Center for Environmental Resource Management And
Department of Civil Engineering University of Texas at El Paso
El Paso, TX 79968 August 15, 2003
Executive Summary
This is a report of visibility research activities undertaken by the Center for
Environmental Resource Management, University of Texas at El Paso on behalf of the
Texas Commission on Environmental Quality (formerly, Texas Natural Resource
Conservation Commission). In this report, the performance data for digital image
acquisition and archiving at each of several urban El Paso sites and West Texas National
Park or rural sites, collectively denoted as “wilderness.” is presented. In general, the
urban El Paso sites operated 80% or more of the time, as they are available for
convenient maintenance of hardware and re-establishment of the dial-up connection to
the UTEP server. Wilderness site up time varied from 70% up-time for The McDonald
Observatory (University of Texas, Austin) site which was enabled to send large (400kB)
files quickly over the U.T. system internet to a nominal 50% time for the National Park
Service images archived also by this project, and thence, to 20% time for the new Big
Bend installation (stopped by lightning). TCEQ Region VI coordinated a new
deployment of a Guadalupe Mountain National Park camera; that site has operated
continuously since installation in July 2003.
In the image parameters section, a site image is presented along with the “regions of
interest” (ROI) for quality assurance and visibility indexing of the systems. The ROI for
each image is integrated and average pixel brightness is determined. The values obtained
for the period of this study (December 2002 to July 2003) are interpreted in terms of a
contrast ratio (CR) where adequate contrast of juxtaposed image sections can be
identified or in terms of brightness variation, which is indexed as the coefficient of
variation (CV) for chosen ROI.
An initial examination of the relationship of image visibility parameters and reported air
quality and meteorological data by TCEQ was performed and will be continued as part of
an ongoing graduate research project. The results of multiple regression analysis with an
image contrast value as the dependent variable and various of the air quality and
2
meteorological values as independent variables indicated that about 70% of the variation
in contrast (R2 ≈ 0.7) could be explained on the basis of gaseous urban aerosol
component concentrations, various meteorological values, and the solar radiation flux for
the preceding hour. The inclusion of PM 2.5 or PM10 from geographically peripheral
CAMS sites did not markedly improve the statistical associations. The presumption at
this time is that the gaseous components are more ubiquitously distributed and measured,
and therefore, represent a better characterization of the urban aerosol in the camera sight
path (for the long term El Paso downtown view from Chelsea). This latter observation is
salient because of the limited PM 2.5 measurement locations. A comparison to the one
short range (meters) visibility monitor data at Ascarate Park (CAMS 37) showed
remarkable agreement with the 6000-meter path from Chelsea 1 to downtown even
though the park was behind the Chelsea 1 site by several kilometers in the opposite
direction. This separation did lead to dramatic differences in visibility on occasion but
agreement was more common than not. This work also is being continued as part of a
graduate research project that is ongoing under separate auspices.
3
Table of Contents
Executive Summary 02
I. Introduction 06
II. Operation Summary for the Digital Cameras 10
III. Regions of interest and Results 17
IV. Visibility Degradation Parameters and Preliminary Data Analysis 46
V. References 54
4
Acknowledgements
We are grateful to Stuart Dattner, who originally conceived and supported the project,
Fernando Mercado who is the current manager for the project at TCEQ, Bethany
Georgeolis, and Erik Gribben of TCEQ (Technical Assessment) for all their efforts.
We would like to thank Archie Clouse, Victor Valenzuela and Joe Saenz and local
program staff of TCEQ, Region 6.
We would also like to thank Henry Del Rio and Jesus – Chuy Reynoso of the El Paso
City for all the help extended from time to time during the duration of the project.
Additionally we thank Gautam Kumar Agrawala (PhD Candidate ESE Program-UTEP)
for all the help extended to us with data analysis; Ritesh Mariadas and Aqiles Ramos for
data acquisition, files maintenance, and other critical information technology related
activities.
5
I. Introduction
This is the final report for a series of visibility activities undertaken by the Center for
Environmental Resource Management of the University of Texas at El Paso on behalf of
the Texas Commission on Environmental Quality (formerly, Texas Natural Resource
Conservation Commission). In this report, the performance data for digital image
acquisition and archiving at each of several urban El Paso sites and West Texas National
Park or rural sites collectively denoted as “wilderness.” is presented in Section II. In
Section III, site images are presented along with “regions of interest” (ROI) for quality
assurance and visibility indexing of the scenes. Where adequate contrast of juxtaposed
image sections can be identified, the values obtained for the period of this study
(December 2002 to July 2003) are interpreted in terms of a contrast ratio (CR).
Otherwise, brightness variation is indexed as the coefficient of variation (CV) existing for
a chosen ROI.
A long-standing interest has been to get to the point where some continuous PM 2.5 data
was available for image sections chosen for either CR or CV. The interest in PM 2.5
derives for the fact that the visible part of the particulate matter size fraction is 0.4 to 0.8
µm approximate diameter and is contained in the PM 2.5 fraction obtained by current
sampling systems based on effective aerodynamic diameter separations. Section IV
shows some early results of an ongoing graduate research project of one of us (D. Raina)
where the correlation of various air quality parameters reported by TCEQ and the image
parameters computed from pixel brightness levels in various ROI are presented.
In this report, we also have examined the correlation image parameters with PM10 values
available from two CAMS sites that geographically bracket the main view path of the
longest installed visibility site (Chelsea 1) in El Paso (Orquiz, Li, et. al. 2001). The mean
value of the PM 2.5 fraction is about 25% by mass of the PM10 in El Paso. Additionally,
we address in Section IV the influence of solar radiation (various times of the year at the
same MST or GMT) on the intrinsic contrast of chosen targets. Finally, the Visibility
6
values obtained from CAMS 37 have been compared to that of the Contrast Ratio values
calculated for the different sites as part of the contract. The CAMS 37 site uses the
Visibility Sensor – Model 6230 A, manufactured by the Belford Co.
