oEPA
United States
Environmental Protection
Agency
Environmental Monitoring
Systems Laboratory
PO Box 15027
Las Vegas, NV 89114
TS-AMD-81035
August 1982
ADIRONDACK LAKES ACID RAIN
MULTISPECTRAL SCANNER SURVEY
Adirondack State Park, New York
prepared for U.S. Environmental Protection Agency
Environmental Research Laboratory
Corvallis, Ore.

-------
TS-AMD-81035
August 1982
ADIRONDACK LAKES
ACID RAIN MULTISPECTRAL SCANNER SURVEY
Adirondack State Park, New York
by
T. H. Mace
Environmental Programs
Lockheed Engineering and Management Services Company, Inc.
Las Vegas, Nevada 89114
Contract No. 68-03-3049
Project Officer
G. A. Shelton
Advanced Monitoring Systems Division
Environmental Monitoring Systems Laboratory
Las Vegas, Nevada 89114
ENVIRONMENTAL MONITORING SYSTEMS LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
LAS VEGAS, NEVADA 89114

-------
NOTICE
This document has not been peer and administratively reviewed within EPA and is
for internal Agency use and distribution only.
i i

-------
ABSTRACT
During the fall of 1981, the U.S. Environmental Protection Agency's Environ-
mental Monitoring Systems Laboratory at Las Vegas, Nevada, conducted an aerial
overflight and multispectral analysis of a portion of Adirondack State Park, New
York with its 11-channel multispectral scanner to determine if relationships existed
between acid rain-related water quality data and scanner imagery . Concurrent water
sampling and laboratory analysis was conducted by Paul Smith's College and the USEPA
Environmental Research Laboratory, Corvallis, Oregon. Strong linear relationships
were found between the multispectral scanner data and the water quality parameters of
Secchi depth and dissolved organic carbon. Weaker linear relationship^ were found
between the scanner data and phaeophyton, chlorophyll a, total suspended solids, jpHTJ
and alka 1 i ri ity.
Radiances in the red and near-infrared regions of the spectrum proved most
useful, and the results of this study suggest that the Thematic Mapper multispectral
scanner aboard the Landsat D satellite may be useful for monitoring of water quality
parameters related to acid precipitation effects.
i ii

-------
CONTENTS
Abstract		iii
Figures		iv
Tables		iv
Introduction 		1
The Nature of the Problem		1
The Current Study - Scope 		3
Conclusions and Recommendations 		4
Conclusions 		4
Recommendations 		5
Methodology 		6
Data Acquisition 		6
Preprocessing of Multispectral Data 		6
Data Analysis		7
Results		11
Regression Analysis 		11
Discussion		22
References		25
Appendices
A.	Multispectral Scanner and Analysis System Characteristics 		26
B.	Multispectral Radiances and Water Sample Data 		30
FIGURES
Number	Page
1	Predicted vs observed Secchi depth (m) 		14
2	Predicted vs observed pH		15
3	Predicted vs observed akalinity (ueq/1) 		16
4	Predicted vs observed total suspended solids (ppm) 		17
5	Predicted vs observed chlorophyll a (mg/m3) 		18
6	Predicted vs observed phaeophyton (mg/m3) 		19
7	Predicted vs observed dissolved organic carbon (ppm) 		20
8	Multispectral Classification of Lake pH, Adirondack State
Park, New York		21
TABLES
Number	Page
1 Best Subset Multiple Linear Regression Summary 	 12
iv

-------
INTRODUCTION
THE NATURE OF THE PROBLEM
The effects of acid rain on the environment are among the most pressing environ-
mental issues facing the industrial countries. Atmospheric pollution, caused by
fossil fuel combustion, is carried across state and national boundaries producing
national as well as international environmental degredation.
Scores of articles have been published outlining the possible effects of acid
precipitation on sensitive areas, and no attempt to reproduce them will be made here.
It may, however, be useful to outline some of the effects suggested by the litera-
ture.
Generally, sulfur and nitrogen oxides are produced in the process of fossil fuel
combustion. These oxides combine with other atmospheric constituents to produce
sulfuric and nitric acids. Normally, these acids would be buffered by atmospheric
carbon dioxide to produce a pH of approximately 5.6 in rain. However, the pH of
rainfall throughout most of the eastern United States has been reported at about 4.0
(Galloway jit a^. , 1976).
This "acid rain" phenomenon has many, often conflicting, effects. Among these
effects are increased erosion of statuary and historic buildings, crop and forest
productivity reduction, increased acidification and leaching of soils, and degreda-
tion of biota in rivers and streams. These effects are particularly severe in those
areas of crystalline bedrock and naturally acidic soils, where cation exchange
capacity (buffering) is relatively low. Lakes and streams in these areas are very
poorly buffered, and the plant and animal species reduction which results is severe.
As the lakes become more acid, organic matter decomposition decreases. Leaf
1

-------
litter increases on the lake floor and a fungal mat may spread along the bottom.
Additionally, magnesium and aluminum ions are liberated from the bottom sediments,
increasing their concentrations in the lakes, while phosphorous decreases. The
effect on biota is dramatic. The distribution and extent of macrophytes, phyto-
plankton, aquatic invertebrates, and fish decreases markedly (Baalsrud et al, 1976).
These effects have been severe in the Adirondack region of New York both in
extent and rapidity of development. "A recent survey found that 51% of mountain
lakes (217 lakes at an elevation above 600m) have pH values below 5.0; 90% of these
lakes contain no fish. In contrast, during the period 1929-1937, only 4% of these
lakes had a pH under 5.0 or were devoid of fish" (Likens, 1976, p. 43).
This rapid change in environmental conditions creates an environmental
monitoring problem. At one extreme, investigators have suggested that water
chemistry measurements should be made in lakes 5-10 times each year and during
periods of limnological change (Dochinger and Seliga, 1976). However, which lakes
should be sampled, and how can we optimize such a costly program? Remotely sensed
measurements may provide a cost-effective environmental monitoring tool to reduce the
number of water samples required for a regional monitoring program. Such programs
have been successful in the past for lake trophic classification (Witzig and
Whitehurst, 1981) , and it is possible that digital analysis techniques may be
applicable to the monitoring acid rain effects as well.
2

-------
THE CURRENT STUDY - SCOPE
This study represents the first of a series of experiments to determine if lake
acidification could be detected directly by digital interpretation of multispectral
scanner (MSS) data. Previous studies (Mace, 1982; Alfoldi and Munday, 1978) have
indicated that regression-based techniques may be successfully used to map a variety
of water quality parameters from MSS data. Therefore, this investigation used
regression-based digital interpretation procedures to establish the relationship
between airborne MSS data collected over Adirondack State Park, New York and a series
of lake measurements.
The area studied consists of 40 lakes within a rectangle bounded	on the east
and west by Raquette Lake and Old Forge and on the north and south by	Cranberry Lake
and State Highway 8. Four adjacent north-south flight lines provided	scanner cover-
age of the 97.5 by 25.0 km study area.
Automated analysis of the digital imagery was used to relate the concurrent
water sample data (Secchi depth, pH, alkalinity, total suspended solids, chlorophyll
a, phaeophyton, and dissolved organic carbon) to radiances detected by the airborne
multispectral scanner.
3