Nine digital image systems operating under the auspices of the Texas Commission on
Environmental Quality (TCEQ) were originally deployed at some time during the period
of this interim report and were continuously acquiring and transmitting images to the
University of Texas Regional Haze archive server (See list in Table I). Since the
original deployment of some of these cameras, two were uninstalled during the fall of
2003 in the course of activities carried out under the auspices of TCEQ Region VI. Of
these two, one was redeployed in July 2003 at in a new location at Guadalupe Mountain
National Park. The other is the Midland camera originally located near the edge of the
Permian Basin.
The UTEP program has archived in the past, five sets of daily images from the Paso del
Norte urban network and 4 sets of daily images from sites in west Texas. In north to
south order, the west Texas sites are Midland-Odessa (Permian Basin), Guadalupe
Mountain National Park (near Carlsbad Caverns National Park), an interstitial area
surveyed from The McDonald Observatory atop Mt. Locke near Ft. Davis, TX, and Big
Bend National Park (the view extends to the Sierra del Carmen in Mexico).
This project has digitally stored images from all the visibility imaging sites. The first,
and most extensive, results are from the archive of urban Paso del Norte digital images
taken at the Chelsea Retirement Home in El Paso with a westerly view of downtown El
Paso and the Sierra de Juarez. The last deployed system in August of 2002 was at The
McDonald Observatory, Ft. Davis, TX, and introduced a changeover from remote
systems using Windows 98 and Kodak DC290 cameras to systems using Windows 2000
and Olympus C2100 cameras.
The performance data for the digital camera systems is given in Section II in terms of
figures and charts showing number of days per month from inception (December 2000)
7
to February 28, 2003 that the unit was judged operational. Generally, this is interpreted
as over 50 percent of the possible images were archived. This study found similar image
quality for the Kodak DC260 used originally in the Big Bend Regional Aerosol and
Visibility Observation (BRAVO) study and currently at BBNP, the Kodak DC290’s
initially deployed for this project, and the Olympus C2100 recently deployed. Improved
reliability of systems (especially those ca. 300 miles from the UTEP laboratories) was the
anticipated result of the changeover to the Olympus C2100. However, improvements in
configuration of DC290 systems have been made and existing such systems are not being
replaced. The Big Bend data is, to date, from the National Park Service camera.
Permission to install a TCEQ camera has been obtained and therefore, images from a
second BBNP camera are expected later in FY 2003.
The Data Appendix DVD accompanying the report contains the spreadsheets with these
data and the image archive and also, a “. PDF” copy of the report. The images and data
for this digital visibility imaging project have been deposited with the University of
Texas El Paso Library, Special Collections, from which they can also be retrieved upon
request.
This project has put in place the initial TNRCC web server technology for the Visibility
Camera Program at http://cams.utep.edu, featuring a satellite image of the Paso del Norte
Air shed with site " yellow dot” icons. Summary of archive numbers are given in Section
II.
The image files are 40KB in size except Mc Donald’s where they are about 400KB
transferred over a LAN connection.
8
Table I.1: TCEQ West Texas Visibility Camera Systems
Name Location Description Camera Type CommunicationChelsea 1 Chelsea
Retirement Center Roof
Westerly View of Downtown El Paso and distant Sierras de Juarez
Kodak DC290
Modem
Chelsea 2 Chelsea Retirement Center Roof
Southerly View toward Ciudad Juarez
Kodak DC290
Modem
Ranger 1 Ranger Peak Southerly View toward Juarez from Ranger Peak Aerial Tramway
Kodak DC290
Modem
Ranger 2 Ranger Peak Southwesterly View of Juarez includes Downtown El Paso and UTEP
Kodak DC290
Modem
GMNP Guadalupe Mountain National Park
Easterly view from Dell City.
Kodak DC290
Modem
UGLC UTEP Undergraduate Learning Center Roof
Southerly view of IH-10, railway and Ciudad Juarez and mountains.
Panasonic WV CP450
LAN
McDonald McDonald Observatory catwalk of 107” telescope.
Southerly view of various mountain peaks and ranges
Olympus 2100
LAN
BBNP 1 Big Bend National Park (NPS site)
Southerly view of Sierra del Carmen Mountains
Kodak DC260
Modem
BBNP 2 Big Bend National Park (TCEQ site)
Easterly view Chisos Mountain Lodge
Kodak DC290
Modem
9
II. Summary of Site Operating Days
The following are the charts that represent the number of days the cameras were
operational, it also includes the time period for the current contract.
05
101520253035
Nov-01
Dec-01
Jan-0
2
Feb-02
Mar-02
Apr-02
May-02
Jun-0
2Ju
l-02
Aug-02
Sep-02
Oct-02
Nov-02
Dec-02
Jan-0
3
Feb-03
Mar-03
Apr-03
May-03
Jun-0
3Ju
l-03
Big Bend National ParkTotal Days with Images by month
Fig. II.1. Big Bend National Park camera (Kodak DC260). National Park Service maintains this camera.
Images are obtained by saving the image from their web site. Single archived images taken at 3:00 pm
local time, can be viewed from the National Park Service website:
http://www2.nature.nps.gov/ard/cams/bibe/bibejpgfram.cfm.
10
05
101520253035
Jul-0
1
Sep-01
Nov-01
Jan-0
2
Mar-02
May-02
Jul-0
2
Sep-02
Nov-02
Jan-0
3
Mar-03
May-03
Jul-0
3
Chelsea 1Total Days with Images by Month
Fig. II.2. Chelsea 1 is the original site in December 2000, but archiving records at UTEP only extend from
July 2001. Back up CD’s from the site does exist.
0
5
10
15
20
25
30
35
Jul-0
1
Sep-01
Nov-01
Jan-0
2
Mar-02
May-02
Jul-0
2
Sep-02
Nov-02
Jan-0
3
Mar-03
May-03
Jul-0
3
Guadalupe Mountain National ParkTotal Days with Images by Month
Fig. II.3. Guadalupe Mountain National Park was deployed in 2001 and uninstalled for a move to a new
location by TCEQ Region VI. It has now been relocated to a new site by TCEQ and UTEP in July 2003.
11
0
5
10
15
20
25
30
35
Jul-0
1
Sep-01
Nov-01
Jan-0
2
Mar-02
May-02
Jul-0
2
Sep-02
Feb-03
Apr-03
Jun-0
3
UTEP Undergraduate Learning CenterTotal Days with Images by Month
Fig. II.4. UTEP Undergraduate Learning Center (UGLC). This site is a live video camera from which
periodic images are archived. The UGLC camera is a Panasonic video camera connected to a Panasonic
NT104 network and Internet server. Images are displayed live via internal proprietary java scripting by
Panasonic. Images are collected from the web page using a recorded sequence includes an image save
routine. Reliability of the image archives is dependent on network and Internet conditions. Local archive
of the images is possible through a video recorder.
12
05
101520253035
Jul-0
1Sep
-01Nov
-01Ja
n-02
Mar-02
May-02
Jul-0
2Fe
b-03
Apr-03
Jun-0
3
Ranger Peak 1Total Days With Images by Month
Fig. II.5. Ranger Peak 1 in the Franklin Mountains overlooking El Paso and Ciudad Juarez.
0
5
10
15
20
25
30
Sep-01
Oct-01
Nov-01
Aug-02
Aug-02
Sep-02
Oct-02
Nov-02
Dec-02
Jan-0
3
Midland OdessaTotal Days with Images by Month
Fig. II.6. Midland-Odessa camera originally placed in the TCEQ offices in September 2001 for testing. In
late August, it was deployed at the Bob Derrington Water Treatment plant south of Odessa with a westerly
view of the plains east of Guadalupe Mountain National Park. TCEQ Region VI uninstalled this camera on
October 10, 2002 for potential relocation. Images during testing are not included in the tabulation. This
camera was relocated to Big Bend National Park (Stopped by lightning).
13
0
5
10
15
20
25
30
35
Jun-0
2
Aug-02
Oct-02
Dec-02
Feb-03
Apr-03
Jun-0
3
Chelsea 2Total Days with Images by Month
Fig. II.7. Chelsea 2 provides a southerly view toward Cd. Juarez.
14
0
5
10
15
20
25
30
35
Aug-02
Sep-02Oct-
02
Nov-02
Dec-02
Jan-03
Feb-03
Mar-03Apr-0
3
May-03
Jun-03
Jul-0
3
McDonald ObservatoryTotal Days With Images by Month
Fig. II.8. McDonald Observatory camera is located on the catwalk of the 107-inch telescope and provides a
southerly view of various mountain ranges and peaks. The system resides permanently on the local area
network and is accessible via the Internet.
05
101520253035
Aug-01
Oct-01
Dec-01
Feb-02
Apr-02
Jun-0
2
Aug-02
Oct-02
Dec-02
Feb-03
May-03
Jul-0
3
Ranger Peak 2Total Days with Images by Month
Fig. II.9. Ranger Peak 2 supplements the images from Ranger 1. It was temporarily removed by TCEQ in
December of 2002.
15
Observations and Conclusions:
The visibility camera systems have presented a number of challenges to provide for
reliability and image quality. The factors include Internet status, computer-camera
software, operating systems, communications, system security and others. In summary,
the Kodak DC290 systems have proven to be reliable with the ability to restart the
systems remotely. The Olympus system has been most robust. Restarting the computer,
loss of connectivity or other factors have no affect on the ability to activate the software
and restart collection of images. In general, the “ARS DIGICAM” software and systems
have been reliable, relatively easy to maintain, and provide high quality images for
documentation and analysis of visibility.
The archive results for this report are not completely representative of the complete
available archive of images from these systems. In some cases, a communication or file
transfer problem exists and images being recorded on the local attached computer were
not transmitted to the ftp and archives at UTEP. During the course of this project,
however, the missing archives will be obtained and added to those available at
ftp://tnrcc.utep.edu for download via the Internet.
16
III. Regions of Interest and Results for the Contrast Ratio and Coefficient of
Variation Analysis for the nine visibility imaging sites
Contrast Ratio (CR) analysis and Contrast Variance (CV) analysis are being performed
on the images at defined times each day that provide acceptable target contrast. The
theoretical foundation for our analyses derives from the classical approach described in
Von Koschmieder’s Habilitationschrift (Koschmieder, 1924, Pt. I and II). The first use
of the present adaptation by us was described by Turner and Parks (Turner, 1998) for
digitized TNRCC video images. Application of an older form of NIH IMAGE©
has been shown to be a facile means to get the basic data out of images as integrals of
pixels brightness from selected ROI’s (Parks 2002, and Sawant 2002). The current work
uses IMAGE-J© (Image 2002), a JAVA language based incarnation of the earlier form
which features a variety of operational improvements.
Contrast Ratio and Contrast Variation are collectively referred to as Visual Air Quality
(VAQ) factors. They are defined by the following equation:
Contrast Ratio: CR = Mean (Bright ROI) – Mean (Dark ROI) Eq. III.1
Mean (Bright ROI)
Contrast Variation: CV = SD (ROI) Eq. III.2
Mean (ROI)
Where;
SD (ROI) = Standard Deviation of the pixel intensities (0 – 255).
Mean (ROI) = the mean of the pixel light intensities.
Where possible (dark and light targets), CR is used. Where no discrete targets exist, CV
is used.