-------
CONCLUSIONS AND RECOMMENDATIONS
CONCLUSIONS
From the evidence produced by this study and a review of pertinent literature,
it is concluded that:
o Weak linear relationships exist between multispectral scanner data and
lake pH and alkalinity.
o Secchi depth and dissolved organic carbon are reasonably well-defined by
linear combinations of the multispectral scanner data.
o A non-linear multispectral model may yield a better fit for chlorophyll a,
total suspended solids, and phaeophyttfn.
i
o All of the regression models would be considerably improved by a greater
range of measured values in the water quality variables.
o Multispectral data from the ultraviolet-blue (scanner channels 1 and 2),
red (scanner channels 6 and 7), and near infrared (scanner channels 9 and
10) provided the best linear estimates of the water quality parameters
studied.
o The Thematic Mapper on the Landsat D satellite has spectral channels
similar to MSS channels 7 and 9, and may provide an excellent water quality
monitoring tool.
A l +
o -Wh-irl-e-the results of this study do not establish the use of the MSS as an
operational tool for monitoring lake pH, the correlations are sufficiently
good to offer promise toward its eventual use. Continued development of
the use of MSS toward the detection and monitoring of acid rain effects is
warranted.
4

-------
RECOMMENDATIONS
The analysis presented in this report yields the following recommendations:
o More sophisticated models need to be developed relating MSS data to a variety
of water quality parameters affected by acid precipitation.
o An atmosphere/sun angle/scan angle correction needs to be developed to reduce
the variance in the spectral data induced externally to the water column up-
welling radiance. This is particularly important for multiple flight line
studies or multi-date analysis.
o Follow-on studies should attempt to conduct water sampling in areas with as
much water quality variability as possible to achive a better description of
the relationships between MSS data and acid rain-related phenomena.
o Follow-on studies should concentrate on seasons which would maximize the
variability of possible indicators of lake acidification.
o If possible, further study should include Thematic Mapper or Thematic Mapper
Simulator data.
o Water sampling should be conducted using a minimum of a three-replicate
system and an error analysis should be provided with the data.
5

-------
METHODOLOGY
DATA ACQUSITION
The multispectral data were acquired using a Daedalus DS-1260 multispectral
scanner flown aboard an Aero Commander 680V aircraft from 0753 to 0919 EDT on Septem-
ber 25, 1981 and from 1253 to 1426 EDT on October 13, 1981. The flight altitude was
approximately 6400 meters (21000 feet) above mean sea level (MSL). The average ter-
rain level below the aircraft was approximately 600 meters (1970 feet) yielding a
nadir ground resolution element of approximately 14.5 by 14.5 meters. More detailed
information about the multispectral scanner may be found in Appendix A.
Additionally, simultaneous 9 by 9-inch aerial transparencies were collected
using Kodak 2448 color film and a Wild RC-8 mapping camera with a 152-mm (6-inch)
focal length lens and a haze-correcting antivignetting filter. This produced stereo
coverage of the study area at a nominal scale of 1:38,000. The film transparencies
and associated film wedge have been archived at the Environmental Monitoring Systems
Laboratory, Las Vegas, for possible densitometric analysis.
All of the imagery from the September 25 mission was seriously degraded due to
atmospheric attenuation. Therefore, only the data from the October 13 mission was
analyzed.
PREPROCESSING OF MULTISPECTRAL SCANNER DATA
MSS channels 1 through 10 (see Appendix A) were decommutated to computer com-
patible tapes and converted to band interleaved-by-pixel format. Simultaneously,
each scan line was corrected for tangential distortion, and scan-line overscan
corrections were made.
6

-------
Further, a correction for illuminant changes was attempted. Samples of 25
picture elements were extracted from the digital data in water areas from adjacent
flight lines. The ratios between the radiances for the adjacent pairs were averaged,
and a multiplicative normalization factor was applied in each spectral band to cor-
rect flight lines two and four to line three, and line one to the corrected line two.
However, scan angle/sun angle/atmospheric interactions proved to be too complex to be
modeled in this way, and this technique was abandoned during the final analysis.
DATA ANALYSIS
The general data analysis approach used in this study was to relate the multi-
spectral scanner data to a variety of water quality parameters, one at a time, with
a multiple linear regression technique. The multispectral data were treated as
independent variables in a step-wise best subsets regression routine.
Lake Sampling
Water sampling in approximately 40 lakes was accomplished by Paul Smith's
College, Ecology and Environmental Technology Division, and Mr. Frank Vertucci of
Cornell University during the period from September 25 through September 28, 1981.
Selected lakes were again sampled on October 17, 1981 to confirm no major changes in
lake chemistry had occurred between the first and second overflights.
In situ measurements were made of temperature and Secchi depth. Alkalinity, pH,
total suspended solids, chlorophyll a, and phaeophytpn were measured using standard
laboratory techniques by Paul Smith's College. Dissolved organic carbon measurements
were made by the EPA laboratory in Corvallis, Oregon. Usually, three locations were
sampled within each lake, and two replicates were taken at each sample location.
Appendix B contains the results of the laboratory analysis for the sites used by the
regression analysis. Values are averages of the replicates.
Additionally, lake radiance measurements were collected simultaneously with
the second scanner overflight. However, these measurements did not correspond to
7

-------
the scanner radiance measurements and were not used in the subsequent analysis.
Multispectral Sampling and Transformations
The ground survey crew marked the location of each sampling station with a code
number on USGS 1:62,500 scale topographic quadrangles. Channel 10 (infrared) of the
multispectral scanner imagery was displayed on a COMTAL image display system and the
location of each sample station was visually determined by reference to the display
screen and the appropriate map. A cursor was placed over the sample location on the
video screen and a 5 X 5 pixel matrix of data values was extracted from the 10
channel data set. This procedure was repeated for both the normalized and
uncorrected data sets.
Mean values for each channel of the 25-pixel MSS data matrices were calculated
and stored in a computer file on a Varian V-75 computer. Thus, each water sampling
station had a corresponding matrix of 10 mean vectors representing the multispectral
scanner data.
The mean vectors for each channel were then transformed to radiances using
equation 1:
R = W (C - GB)	(1)
G (A - B)
radiance (y watt/cm2 - nm - str) for a channel
mean vector for a channel
scanner gain setting
scanner calibration values provided by
empirical measurements conducted by NASA at
a scanner gain setting of 1
These computed radiances formed the basis for all subsequent processing and analyses.
Where: R =
C =
G =
A, W, and B =
8

-------
Best Subsets Multiple Linear Regression
The multispectral radiances and the water sampling data were combined into a
single file (Appendix B) for use by the Biomedical Computer Programs for statistical
analysis (Dixon, 1977). Program P9R (Best Subsets Multiple Linear Regression) was
used to establish the relationship between the MSS data and each of the water quality
parameters. Hallow's Cp was selected as the best subset criterion (as suggested by
Whitlock, 1977). The decision rule for the regressions was to select MSS channels as
independent variables such that the value of Cp was minimized.
It has been suggested in the literature that ratios of multispectral data may
improve the relationship between the MSS data and concurrent water sample data
(Alfoldi and Munday, 1978). Therefore, several ratio transformations of the multi-
spectral radiances were included in the analysis.
Chromaticity ratios were calculated for each channel. The chromaticity ratio
follows the form:
Chi/EChi	(2)
Where: Chj^ = the radiance for a particular channel (^)
EChi = Chx + Ch2 + Ch3 + . . ,Ch10
Physically, these ratios represent the percent reflectance for a channel relative to
the total reflectance sensed by all channels. The chromaticity ratios for all chan-
nels were included on the regression analysis. Additionally, the albedo (ECh^) was
included as an additional variable. The only exception was that Ch3 had to be
dropped from the independent variable list, as it was a linear combination of other
variables, producing a singular covariance matrix.
Also, adjacent-channel ratios of the form:
Chi/Ch(i + 1}	(3)
were calculated and used on independent variables. The ratio Ct^/Ch^o was also
included in this group. It was hypothesized that some of the MSS radiance
9