17
In some instances the CR values obtained are negative, this is attributed to the fact that on
that particular day the sky was darker as compared to the ridge and hence the change in
values of the mean and thereby in the CV.
Methods & Materials:
1. The images were analyzed using ImageJ 1.29x Analysis software.
2. The data in Section IV has been analyzed using SPSS statistical package.
18
Chelsea (Site 1):
This camera located atop the Chelsea Retirement Home, produces a westerly view of
downtown El Paso and distant Sierra de Juarez. The classic view has now been used for
over 10 years, it is the best-studied and understood site. Regions of Interest are marked
and labeled in the various figures.
CR Analysis - The four ROI’s that have been used for the TNRCC (now TCEQ)
contracts for 2001 and 2002, have been retained.
Fig. III.1
CR values are being computed for the downtown target buildings (A; B) located
approximately 5 km from the Chelsea site. CR is also being computed for the edge of the
Sierra de Juarez for a ridge-sky two-target basis (C; D) at ca. 12 km.
CV Analysis - Contrast Variance (CV) would be used to look at the spread of color pixel
values for urban zones in El Paso (E) and Ciudad Juarez (F).
19
A ROI encompassing both the black and the white target has been evaluated and can be
used to calculate the CV values, for this view.
In addition the complete image could be used as ROI for CV or Fourier Transform and
spatial frequency evaluation. Spatial frequency can approach zero in parts of the image
during heavy inversions and the total image during sand storms (wind events).
20
Results:
CR & CV Values For BlckWht Bldg Dec 2002
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
12/1/
2002
12/3/
2002
12/5/
2002
12/7/
2002
12/9/
2002
12/11
/2002
12/13
/2002
12/15
/2002
12/17
/2002
12/19
/2002
12/21
/2002
12/23
/2002
12/25
/2002
12/27
/2002
12/29
/2002
12/31
/2002
Days
CR
/CV CR BW
CV BW
Fig. III.2 CR & CV Values for the Black & White building-Dec 2002
CR & CV Values for Blck Wht Bldg Jan 2003 - Mar 2003
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
1/1/
2003
1/8/
2003
1/15
/200
3
1/22
/200
3
1/29
/200
3
2/5/
2003
2/12
/200
3
2/19
/200
3
2/26
/200
3
3/5/
2003
3/12
/200
3
3/19
/200
3
3/26
/200
3
Days
CR
/CV CR BW
CV BW
Fig. III.3 CR & CV Values for the Black & White building-Jan – Mar 2003
21
CR & CV Values for Blck Wht Bldg Apr 2003-June 2003
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
4/1/
2003
4/8/
2003
4/15
/200
3
4/22
/200
3
4/29
/200
3
5/6/
2003
5/13
/200
3
5/20
/200
3
5/27
/200
3
6/3/
2003
6/10
/200
3
6/17
/200
3
6/24
/200
3
Days
CR
/CV CR BW
CV BW
Fig. III.4 CR & CV Values for the Black & White building-Apr – June 2003
CR & CV Values of Blck Wht Bldg July 2003
0
0.1
0.2
0.3
0.4
0.5
0.6
7/1/20
03
7/3/20
03
7/5/20
03
7/7/20
03
7/9/20
03
7/11/2
003
7/13/2
003
7/15/2
003
7/17/2
003
7/19/2
003
7/21/2
003
7/23/2
003
7/25/2
003
7/27/2
003
7/29/2
003
7/31/2
003
Days
CR
/CV CR BW
CV BW
Fig. III.5 CR & CV Values for the Black & White building-July 2003
22
Quality Assurance for the Equipment:
The quality assurance protocol assumes that a sufficiently close target to the camera will
be minimally affected by haze. This is approximately true except for unique events like
sand storms, snow etc. An example from Chelsea 1 is given. An ROI under the bridge
next to the freeway is selected and CV Analysis revealed that the camera functioned
satisfactorily.
Fig. III.5.a
23
The result of the QA test is displayed below;
CV
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
12/5
/200
2
12/1
9/20
02
1/2/
2003
1/16
/200
3
1/30
/200
3
2/13
/200
3
2/27
/200
3
3/13
/200
3
3/27
/200
3
4/10
/200
3
4/24
/200
3
5/8/
2003
5/22
/200
3
6/5/
2003
6/19
/200
3
7/3/
2003
7/17
/200
3
Days
CV CV
Fig. III.6 CV values for the Contract Period for ROI under bridge
24
Chelsea (Site2):
This camera is located on the Chelsea Building but has a different view that in the
Southerly direction toward Ciudad Juarez.
Fig. III.7
CR Analysis - The CR Analysis has been performed on the mountain (A) and the sky (B). CV Analysis - The ridge of the mountain with respect to the sky (C) has been used to calculate the value of CV.
25
RESULTS:
CR & CV for Ridge-Sky & CV for the common ROI (Dec 2002)
-0.05
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
12/1/
2002
12/3/
2002
12/5/
2002
12/7/
2002
12/9/
2002
12/11
/2002
12/13
/2002
12/15
/2002
12/17
/2002
12/19
/2002
12/21
/2002
12/23
/2002
12/25
/2002
12/27
/2002
12/29
/2002
12/31
/2002
Days
CR
/CV
CR BW
CV COMMON
Fig. III.8 CR & CV values for the Ridge-Sky ROI-Dec 2002
CR/CV for Chelsea 2 Ridge-Sky and Common ROI
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
1/1/20
03
1/8/20
03
1/15/2
003
1/22/2
003
1/29/2
003
2/5/20
03
2/12/2
003
2/19/2
003
2/26/2
003
3/5/20
03
3/12/2
003
3/19/2
003
3/26/2
003
Days
CR
/CV
CR BW
CV COMMON
Fig. III.9 CR & CV values for the Ridge-Sky ROI-Jan – Mar 2003
26
CR for Ridge-Sky and CV for common ROI Apr 03-Jun 03
0
0.05
0.1
0.15
0.2
0.25
4/1/03 4/10/03 4/19/03 4/28/03 5/7/03 5/16/03 5/25/03 6/3/03 6/12/03 6/21/03 6/30/03
Days
CR
/CV
CR BW
CV COMMON
Fig. III.10 CR & CV values for the Ridge-Sky ROI-Apr – June 2003
CR Ridge-Sky CV Common ROI Jul 2003
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
7/1/20
03
7/3/20
03
7/5/20
03
7/7/20
03
7/9/20
03
7/11/2
003
7/13/2
003
7/15/2
003
7/17/2
003
7/19/2
003
7/21/2
003
7/23/2
003
7/25/2
003
7/27/2
003
7/29/2
003
7/31/2
003
Days
CR
/CV
CR BWCV COMMON
Fig. III.11 CR & CV values for the Ridge-Sky ROI-July 2003
27
Ranger Peak (Site1):
This camera has a Southerly View toward Juarez from Ranger Peak Aerial Tramway.
Fig. I.12
CR Analysis - Fig 3 shows the region of r the view from ranger peak 1. The contrast was taken between the mountain and sky nd black building and White building.
in
owntown El Paso and another Block in Ciudad Juarez. Also, a new ROI is to be added
II
interest fo a
CV Analysis - The Coefficient of Variation analysis was performed on the block
D
that encompasses the mountain ridge.
28
Results:
Ranger1 CR Black-White Bldg. & CV Sky-Ridge Common ROI Jan 2003- Mar 2003
0
0.1
0.2
0.3
0.4
0.5
0.6
1/1/20
03
1/8/20
03
1/15/2
003
1/22/2
003
1/29/2
003
2/5/20
03
2/12/2
003
2/19/2
003
2/26/2
003
3/5/20
03
3/12/2
003
3/19/2
003
3/26/2
003
Days
CR
and
CV
CRBW
CVCOMMON
Fig. III.13 CR & CV values for Ranger 1 Jan – Mar 2003
Ranger1 CR Black-White Bldg. & CV Sky-Mountain Common ROI Apr-June 2003
0
0.05
0.1
0.15
0.2
0.25
0.3
4/1/20
03
4/8/20
03
4/15/2
003
4/22/2
003
4/29/2
003
5/6/20
03
5/13/2
003
5/20/2
003
5/27/2
003
6/3/20
03
6/10/2
003
6/17/2
003
6/24/2
003
Days
CR
and
CV
CRBWCV Common
Fig. III.14 CR & CV values for Ranger 1 Apr-May 2003
29
Ranger1 CR Black-White Bldg. & CV Sky-Mountain Common ROI July 2003
0
0.1
0.2
0.3
0.4
0.5
0.6
7/1/20
03
7/3/20
03
7/5/20
03
7/7/20
03
7/9/20
03
7/11/2
003
7/13/2
003
7/15/2
003
7/17/2
003
7/19/2
003
7/21/2
003
7/23/2
003
7/25/2
003
7/27/2
003
7/29/2
003
7/31/2
003
Days
CR
and
CV
CRBWCV Common
Fig. III.15 CR & CV values for Ranger 1 July 2003
30
Ranger Peak (Site2):
he Camera has a Southwesterly View of Juarez including Downtown El Paso and
TEP.
Fig. III.16
CR Analysis – The contrast ratio was performed for the mountain ridge with respect to
e.
T
U
the sky. Two different lengths of ROI’s were taken as shown in the Fig. III.16 abov
31
RESULTS:
RANGER2 CR Ridge-Sky Shrt & CR Ridge-Sky Long Apr-Jun 2003
0
0.05
0.1
0.15
0.2
0.254/
26/2
003
5/3/
2003
5/10
/200
3
5/17
/200
3
5/24
/200
3
5/31
/200
3
6/7/
2003
6/14
/200
3
6/21
/200
3
6/28
/200
3
7/5/
2003
7/12
/200
3
7/19
/200
3
7/26
/200
3
Days
CR
RIJ
SKY(
SHR
T&LN
G)
CR RIJSKY(SHRT)0.3
CR RIJSKY(LNG)
Fig. III.17 CR Ridge-Sky short & CR Ridge-Sky long ROI Apr-Jun 2003
RANGER2 CR Ridge-Sky Shrt & Ridge-Sky Long July 2003
0
0.05
0.1
0.15
0.2
0.25
7/1/20
03
7/3/20
03
7/5/20
03
7/7/20
03
7/9/20
03
7/11/2
003
7/13/2
003
7/15/2
003
7/17/2
003
7/19/2
003
7/21/2
003
7/23/2
003
7/25/2
003
7/27/2
003
7/29/2
003
7/31/2
003
CR
RIJ
SKY(
SHR
T&LN
G)
CR RIJSKY(SHRT)
CR RIJSKY(LNG)
Fig. III.18 CR Ridge-Sky short & CR Ridge-Sky long ROI-July - 2003
32
33
UGLC:
This is the Southerly view of IH-10, railway and Ciudad Juarez and mountains.
Fig. III.19
CR Analysis: The CR analysis was performed for the mountain ridge with respect to the
sky.
CV Analysis: The CV analysis was performed by taking a ROI that encompasses the
ridge of the mountain and the sky.
RESULTS:
UGLC CR Ridge-Sky & CV Common ROI Jan-Mar 2003
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
1/1/20
03
1/8/20
03
1/15/2
003
1/22/2
003
1/29/2
003
2/5/20
03
2/12/2
003
2/19/2
003
2/26/2
003
3/5/20
03
3/12/2
003
3/19/2
003
3/26/2
003
Days
CR
RIJ
SKY
AN
D C
V C
OM
MO
N
CR RIJSKYCV COMMON
Fig. III.20 CR Ridge-Sky mon ROI Jan-Mar 2003 & CV Com
UGLC July 2003
0
0.05
0.1
0.15
0.25
7/1/03 7/6/03 7/11/03 7/16/03 7/21/03 7/26/03
Days
CR
RIJ
SKY
& C
V C
OM
MO
CR RIJSKY
CV COMMON
Fig. III.21 CR Ridge-Sky & CV Common ROI July 2003
0.2
N
34
McDonald Observatory:
This view gives the southerly view form the observatory of various mountain peaks and
ranges.