-------
measurement error (due to atmospheric scattering) could be reduced in this manner.
The radiance data, chromaticity data, and band ratio data were analyzed as three
distinct groups of independent variables. Both the normalized and untransformed data
sets were treated in this manner.
Multispectral Classification
The regression equation obtained for the "best subset" describing pH was then
applied to each pixel of one of the multispectral data tapes (approximately 2,000
scan lines by 742 elements). This transformation is described by equation 4:
pH = - 1.195010-1! + 9.10103Ch2 - 11. 8644Ch3 + 8.8001.8Ch5	(4)
- 8.94497Ch9 + 4.91758
Where: pH = estimated pH for a pixel
Chi = the radiance value for channels i = 1,2,3,5, and 9
The calculated pH for each pixel was then multiplied by 10 and rounded to the nearest
8-bit integer (0-255). Thus, a single channel of pH values was synthesized from the
five multispectral channels suggested by the best subsets regression.
Channel 10 (infrared) of the corresponding data was copied along with the
synthesized (transformed) pH channel to form a two-channel file. This permitted the
classifier to operate separately on water and non-water pixels, as low values in
channel 10 uniquely delineate water. It also permitted a useful grey level repre-
sentation of the land pixels as the infrared channel is sensitive to forest cover
differences.
A parallelepiped (box) classification was then performed on the two-channel data
set, catagorizing the pH values of water pixels into 0.5 pH increments and the land
pixels into five levels of brightness based on infrared reflectance.
10

-------
RESULTS
REGRESSION ANALYSIS
The results of the best subset regression analyses are presented in Table 1.
The input data were radiances (not normalized), chromaticity ratios of radiances, and
adjacent band ratios of radiances. Numbers 1-10 indicate MSS channels used with the
following exceptions:
1)	For chromaticity ratios, I is the sum of the radiances of all channels.
2)	For adjacent band ratios, 10 indicates channel 5 radiance/channel 10
radiance.
R2 is the coefficient of determination, and S is the standard error of the esti-
mate. The order of predictors indicates the relative order of each independent
variable using only one predictor. Although this indicates the relative importance
of variables as predictors, it should be noted that no single channel was a good
predictor. The measured range and measured mean refer to the 102 water sample
measurements of the dependent variables.
11

-------
TABLE 1. BEST SUBSET MULTIPLE LINEAR REGRESSION SUMMARY


Secchi
Depth
(m)
PH
Alkalinity
(p eg/1)
Total
Suspended
Solids
(ppm)
Chlorophyll a
(mg/m 3)
Phaeophyton
(mg/in3)
Dissolved
Organic
Carbon
(ppm)
R
R ^
0.796
0.383
0.309
0.494
0.424
0.502
0.807
A
D
S
1.436
0.593
39.738
0.935
1.649
0.820
0.824
I
A
N
Channels
used
2,4,5,
6,10
1,2,3,
5,9
1,2,3,5,
6,9
2,7
2,7,10
1,5,7
1,2,4,
7,9
C
E
Order of
predictors
7,6,8,5,
9,10,4,
2,1,3
9,10,6,
5,7,4,8,
2,3,1
10,9,6,5,
7,8,2,4,
3,1
7,8,6,9,
5,10,4,1,
3,2
7,6,8,5,
9,10,1,
4,3,2
7,8,6,9,
10,5,1,
4,3,2
7,6,5,2,
8,3,4,
10,1,9
C








H
R

0.801
0.390
0.321
0.510
0.409
0.504
0.793
0
M
S
1.434
0.593
39.404
0.920
1.661
0.814
0.854
A
T
I
Channels
used
2,1,2,4,
5,6,8
1,2,5,7,
8,10
1,2,5,6,
8,10
2,7
2,7
2,7
2,1,2,
7,9
C
I
T
Order of
predictors
7,6,8,2,
5,1,4,9,
2,10
9,10,5,
6,8,7,2,
4, 2,1
10,9,5,2,
6,8,7,1,
2, 4,
7,6,8,1,
5,9,10,
2,2,4
7,6,8,1,
5,9,2,10,
2,4,
7,6,8,1,
9,5,10,
2,2,4
7,6,4,I,
1,2,10,
8,9,5
Y









R*
0.786
0.394
0.306
0.574
0.393
0.448
0.788
R
S
1.489
0.585
38.432
0.884
3.001
0.877
0.896
A
T
I
Channels
used
3,4,5
1,2,4,5,
7,8
1,2,4,
5,8
2,4,5,6,
9,1 0
2,4,6,8
1,4,5,6
9,10
1,2,3,4,
5,6,7,8,
0
Order of
predictors
4,5,6,3,
1,2,8,7,
1 0,9
7,8,4,
10,9,3,
2,1,5,6
8,4,7,2,
10,9,1,
5,6,3
6,4,5,1,
2,9,10,
3,7,8
6,4,5,1,
3,2,8,
10,7,9
6,5,4,1,
2,8,7,3,
10,9
5.6.4.7,
2.3.1.8,
10,9










Measured
Range
1.5-16.0
4.35-7.73
-39.000-
170.650
0.02-6.29
0-10.41
0-5.39
0-7.0

Measured
Mean
5.0
5.74
27.397
1.667
1.867
1.076
3.314
12

-------
Figures 1-7 are plots of the observed (measured) values for the dependent
variables on the X-axes and the values predicted by the radiance regression models on
the Y-axes. The straight line drawn on each plot indicates the line of perfect
correlation. Points falling above this line indicate overestimation by the model,
and points falling below the line indicate underestimation.
13

-------
o
o
SECCHI
(M)
o »
H
Rt 0.796
CH»2,4,5,6,10
S=1.436
• •
o o a
4 DO
1 & 0 0
8 0 0
16.00
SO . 0 0
a 4
OBSERVED
Figure 1. Predicted vs. observed Secchi depth (m).
14

-------
PH
••
• •
R« 0.383
CH° 1,2,3,5,9
S= 0.593
3 20
4 0 0
9 0 0
4 80
5 6 0
6 4 0
? 20
OBSERVED
Figure 2. Predicted vs. observed pH.
15

-------
©
o
ALKALINITY
(JJEQ/L)

a
UJ
i-
o
H
~
Li]
Q£
CL
Ri 0.309
CH'1,2,3,5,6,9
S=39.738
- 40 0 0
0 0 0
40 0 0
80 0 0
20 0 0 0
16 0.00
OBSERVED
Figure 3. Predicted vs. observed alkalinity ( p eq/1).
16

-------
TOTRL SUSPENDED SOLIDS
•	(PPM)
R^=0.494
CH=2,7
S=0.935
ooo	a.o o
~T	I	l
4.00	600	800
OBSERVED
T
"I
10 0 0	12 00
Figure 4. Predicted vs. observed total suspended solids (ppm).
17

-------
CHLOROPHYLL fi,
(MG/M3)
Rl. 0.424
CH=2,7,10
S=1.649
i	1	1	
000	200	400	6 0 0
OBSERVED
8 0 0
I
10 0
12.00
Figure 5. Predicted vs. observed chlorophyll a (mg/ra3).
18

-------
PHflEOPHYTON
(MG/M3)

RoO.502
CH=1,5,7
S=0.820
0 00	100	2.00	3 00
0BSERVE0
4.08
I
5.00
6 0 0
Figure 6. Predicted vs. observed phaeophyton (mg/m^).
19

-------
DOC
(PPM)
4 0 0
8 0 0
10 0 0
OBSERVED
Figure 7. Predicted vs. observed dissolved organic carbon (ppm).
20

-------
The spatial form of the model is depicted by Figure 8. The results of the
multispectral classification are presented for pH in the area between Fourth and Big
Moose Lakes. Colors represent modeled pH levels in the lakes and streams, and
black-and-white tones represent various forest cover classes.
LAND
ALKALINE
NEUTRAL
a#*

ACID
Figure 8. Multispectral Classification of Lake pH, Adirondack State Park, New York.
21

-------
DISCUSSION
Multispectral Relationships
Review of Table 1 indicates that only Secchi depth and dissolved organic carbon
provide reasonably good coefficients of determination (R^) values. Alkalinity and pH
provide the worst R^ values. However, weak, positive relationships are present even
for these variables, and the standard error of the estimate (S) for each of the
dependent variables seems to be of an acceptable magnitude.
There seems to be no distinct advantage to the use of ratios rather than rad-
iances directly. Although, slight increases in R^ were noted in a few cases, no
consistant pattern developed to justify the increased computational difficulty.
Although no single MSS channel was a good predictor of any of the water quality
parameters, several MSS channels were consistantly more useful. It was initially
hypothesized that channel 1 (0.38-0.42 um) would be a valuable predictor of at least
dissolved organic carbon. Taken singly, this was not the case. However, either
channel 1 or adjacent channel 2 was included in every best subset group.
The red channels (6 and 7, 0.60-0.65 and 0.65-0.70 ym) , representing chloro-
phyll absorption bands, proved very useful for all but pH and alkalinity, while the
infrared channels (9 and 10, 0.80-89 iim and 0.92-1.10 pm) were most influential in
discriminating levels for the latter. This may provide a clue to the process of
relating MSS data to acid rain effects. It may be that the chlorophyta (which are
detected by channels 2, 7 and 10) are not the key to remote sensing of lake acidifi-
cation effects. Nor are we likely to find a surrogate in the phaeophyta, as the
chlorophyll absorption band is an important predictor as well. However, it is likely
that some plant response, or combination of responses is responsible for the relative
importance of the infrared channels in the best subsets regression. Infrared energy
is absorbed within the top few centimeters of the water column. Therefore, some
near-surface phenomenon must combine with phenomena throughout the water column to
produce the relationship between MSS-detected spectral response and lake acidity.
Measurements of chlorophyll a, b, and c will be made during the spring conditions
22

-------
follow-on to this study and may provide further clues to the detectable phenomena
related to lake acidity.
It is also encouraging that the Thematic Mapper aboard the new Landsat D and D'
series satellites have MSS channels similar to the most useful channels of the air-
craft MSS. Channels 3 and 4 of the Thematic Mapper describe energy within the 0.63-
0.69 ym and 0.77-0.90 wavebands, respectively. These channels are approximated by
the aircraft MSS channels 7 and 9 in this study. With its 30-meter spatial resolu-
tion, the Thematic Mapper should prove to be a very useful water quality assessment
tool.
Scatter Plots
The scatter plots, represented by Figures 1-7, are visual representations of the
regression results. Generally, they confirm the relatively poor values for the
pH, alkalinity, chlorophyll a, total suspended solids, and phaeophytjkn regressions.
They also indicate that the chlorophyll a, phaeophytjfln, and alkalinity regression
values may be improved by some transformation (perhaps In) of the dependent variables.
Deviations in the plotted data from the correlation line represent errors in the
regression-based predictive model. These errors may be induced by mensuration errors
in radiometry as well as water chemistry. During at least one previous study, the
model estimated error was within standard laboratory measurement limits for N-NO3,
total phosphorous, total suspended solids, and conductivity {Mace, 1982). This
probably was due to relatively wide ranges in the water sample data. The present
study, however, dealt with relatively low concentrations of chlorophyll a, suspended
solids, organic carbon, and phaeophyttfn. Multispectral analysis techniques may not
be sensitive to extremely low levels of some materials in water. Even the laboratory
measurements revealed some sample replicates differing from each other by a factor of
10. Since only two replicates were taken, spurious values could not be eliminated.
Also, no quality assurance analyses were provided for the laboratory data. Future
studies should attempt to reduce the variance due to water chemistry measurement
differences by at least a three-replicate procedure, and a quality assurance program
should be instituted.
Variances induced by errors in radiometry are most probably caused by a sun
23

-------
angle/scan angle/atmospheric interaction. Corrections for these effects are not
straightforward, and more research needs to be performed before reliable corrections
can be made to the MSS data.
Multispectral Classification
The multispectral classification enclosed in this report represents a small por-
tion of the imaged area. Moreover, the relationship between lake spectral response
and pH was not strong. However, Figure 8 is included in this report as an example,
in image form, of the eventual application of this technology. Clearly, the model
must be improved before an operational monitoring technique is developed. The image
does, however, illustrate general differences in acidity between lakes and may serve
as a qualitative tool. For example, the mix of yellow and green mapping levels in
Fourth Lake (lower right) favor more neutral conditions, while the mix of colors in
Big Moose Lake (upper right) favor more acidic conditions. The speckled nature of
the classification reflects the statistical observation that the variance in the
estimate (Figure 2) approaches the range of observed values. Nevertheless, a weak,
positive relationship is expressed in the image as well as the statistical data.
General Considerations
Considering the magnitude of the problem, and the benefits to be gained by the
development of an operational monitoring technique, the monitoring of lake acidifi-
cation by remote means provides sufficient promise of success to be worth additional
study. Overflight timing may be the critical element. Levels of the various water
quality parameters examined in this study were relatively low during the fall sam-
pling for this study. A spring follow-on to this study will be undertaken, but some
of the concentrations of the surrogates to lake acidity may again be low. The exper-
iment should be repeated at least once more - during mid-late summer. This would
permit the temporal variable to be examined, as well as full development of plant
responses to the lake pH values. The cost of such a continuation of this study is
relatively low compared to the cost of developing an extensive water-sample-based
system. If appropriate models and timing can be found, tremendous savings can be
made in the establishment of an operational monitoring program.
24