Fig. III.22 The McDonald Observatory, Department of Astronomy, University of Texas at Austin, atop Mt. Locke (altitude ca. 2200 meters). The central peak in the “notch” of Twin Mountains is Cathedral Peak in Big Bend National Park at a distance of ca. 65 km.
CR Analysis:
This was performed by taking a ROI on the twin mountain and compare it to that of the
ROI of sky above it.
The same procedure was adopted for the Cathedral Peak, which is at a distance of 65 km
from the camera site.
CV Analysis was performed by taking a ROI encompassing the twin mountain ridge and
the sky together.
35
RESULTS:
C R T w in M o u n t/C R C ath ed ra l Peak F eb 2003 - M ar 2003
-0 .2
-0 .1
0
0 .1
0 .2
0 .3
0 .4
0 .5
2/1/20
03
2/8/20
03
2/15/2
003
2/22/2
003
3/1/20
03
3/8/20
03
3/15/2
003
3/22/2
003
3/29/2
003
D ay s
CR
/CV
C R T MC R C P
Fig. III.23 CR Twin Mountains/CR Twin Mountains Feb – Mar 2003
CR Twin Mount/CR Cathedral Peak Apr-June 2003
0
0.3
4/1/03 4/9/03 4/17/03 4/25/03 5/3/03 5/11/03 5/19/03 5/27/03 6/4/03 6/12/03 6/20/03 6/28/03
0.25
0.15
0.2
CR
/CV
0.05
0.1
0.35
0.4CR TM
CR CP
Fig. III.24 CR Twin Mountains/CR Twin Mountains Apr – Jun 2003
Days
36
CR Twin Mount/CR Cathedral Peak July 2003
-0.25
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2
0.25
7/1/20
03
7/2/20
03
7/3/20
03
7/4/20
03
7/5/20
03
7/6/20
03
7/7/20
03
7/8/20
03
7/9/20
03
7/10/2
003
7/11/2
003
7/12/2
003
7/13/2
003
7/14/2
003
7/15/2
003
7/16/2
003
7/17/2
003
7/18/2
003
7/19/2
003
7/20/2
003
7/21/2
003
7/22/2
003
7/23/2
003
7/24/2
003
7/25/2
003
7/26/2
003
7/27/2
003
7/28/2
003
7/29/2
003
7/30/2
003
7/31/2
003
Days
CR
/CV
CR TM
CR CP
Fig. III.25 CR Twin Mountains/CR Twin Mountains July 2003
37
Quality Assurance for the Equipment:
The Quality Assurance was performed by taking an ROI on the handrail, which is at a
distance of about 2 meters from the camera, and a CV Analysis revealed that the camera
functioned satisfactorily during the duration of the project. A couple of images were
taken every month, for the entire contract period.
Fig III.26 The ROI on the handrail at Mc Donald Observatory
Results:
CV
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
2/5/
2003
2/19
/200
3
3/5/
2003
3/19
/200
3
4/2/
2003
4/16
/200
3
4/30
/200
3
5/14
/200
3
5/28
/200
3
6/11
/200
3
Day
CV CV
s
Fig. III.27 CV values fo e ROI on the handrail
r th
38
BBNP:
39
Fig. III.28
R Analysis: The contrast ratio analysis was performed by taking the contrast of an ROI
untain with respect to that of the sky.
CV Analysis: was performed by taking the ROI encompassing both the sky and the ridge
together.
C
on the mo
Results:
CR Ridge-Sky/ CV Common Mar 2003
-0.2
-0.1
0
0.1
0.2
0.3
0.4
3/1/03 3/4/03 3/7/03 3/10/03 3/13/03 3/16/03 3/19/03 3/22/03 3/25/03 3/28/03 3/31/03
Days
CR
/CV
CR RIJSKY
CV COMMON
-0.3
Fig. III.28 CR Ridge-Sky/CV Common ROI Mar
2003
40
CR Ridge-Sky/CV Common ROI Apr 2003-June 2003
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
4/1/
2003
4/8/
2003
4/15
/200
3
4/22
/200
3
4/29
/200
3
5/6/
2003
5/13
/200
3
5/20
/200
3
5/27
/200
3
6/3/
2003
6/10
/200
3
6/17
/200
3
6/24
/200
3
Days
CR
/CV
CR RIJSKYCV COMMON
Fig. II 003 I.29 CR Ridge-Sky/CV Common ROI Apr - June 2
CR Ridg-Sky/CV Common ROI July 2003
-0.04
-0.02
0
0.02
0.04
0.06
0.08
0.1
7/1/03 7/5/03 7/9/03 7/13/03 7/17/03 7/21/03 7/25/03 7/29/03
Days
CR
/CV
CR RIJSKY
CV COMMON
Fig. III.30 CR Ridge-Sky/CV Common ROI July 2003
41
42
New Big Bend Site (Chisos Lodge) – Westerly View:
Fig. III.31
CR Analysis: This was performed by taking a ROI on the mountain and compare it to that
of the ROI of sky a
CV Analysis: This was performed by taking an ROI encompassing the ridge and the sky
at the distance.
bove it.