-------
REFERENCES
Alfoldi, T. T., and J. C. Munday. 1978. Water Quality Analysis by Digital
Chromaticity Mapping of Landsat Data. Canadian Journal of Remote Sensing.
4:108-126.
Baalsrud, K., T. S. Traaen, M. Laake, and G. Raddum. 1976. Acid Precipitation:
Some Hydrobiological Changes. AMBIO. 5(5-6):224-227.
Dixon, W. J. (ed) 1977. BMDP-77: Biomedical Computer Programs - P Series.
University of California Press, Berkeley, California.
Dochinger, L. S., and T. A. Seliga. (eds). 1976. Acid Precipitation and the Forest
Ecosystem. Unpublished workshop report. USDA Forest Service, Ohio State
University, and the National Science Foundation. 38 pp.
Galloway, J. N., G. E. Likens, and E. S. Edgerton. 1976. Acid Precipitation in the
Northeastern U.S.: pH and Acidity. Science. Vol 194:722-724.
Likens, G. E. 1976. Acid Precipitation. C & EN Special Report. :29-44.
Mace, T. H. 1982. Characterization of Lake Water Quality Parameters With Airborne
Multispectral Scanner Data: Flathead Lake, Montana. Proceedings of the 48th
Annual Meeting of the American Society of Photogrammetry. ASP, Falls Church,
VA. :375-387.
Whitlock, C. H. 1977. Fundamental Analysis of the Linear Multiple Regression
Technique for Quantification of Water Quality Parameters From Remote Sensing
Data. PhD Dissertation, Old Dominion University. University Microfilms
International, Ann Arbor, MI. 176 pp.
Witzig, A. S. and C. A. Whitehurst. 1981. Literature Review of the Current Use and
Technology of MSS Digital Data for Lake Trophic Classification. Technical
Papers of the American Society of Photogrammetry: ASP-ACSM Fall Technical
Meeting. ASP, Falls Church, VA. :l-20.
25

-------
APPENDIX A
MULTISPECTRAL SCANNER
AND
ANALYSIS SYSTEM CHARACTERISTICS
26

-------
MSS WAVELENGTH BANDS
Channel
Wavelength
Band
Color/Spectrum
1
0.38-0.42um
Near Ultraviolet
2
0.42-0.45um
Blue
3
0.45-0.50pm
Blue
4
0. 50-0.55um
Green
5
0.55-0.60ym
Green
6
0. 60-0.65pm
Red
7
0. 65-0.70ym
Red
8
0.70-0.79ym
Near Infrared
9
0.80-0.89pm
Near Infrared
10
0.92-1.lOym
Near Infrared
11
8.00-14.00pm
Thermal Infrared
27

-------
MULTISPECTRAL SCANNER
IMAGING CHARACTERISTICS (SIMPLIFIED)
nnu (UD OF WW
J DIRECTION OF SCAN
GROUND
RESOLUTION
ELEMENT
SCAN UNE
28

-------
MSS DATA PROCESSING (SIMPLIFIED)
HARD COPY IMAGE
•	COLOR / B & W FILM
•	ELECTROSTATIC PLOT
TV DISPLAY
•	COLOR I B & W IMAGE
•	STATISTICS
DATA
PRE-PROCESSING
MSS
SENSOR
TAPE
LINE PRINTER
• STATISTICS
TV DISPLAY
•	IMAGE
•	STATISTICS
HARD COPY
•	B 4 W FILM
•	ELECTROSTATIC PLOTS
•	STATISTICS
IMAGE ANALYSIS
•	IMAGE ENHANCEMENT
•	CLASSIFICATION
POST ANALYSIS
PROCESSING
•	COLOR ASSIGNMENT
•	ANNOTATION OVERLAYS
•	RECTIFY TO MAP BASE
29

-------
APPENDIX B
MULTISPECTRAL RADIANCES
AND
WATER SAMPLE DATA
30

-------
APPENDIX B
MULTISPECTRAL RADIANCES AND WATER SAMPLE DATA
IiEmIi 1
2. ; v.
2. i 1
.' 00 2
1 , '1 ' 1
1 . Off
0 - '41'
<¦. r-7 l
r.
5 1,:
0. 1 2'
. '. ,14,
4.1"
1. i9
-.r'.ix
J , '.4
1 i.',r
I . 0 1
"5.0 1
1 > b. AI \
. 5 70
.'.oi :
1 .fiA.'
t . ¦! 4 '
0, v.;
0 . <-'0 '
A . *)'•'/
a
^1."?
¦M.'.
. \j 4 V
*!
•J. i
- <9,0<
1 -17
I - 1 A(
0. ft 1.
5.0 1
Hf.-.h' <
'J. -1VI
1 . 97
1 .HI 7
1 , 1 "
•> ^,T.
O.'Jti,
' - 4 H ,
0
. <0"?
0 12.
a . A'i:
1
<»~
!'.'. 61
1 . aa
1 , iiik
0 ft<: 0
3.0 1
III-AV 1
'.5 In
r.'tr.
2 1 ("i.
i .;
1 .
t . 'M'
a.,SO A
0

A , 12
A. 0 4 -


22 .At;
4, n
1 ."fti
"'.5 '0
1 - 15.'
"'.0 1
Hl'AO"
I' - J',' '
>."5ft T
". >12
1 ./
[ ..)V
1 , A »/
0 ,' ¦ ? ^
A
. 5 .1
0.1 ¦<
A - 0 1 \J>

r., V/j

1 . !]f.
'.4hiO
'.0 1
KE.W :
" < "04
'.442
r • 26v
L . 1
I.I,"
1 ,*0'."
0.
tl
. 9/
• •, V r S
0. ')o'
. k
. 'L~ t
".Of
0 . " 1 1.
, ¦,
\
. 1 0
0, v
A, A-.A
a , "
A I
i.hij'rji:
3 ,JF!
_>, ,'V 1
I , -, "

A."''1

0 . 1.


.A' 2
' 1 . 1 4 t'l
. > ~,y>
' •. ^;
V V ~
' > (. *' .
• 1 , A » A
0 . ' ..50
- 1
l.tfllS'-ll'
"5. OW

:• 'H.
1 1
1 , ,,
A /> ¦ ,
'i.'.1 *

•'¦3
" AKI
• ' , A 11

! > ~ 0 5

(>,SA
0 . 1 ,-.0
A . 1
1
i \ r 11 b i
: , V7
1 -"-'1
i . ;'< i ¦
1 !
f. , / 1 A
" .
A .)o!
.>

. i . i ,
" . (¦ 'C
,' ¦

I A, ()4
0. ",
A , \,),i
A. tftA
5 . A I
i.un>2
, .>!'.
I. y 7 <
i r; i ¦
1 .
r, <_¦()/
O.i' ^ 1
A . 1.) ,
t\
:1
¦ • 1
0 • • 'i A

. '..A
A> . fl 1
1 , 2A
ft , ,'A(
0.11 0
5,0 1
Mil h 5
' . i
1 .
1 -V/
1 .
0 1
A, !i
a 4 > 1
n

U.A'- 1
a . 0-1A
¦>T * ' 1
' 1 *
H .
1 , 2r.
A , ' .ft A
0 . -i '0
. A 1
non i
¦. i ii
'. Ol.i?
I . •/
¦ •, 1
1 ,
1 . •' a 1
A . • ,9-'
(1
. W,-i
A
A A,< 1
1
^ . 1 A
'•0. (<2
1 ,-M
n . ,'.l
4 .5 5 A
6.0 I
i.OG >
2 • 090
1 >'(>/
\ .
I
O.VO',
A
A a r's
. >
'",".4
' ¦. 0 / J
A . A ~ \.
1
(i, t »\»
<
' .2'->
7.5 10
4 . 5/0
6.0 1
i.inn
1 17
1 -iii;
i, ' i
i, '
0 - <
A , 9.1A
' » * ' i. '"
i;
'.'"ii
0 , ASA
" , ,> |
f .1.
/> *
1 0
j , '1
10.41"
".ft',0
5 . ''' 1
nr-Acm i
!ft
2 ¦' " >
1 /Vi
1.4 !
t., y., /

0. » A


... ii'.
" . 01;,
1 :.o

A . 4
0 .20
0 . '"60
0. . 1 ftA
1 . 0 I
IIRANII"
. 702
2. i:'0
I .'-'-i/
i.
0 . r'KV

0. >.9|
A
1
O.O-V.
A.m.'2

1 i
'j. 4:"
a. on
a. 2ft"
0 . 1 4 0
1 0 1
1 . 0 1
I'fcANU.l
2. 5«>,
,' - Oil,',
l.I
J . 
. :;n
0.02
0.660
0. 140
CANAIU
1 , H71
i
I -
i . * i-j
0 > " 1 2
0 . H I 6
0. 12?
¦ )
20 1
0. Ofll
0.0
i « *

i !. :i40
I , A 1
I AiVA in
1 1 1
I
l. ;
1 tv 1
n . |
(¦ >: >; • v
A. 12,'
1)
2L. 4
0..jS:3
A . A 4

. 7H
1 1 . 10
I . 7.5
I . 7H0
0. ft',.0
1 .A 1
i A "¦"> T t
2..5'Y>
1 , -IV)
i .
1 , < 0
A ?0f?
0.1-I49
0. V. '
0
r (*>'.'
A . 0:! 1
a , 0 ¦'1 '
'j. 0
/). .04
4*. '1,'t
0 .
40
A . V 'O
-.0 1
CAr.r,2
2.191
1 .85v
1 . /4'".
1. ? •'
o, -?o;i
0. H
0. I',1:
i >
4
0,0° 1
','.04/

<'.. 9
.'.r,, r;
o.nb
0 . ft 20
A . 7 'A
3.0 1
t Ar;i':i
2 . 17 1
1 ..i46
I . >'•
1 . )
0 , 1 4
o.,;m
0. 17"
x'\ 4
<10
0. u
U.A.,4
1 > *n
A.
4 lj . A9
0 . 90
I , 2u0
0.0 "50
3.0 1
1,1 K 1 «
1 .11-11'
1 .?
0.
2'j 4
•' - K'fi
1 ' . ' '4c>
l'\S
",.11
l ft. AH
0. VO
0 . 5R0
0 > 000
0,0 1
ci. f.::h
I .("I/O
i
i
1 , i i,-.
O.liv H
0
A, 1''
0
S 4
¦ > . Of| £
0 . 04 i
( M . S
1 1
I '.ill
0 ¦ 0 7
0. 4ft0
" .'.'AO
A , 0 1
i nw i
5
1 , 9 '5
i. i
1 . 4
y. RH -1
A51
0. 14 1
•),
22 1
0."4 1
A ..AO

10
111,2ft
0. 10
•). 750
0. 1 "A
5 . 0 I
1 kh'^2
^ » '"I '
2, 029
l
I >'¦ .
0. >'A!i
A,(I,'.
0. -1
1)
2 -17
L. ,nj 1
•J AA11
^
(5.2 9
JH . Of)
1 .00
I . 0 50
0. 130
5.0 I
1 FAN 5
2.205
1' . 11"!
1 .Vfl'l
i .-r;.i

A.KH/
A, 1 ^
<>
.'j 1
0 . 1
¦ >.000

A . j^,
.Ul. 41
0.4.?
1 .050
0 . .500
5.0 I
HAK 1 1 I;
-'9ft
'.:' 11
.',o ¦
i
1*01 0
. ?4 '
0 . 'Vi;-!
0.
_w:j
. 12"
" . A 4 6

">. IV
'./6
1 ,42
0.4 40
0. <50
3.0 1
l'Aft l"'i*

..in
1 . "-'4
i ¦
A v>y,)
0.91 S
A,CI,'
0.
. A,'
0 . 00
.: 11
I . 2\.
A. 4,60
0. 1 AO
3.0 1
nAlv 1 j4
0-1/2
0 ,

0 .10 4
U , A 44
4.5
ii.1
1
1 .20
0.400
0. '(90
3.0 I
rmii' i
2 - ,'»9
1 . 9 1
1 .,-<4?
i . i
0. '-1 -i; 1
0. B. 10
0.4
.1.
:.;-.4
0 .,
0 ."11
5.0
7 .17
n.\ 14
t .45
1 . 920
0.4 no
3.0 I
runrv.'
>.69'/
\u9
1 . V'JO
1 (
O.'-fi!)
a . t:9'j
0 . 1 "A
/1 ,
2M
0 - 011
A A(j)
S.(»
7. 1 "t
1 5S.00
I .75
1 . 700
0.580
3.0 I
1 IIIJK 5
2. 1 72
1 ,l>39
1 • / I' i
j. <
0 ' Oi,
0^12?
A.-->31
0.
24 2
A .041
0.000
•;. 0
7, 2J
I70.fi-3
1 .110
1 .940
0.350
3.0 1
HCNN 1 I.

'.'.07,'.
1 . V,J<6
1 . 1<"-'0

0.D20
0.4"! 7
A .
2'.4
0-114
A. 0-i A
16»n
4.91
- 1. 7ft
0.5/
0 . "4A
0.000
0.0 I
H0NN2
2 . HUD
.' 'J-Vi
2 ,I,M
l . 'if
0.91."!
0 .! 12 y
O.-I^l
0 .

0.. Ill
0 - A 4.1
1A.0
4.92
-7.4ft
0. 4ll
0. 0!lO
0. ftf!0
0.0 1
Hur'rt 5
H 1 1" l~
* ..«»«•
. 094
1 . ¦¦>(»
1 ,4.', 1
0 H7,'.
0. v,>
0 . 1 c s
11 (
'44

A *016
1.A.0
4.915
ft. JO
0. 4 h'
O.060
0. 190
0.0 1
(NUt
.'. 2 >tf
2. 16a
.'(•
1 ,'¦"/>
1 . .'00
0. v-; '
A.'.","
0 ¦
I (1
A. 122
A . A"' 1
' r
: j. 1 /
I' >. L 7
2. llj
A.ftOO
O.ftOO
A.O 1
CNfiC
2 . "4 5
2.2.5 <
2.071
1 .' .S,!
1 . M
a , ',y
A.'-.SS
0.
"1111
a , 1
A.O1'

5. 1"
14 I !
2. 1 A
a. 480
0. a so
6,0 1
[NP (
2. 264
ii r,
2.041
I . ^ 5 1
0.9V/
O./M
A. i;.k
('
12 <
A. 122
A .OS.S
' •
II!
[-1.01
"5.12
0. WO
1 .060
6.0 1
NAN 1 h
i ,ri62
1 . 94 1
1 . fM /
I , -i I <>
0 i u
0. v
0.'-. v.
0.
"5 'S
0 . 1 Iff
A.O'i 1
r«
S.09
-1.92
2.40
7.520
5 . *5 f>o
3.0 1
KAN.'
1 . 927
i . 955
1 . 0.'>4
t. i
1 .'>1 0
1 A 1 v
0..S V7
0 .
S19
1. W-l
0. 2^1..'