Results:
CR Ridge-Sky/CV Ridg ommon ROI July 2003
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
7/13/2003 7/14/2003 7/15/2003 7/16/2003 7/17/2003 7/18/2003 7/19/2003 7/20/2003 7/21/2003 7/22/2003 7/23/2003
Days
CR
/CV
CR RijSky
CV Ridge
e C
Fig. III.32 CR Ridge-Sky/CV Common ROI July 2003
43
Guadalupe Mountain National Park (New):
Fig. II 33
ast 2002 views were early December 2002. This camera was uninstalled by TCEQ
egion 6 for re-deployment at a location approximately 15km west of the current site.
he Camera has been reinstalled at a new location in July 2003 by TCEQ and UTEP.
R Analysis. : The CR was computed using the ROI’s shown by the small squares and
quation III.1.
V Analysis: This was performed by taking an ROI encompassing the ridge and the sky
t the distance of the mountain and Equation III.2
I.
L
R
T
C
E
C
a
44
Results:
CR Ridge-Sky/CV Common ROI over mountain and sky July 26 - July 31 2003
0
0.05
0.1
0.15
0.2
0.25
7/26/2003 7/27/2003 7/28/2003 7/29/2003 7/30/2003 7/31/2003
Days
CR
/CV
CR RIJSKY
CV COMMON
Fig. III.34 CR Ridge-Sky/CV Common ROI July 26-31, 2003
45
IV: Additional Analysis of Air Quality Parameters and Visibility
Visibility impairment primarily takes place as a consequence of light scattering by
particles in the Accumulation Mode. It has been reported recently that for El Paso there is
a significant correlation between CO and both ultra-fine and accumulation mode (those
between 0.1 and 1 microns in diameter) particle count. The Pearson correlation
coefficient (r) values reported are 0.81 (r2 ca. 0.64) and 0.87 (r2 ca.0.7) respectively.
(Noble et. al. 2003). This finding is particularly important in the examination of the
relationship of TCEQ reported air quality data and the results of visibility image analyses
for the traditional downtown black-white building target (Chelsea 1). Data sets from most
CAMS stations contain CO, whereas, PM 2.5 (a putatively important indice of visibility)
is only available at field of view.
he analytical work in this section is largely derived from a current, in progress Master of
cience project by one of us (Raina 2003). It is included as a useful tool for the possible
pplications or interpretation of the analytical data described in this section. The work
escribed herein is based on the long term (12 years) traditional westerly field of view
om Chelsea 1 toward downtown El Paso initiated by Stuart Dattner (TCEQ Technical
nalysis Division).
n a priori interest in PM 2.5 derives for the fact that the visible part of the particulate
atter size fraction is 0.4 to 0.8 µm approximate physical diameter and is contained in
e PM 2.5 fraction obtained by current sampling systems based on effective aerodynamic
iameter separations. Various multiple linear regression analyses have been performed
ith the data from the contract period with variables, PM 2.5, PM10, CO, NOx, Wind
peed, Relative Humidity, Sun Angle and Azimuth against the dependent variable,
ontrast Ratio (CR)(see for example, Table IV.1). The CR data was that of the images
nalyzed. The air quality data was obtained from the TCEQ monitored CAMS sites in El
aso (CAMS 12, 37, 40,41).
CAMS sites peripherally positioned to the Chelsea 1
T
S
a
d
fr
A
A
m
th
d
w
S
C
a
P
46
The additi tors
fluencing CR in the downtown Chelsea images over the simple comparison of CR-
omplement (1 – CR chosen for presentation convenience) and PM 2.5 from CAMS 12
izal
roup of warm days. The difference in winter and summer visibility situations is readily
ry
plot of the uncertainty distribution about the curve of maximum
kelihood (straight lines in the 2D plots). With the geographic constraints on PM 2.5 and
)
ion
ip of CR and typical reported air quality and
eteorological values.
onal variables notably improved the understanding of the important fac
in
C
(UTEP) and CAMS 41 (Sun Metro) or the comparison to average PM10 from Cham
(CAMS 40) and UTEP (CAMS 12). The latter two CAMS sites are located on north and
south sides, respectively, of the field of view of Chelsea 1.
Preliminary Data Analysis:
The search for correlations of the CR data with air quality and meteorological data was
addressed with the statistical software, SPSS. The various analyses described herein were
performed for a December, January, February group of days and for a May, June, July
g
apparent from the images (Section III) and the hours of sunlight prior to the 9 AM
analysis time chosen vary from 0.5 to 3.
The first results obtained are presented in Figures IV.1 a, b, c, d. They are the elementa
comparison of PM 2.5 or PM10 with CR. These results are presented two ways. One as a
plot and one as a 3D
li
PM10 measurements (deriving from CAMS site location), it is not surprising that the
Correlation Coefficients (indicating the variability of Y explained by association with X
for CR versus PM are ca. 0.15 to 0.3.
Subsequent addition of all the air quality and meteorology data improved the correlat
coefficients to 0.6 to 0.7. (See Figure IV.2). However, it was suspected that the
correlation of CR with vapor phase components of the urban aerosol — notably fine
particle coupled CO — available from most CAMS sites, might represent an improved
demonstration of the relationsh
m
47
Multiple regression results for three sites, CAMS 12, CAMS 37, and CAMS 41,
representing sites to the north, behind, and ahead of the field of view of Chelsea 1 are
given in Table IV.1. These results do not include any PM data. The
t
t” through the field of view
at is integrated by the methods used herein and described in Section III.
R2 value, in at leas
one case (Sun Metro), improved by removing the PM 2.5 data as a variable. This is
presently attributed to the site of PM 2.5 data collection being physically located behind
the downtown buildings and removed from the “line-of-sigh
th
48
ig. IV.1a. Scatter plot of C-Comp versus PM 2.5 from CAMS 12 (above)
ig. IV 1b. Probability distribution about the red line of Fig IV.1a above.
F
01.
12.
23.
34.
45.
56.
67.