5.10
-4.8ft
ft .00
ft. 9 20
4.650
3.0 1
1* AN i
1 . 9 1 5
I . 9/, y
l .87',
1 , 4 4*
0.S\<.r-'
v). -rtU
c>./ i
0.
JS9
1.'. 1 
;. 11
-11.1 <
4.00
ft . 9 90
.5. 9)!0
3.0 I
I.1LA I
2.:54 2
2. >"n
J . 1 4 7
1 .1 <•
I .04"
0 . 9T,
o.r.'jv
0.
"<1S
'J.Ot 1
'.046
_ CJ
t. .,<2
50 , 1 4
1.;
O.SftO
1 . 250
6.0 1
1.1 L A".'
..116
2.0 4'.
1
i , 11 .-I
o.9i :j
o.on:
A. -'!i4
A ,
.4
3.042
•J. "OS
w' • L'
ft. '.Tl
41 . 45
4.50
1 . 1130
0.970
6.0 I
1.11 A 5
2.24-1
2. L95
2.0-14
1 . L. 1 A
0 .
0.9 i7
0.^1 5
0.
'•i4
<} . 06 r-i
0 . 007
2.'i
S. 27
"4, I ft
J.40
1 . OH a
I . 550
6.0 I
1.1 M£h I
2. 1ftH
L . 71 l.t
1 . 0 •<>
1 •".sc.
1 - "A4
0 . 9,'.4
o.&r.i
0-

0. a 1j 3
-.>.04,'.
8.'^
j. 1)0
t 1 .no
0. ?A
1 .ft 00
1 . 4 50
2.0 I
1. 1 ME l\ 2
".'.146
1 .'-'10
L , tW.'
1 .4!.'"'
0 . 9 ,l' i
0 - 0 '9
0.15:
0-
"»L.4
0.04"
A . 0.5 ^
fi.o
CJ . 6H
ft . 41
A . rtO
a. 560
0 . 0 70
2.0 I
MltFh S
1 Oft
1 . 90
1 . f'4 --
1 . C L
2
0. (J~5
A. 444
-),
"1.1
^. a;J 1
>, O-i.i
'""t. ^
7.7-. |
1 I . 44
0 . 1 ft
". '.ftO
0.01 0
-".0 1
31

-------
APPENDIX B
MULTISPECTRAL RADIANCES AND WATER SAMPLE DATA
L rdfl 1
. h •
| ..0 14 I . A-ll

U . V '.
Q Jjl t
0 . 4 V
1 '
0 ¦ 1 .2
(',0-7
1.:,
4.S.-*
-,.•4,
1 .40
0.!!"0
A .A'.O
2 .
LYON.'
2 . S 72
''.OOtJ l.b'42

0 . v"_\>
'.. l-'ti1
0 '"ill I
0. ?1(!
A. 12'
0 . At.'A
4,f;
4. 'j 1
- "VIA
1 ,\'"i
0. 7!)A
0 . '!'20
2 .0
(.TON i
2.5 SO
2.00 ! t.t; 13

0 . f '•<>
0 . ,c' , ^
A,'-AM
0, rtl-i
0.1:2
'J. A2?
4 ¦ ¦<
Jl ', o
- 2*-.. "i
! , '?n
0. noo
0. 4 40
2.0
Mf- Tj
:». sa.2
->.A42 2,."
0 . I""! "J
A.StS
(1. 1 2 '
A.O""t
s. 0
r".. 1 V
''. 2 j
0. A3
0.200
O.J AO
I , 0
ftOSS t
~S . 1 »f!
2. v-,:; 2.1'M
1 .AO'*
1 . 0
A , .y a *.
0 , j'"j
A <10
0 . L)t* 1
<>.0 4 A
4.:.
A.c,3
A 2 12
0. f i
0. ''40
n , ^ ()
4 , A
(1US 3 '
;>, |
'4V 2.02"
1 .1.: y
0 .'-'Cl.
0 . ^ 10
. Sno
Ao2o
A , T.11
O.A4.S
4.'.
0 . LjL.
ri A ,VA
1 ,23
0. S'20
0 ,4''0
4. A
nci':'. i
'i.O'.,.
2 . '' A 1 ".OA-}

1 .02.
0 .' / /
o.1-, _'!!
•> "(I.J
O.OPA
0 . 0 4 A
4 .-.
A . l~4

1 . ,i"i
1 . 1 20
0. 1 SO
4,0
lif GMi 1
",l ^
L .(til 1 . ",A

(>. y2->
0 . LI - ^
0. 4Aft
A. 'vi
A.; •"
0.04 6
!.'!
¦i '-V)
I A , A"i
2.00
. o.'.o
0 . A4A
" .0
Nf Oh (1 '
'.'.'¦I-.
1 . V I r-
1,414
0- / !.f
¦j- sJU'1
('•.
O.'JS/
0 . 1 OA
0 . 04A
^ * Ll

I!. A 7
I . i^O
>.wAA
0. a y 0

NC OKI <
2. -l1- <">
l.VA'J 1 . i! 'i''
1 4'-. i
n.
0, !,:i4
0 *<< '
A .
A. 1 'A
0 . 04V
"!.

in-:,"/
1 . 40
2. f(
02. to
'•. 'i
NOh ri 11
2. aao
] ,in • 1./21:

O.;:!"1!)
(*,
A , 4"-'0
0. '.4
A . A'\S
1). A * f A
2.' 1
. 1 _!
".4A
2 .00
1 . 4AA
L
' 1 t V >
fll 'R I H2
2."
1 .<•-'! i 1 .,14 1

0 V|"
a .w;
0 -*r-
A.JIS
0 . ¦ ^a
0,04,-,

S.I?
4 4r.

1
1 . '00
4.0
nokrn;
2 . 1 12
> . 0 •' ¦ 1," tv

0.
A. 5()V
0 . S 1 1
0. in;
0 - 1 22
0 . 04 7
2 r„,
1 .AI
2 . ' .
A. 2!j
',).) J: J 0
0.7 10
s. 0
(IIIEL 1
.'. 1 " -
1 .iM4 ;. /1 /
1 iSo
O.Rrt •'
A , D".'
A , J'V.
0.2' ¦ 4
0 021
'i,0
¦.. '1
1 ,. Y>
-22
1 . 10
t . ','((>
0 , 4 7 A
2 . •>
nutTi-
1 isf!
l,;mk i./i',
1 . i' .0
0 .(J,-,;,
0.
0 . 4 2"
0.-2,(4
{\ » J
0 - "4A
0.11
4 , U'J
-2 4
0.60
1 , "AO
A ¦'! A
" . 0
(IUEE3H
1 . 1 1 ¦>
1. .(Ml.

".!•!!! 1
0 H21
A , 1 U
0 -: '..4
A. 0"! I
A . AJ4
: -' >
L!, ! 1
.!. A 2
0,
I . iOA
O.S-'O
2'
I.OfJli 1
1
£,/2,f

'' ,cfJrl
0. f;;."-
0 . I ' 1
0.. ',4
a .1
0. 0a6
Z,.'!
LT. AO
1 H. 24
1. >0
A. "40
0 . y'jO
5. A
K0NH2
1.^1 A
1,7! 11 1.74!