7
8.8
9.9
11
12.1
13.2
14.3
15.4
16.5
0
0.65
00.10.20.30.40.50.60.70.8
0.9
1
Rel. Freq
PM2.5
Ccomp
CR Complement with PM2.5
F
49
ig. IV.1c. Scatter plot of PM 2.5 average data from CAMS 12 and CAMS 40 versus C-comp
Fig. IV.1d. Probability distribution about the red line f Figure IV.1c.
F
0
9
18
27
36
45
54
63
72
81
90
99
108
117
126
135
144
0
0.55
0
0.02
0.04
0.06
0.08
0.1
0.12
Rel. Freq
PM10
Ccomp
CR Complement with PM10
o
50
R2 Values
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
PM2.5 w ithRel.Humidity
PM2.5 w ith Met Data PM10 w ithRel.Humidity
PM10 w ith Met Data
Variable
Con
cent
ratio
n
Winter
Summer
Fig. IV.2. Comparison of Correlation Coefficients from multiple regression of either a limited air quality
data set or data sets (marked “with met”) that include all the meteorological and air quality data.
uring the winter (Dec, 02-Feb, 03), PM 2.5 and relative humidity explains only 30% of
e variability associated with visibility. On the other hand up to 70% of the variability
an be explained with all meteorological and air quality data.
able V.1. Multiple regression coefficients (c) and Coefficient of Determination (R2) for comparison of
R to various gas phase urban aerosol components and meteorological data
Parameters/coefficients CAMS12 (UTEP)
CAMS37 (Ascarate)
CAMS41 (Chamizal)
D
th
c
T
C
Constant 0.553955315 0.331308595 0.637027331 CO_1_PPM -0.144260345 0.032924972 -0.252941274 NOx_1 ppb -0.001014248 -0.001524366 0.000884885 O3_2_PPB -0.003454084 -0.002079178 0.002681624
WSR_1 mph -0.01093396 0.002399688 -0.013540955 Out Temp_1 deg F -0.000834302 -0.002557625 -0.003147041
Rel. Humid_1 % -0.003277 48 -0.002779647 -0.003139329 0Solar Rad_1 Ly/min
20.673898816 0.890823854 0.602488015
R 0.643 0.541 0.762
51
52
In the summer (May, 03-Jul, 03) variability up to 15% can be explained when PM 2.5 is
analyzed with relative humidity. Similarly the explainable variability for all
meteorological factors is 50 to 70%. Subsequently, three CAMS sites data sets for the
vapor phase and meteorological data alone (no PM) was compared to CR. These
intriguing results are contained in Table IV.1. These preliminary results suggest that
indeed, the vapor phase fraction (most likely CO) is an important influence on CR and
visibility.
Lastly, it has been possible to make a comparison of CR for the downtown Chelsea 1
field of view with the Visibility data reported on the TCEQ air data web URL from
Ascarate Park. Ascarate (CAMS 37) is located physically behind Chelsea 1 in an easterly
direction. The visibility unit has a relatively short path length (on the order of meters) in
comparison to the path length from Chelsea 1 to downtown, which is ca. 6000 meters.
The results are shown in Figure V.3. Extensive data cleaning has not been performed.
For example, the one point a approximately CR = 0 is presumed to be an event that
dramatically affected the 6000 meter Chelsea 1 view and not the orders of magnitude
orter Ascarate Park measurement. Given the physical separation and the different
Fig.
iles from the monitor at Ascarate Park (CAMS 37).
sh
measurement methods, the agreement is notable.
y
0.6
0.7
values from C -White building d Visibility
CR V S V is ib ilty
= 0.1695x + 0 .1304R2 = 0 .4224
0.5
0.8
1 2.5 4
y
V.3. Comparison of CR helsea 1 (Black ) and the reporte in
00 0.5
0.1
0.2
0.3
0.4CR
1.5 2 3 3.5
V is ib ilt
m
The four outliers correspond to days, which are relatively infrequent in El Paso. They
represent situations, which are typically “off-scale” for the present 6000-met
to compute CR or CV for downtown. They are included here for completeness. Withou
these points, R2 is ca. 0.45.
er path used
t
53
54
29x. Rasband W., National Institutes of Health; Washington,
.C.
oschmieder, Von H. (1924). Theory der horizontalen sichtweite. Beitrage zur physik
er freien atmosphare. Vol. 12, p.p.: 33-53 (Part I) and 171-181 (Part II).
i (2001). Li Wen-Whai, Orquiz R., Garcia J., Espino, T., Pingitore, N.E., Gardea-
orresdey, J., Chow, J., and Watson, J.G., – “Analysis of temporal and spatial
ichotomous PM air samples in El Paso-Cd. Juarez air quality basin” – AMWA Nov.
arks (2002). Parks, N.J., Gray, R.W., Li, W.W., Sawant, R.R., – “Visibility Research ith the Texas Commission on Environmental Quality Camera Systems in the Paso del orte and West Texas Regions 2001-2002”.
oble (2003) Noble, C.A., Mukerjee, S., Gonzales, M., Rodes, C.E., Lawless, P.A.,
atarajan,S., Myers, E.A., Norris, G.A., Smith, L., Ozkaynak, H., Neas, L.M. –
Continuous measurement of fine and ultrafine particulate matter, criteria pollutants and
eteorological conditions in urban El Paso, Texas” – Atmospheric Environment, 37,
sue 6, February, Pages 827-840.
awant (2002) Sawant, R.R. – “Haze and Visibility in wilderness and urban areas” –
aster’s Thesis, University of Texas at El Paso.
urner (1998) Turner, C.D., Parks, N.J. – “Trans-Boundary visibility Analysis” –
outhwestern Consortium for Environmental Research and Policy, 1995-1996.
V. References
IMAGE-J (2002). ImageJ 1.
D
K
d
L
T
d
PwN
N
N
“
m
Is
S
M
T
S