•J.., '¦
0 . b' '
0 ,41,.
0 , 2'"' 1
A. A.^! >
(1. A JA
. 0
S . 'j4
¦ i. 14
0 <-•-<
O.S2i.i
0.4 70
1.0
I1UNII3
1 . KiST
1.A4A 1,7"./
J.J'"!
o.vo '
A.
0,4'.,.
A . 21.4
0 . ¦1 f 1
A , A 44
CJ . 0
S . S4
7 . A 4
1.00
0 , 4 A f'
0, !'70
,5.0
:>ai ML
l.'Xtl
1.7 <¦» I.AA2
t .
O.d'.ft
A
0. 1".!
A . 2'". 4
•> • A," J
0 . 0 Kb
£.
5.44
A. 71,
1 - 35
1 . -'A'O
0. i/'O
4. 0
ortLfl :
1 . V77
1.7V"! 1.6,-j 7

0 - U 4 (•)
a . £? .
A .
¦).:'",4
A , A.!i
0 . ¦ '4A
. <
j . 44
I". IA
I . '.'O
I , >">')
0.0/0
:i. 0
SEOl
2 . 2 4 j
1. ,'2 L 2.0 Aft

1 .00 5
i'. t'40
0 . t'OS
0 . 2-
0 . 'jfS 1
0,0. 2
A.J
a t \'i y
1
O.DO
O.ffAO
O.SAO
4 . 0
SEM2
2.Jav
2. I1-"1 2.2 HAUU
\ .Mfl
2-'Ml 2.0-1,'.

1.00,
0.5.'".
¦ i , S t 1
O. <10
". >''¦51
') » 0 -*1
[.
ft ¦ 1 '1
4 !. 0 I
,i. 'S
/ , " 1 A
2 - S'A'i
7.0
>HmI II
2. W
' * i 7. 1 i 1
1.
1 ,o',
1.. 0 ¦)
a . S72
A "(IS
0 . Arf,-
0 . 04,1
i. '7
A . 1II
17,44
2. A 7
! > , A | A
2.'' JA
7.0
¦THAI 111
: .-in
2,4: 0 ,M1
1, '
t 1
t . Of L
0 - '> "7
0 *:
7.0
:HXf2 <
2. ",'n)
i.'-as 1, h r:.

0. «-)¦;
O.fJ.M
, to j-
a . i; 17
'"'.11 4
0.04/
c.,0
A . A 9
1 lA.A.'
i,r".
A. A AO
o.yao
3.0
J 1 *2
2.0,-jO
I,*'.', j l.iilTj

Or, 4""i
• J. sVifl
0. 4""
A . ,1 trl
'.'.11 ->
0 ,04.
,N .
6 . > 4
148,A«
0, i",
2.4AA
0.4 AO
1.0
X i
2. oao
1.^/2, i .fit1-"

¦). y 4 •.

"I
A ,Wf
A ' 1 4
A ,".1 7
A 0
6, "I
14?.AO
0 ,HS
•). ' 1A
0 .020
,5. 0
SlHJTfl [
1. -><> f
1

0. 1 -
•1, n
A, .1 'a
A. 2'v)
A,.»8t
li AIV
3-0
u...
4.
0 . A 4
A .
0. 7OA
0.0
\.:n!jni2
\ 0
1,
1 . 1 1.1
0. ,iJ,:
A ,H 14
A , -1 .t
A . 2S4
0. "S1
A
S'-O

1 . 44
0.1)0
0. 4!jO
0. 400
1 .0
¦ion rti i
2 . 1
1 , >•'>'/ 1 , >iO 1

. W'li
0 . ^ 'V
0. 1 !l
a . •', 1
0 "b i
a, .;.d,5
3. A
If!
-2. "7
0 * \ > 4
U '.! i 0
0 . SSO
1
1
2 -IV
2 1 4'".(S ,J * 4.Ni
1 . '¦>
t. y"i
1 . 1 .i~
0 <'¦>
A. 1
0 . 1 22
0 . A4 A

A.'. 1
2 1 . 44
4 . ,"J0
4 . A1 0
2 .510
! + "
SC'{IAUI2
2 .4 44
. 4HO 2. <"4
! ^'.2
i • - < <
1 1 /A
0 • A 4 ft
0..501
0.12 ¦;
0.0 1A
2.1 ¦
A . . 0
li).HS
4.
'j. 1 2')
2. 100
.1. 0
SHUALTi

'.'".1^ 2.L.0 1

1 . SAC.
L . 22'-.
A . Ail •
¦¦. !,¦.!
0 . 1 ',2
A , 0.12
2,: •
A. L'-'
1''. 0 ¦
4 » 2 L.
4 .' j A 0
2,020
¦'
JT Lf. LI
. .',*'11
2,117 t .^'72

O.^H4
0 ¦'12
0. _• 1
A • .">•
A . • I |
a .)L,<
, ,
,.. jO
" 7 . A '
. "'•)
".AAA
1 ,430
A . 'J
STILL 1
1 . /'/M
1..V..9 t,'iV2

0 .:->!)
0.-. y
ti, 4'„.;;
a.; '. t
" .1^4
0 , U 46
2 .1 ¦
O ; 4'*
. a .'
1 . ! 'A
1
1 . j to
s, 0
ST U L J
1 nljV
2. CO 1,W.'
1 . J 7 i
0 . ri),'.

0 SO'
'.) . ."¦! 1
A. -11
0 . 14
j j j
A . 1 2
*j'.
1 . b<>
t , .'10
I . VIA
s.o
l"A IA
1 . -• • "
1 . v 2 I

0 . 'J .
0 . HU4
0.4'-2
A, ilj
0 1 .2
0, A v.
.1 ¦
£ . '.A
- !,').7,'
1 '
4 , ?AA
4 , S 1 ()
2.0
ra in:
I . v. '
L.vl.' L.H4'.

0 . 2 '
O . ,t2-j
0 , 1 ¦
" . 1
¦ •. 1
' 1 , • '4i'
. '!
4 c, 1
?7. AO
! . "I i
4 . 2i)0
2.410
2.0
IA JA
I .'' 2
l.'.M,' J C4'.

it - -

(1, IV 1
A. 5t !
A. 122
0 0 4 0
i , '"j
i.'.4
-A'j , 1 A
2.1.0
i.2.10
2.210
2,0
fHAYI
2.S'-'4
_>.0. V l.n:?

0 - '-\j )
^ . st)n
a _¦(>

»rj
0 '/-'A
2.
A , 78
74 . <10
1
2.iviO
0 . 700
A , A
THA f,'
: sn;
1. . v -f,

0 . v,V7
A . V /
0 - s 11

A-OSl
0 , >j4o
2 .
A . !<4
>:!. y '
1 , - A
4 . 1 0"
0 . / 1 0
A. A
m.ns
2.6'.-!
.".o,1.n'Ai


a, 744
0. '520
0 . 2A4
•>. V.J 1
0. j-lc<
" 2 . ,i
A .SO
101.74
I .no
4'70
J.270
A.O
rM 1 r l
2, I U5
I .8:17 1. 762


0 .
. 4 <
0
• 1, 0 ' *
0 ,04A
4 . 0
¦;. 17
J. -'2
t ,-,5
O.'.JJO
0 . V1 A
, A
run :
2. 137
L.H'V/ |. 757,
1 , t6
0 ,V" 7
A. ;, :
A . 4 r "!
•) 2M
0 . O'? 1
0 . A4A
. L1
. 1 /
0 . _v.
1 .41!
0 . A2A
l.'VO
S.O
rui 1 7
2.o:\{
l.M-M
1 >0
¦ ¦
' ' r,
-------