U.S. ENVIRONMENTAL PROTECTION AGENCY
NATIONAL EUTROPHICATION SURVEY
WORKING PAPER SERIES
TESTS OF THE
DUAL DIFFERENTIAL RADIOMETER
WORKING PAPER No, 473
PACIFIC NORTHWEST ENVIRONMENTAL RESEARCH LABORATORY
An Associate Laboratory of the
NATIONAL ENVIRONMENTAL RESEARCH CENTER - CORVALLIS, OREGON
and
NATIONAL ENVIRONMENTAL RESEARCH CENTER - LAS VEGAS, NEVADA
-------
TESTS OF THE
DUAL DIFFERENTIAL RADIOMETER
WORKING PAPER No, 473
ENVIRONMENTAL MONITORING AND SUPPORT LABORATORY
LAS VEGAS, [NEVADA
DECEMBER 1975
764
-------
Effective June 29, 1975, the National Environmental Research
Center-Las Vegas was designated the Environmental Monitoring and
Support Laboratory-Las Vegas, and the National Environmental
Research Center-Corvallis was designated the Environmental Research
Laboratory-Corvallis.
The mention of trade names or commercial products does not
constitute U.S. Environmental Protection Agency endorsement or
recommendation for use.
-------
CONTENTS
Page
CONCLUSIONS 1
INTRODUCTION 2
THEORY OF OPERATION 3
PROCEDURES 4
RESULTS 7
DISCUSSION 12
SUMMARY 17
REFERENCES CITED 18
APPENDIX A - SENSOR BUNDLE FIELD OF VIEW
GEOMETRIC VALUES FOR VARYING h, a 20
-------
TESTS OF THE DUAL DIFFERENTIAL RADIOMETER
by
Robert W. Thomas
Water and Land Quality Branch
Monitoring Operations Division
Environmental Monitoring and Support Laboratory
Las Vegas, Nevada
Working Paper No. 473
U.S. ENVIRONMENTAL PROTECTION AGENCY
December 1975
-------
CONCLUSIONS
The differential radiometer was designed to provide real-
time remote measurement of the near-surface chlorophyll a_ content
of water bodies. It effectively spans the range 0.03 to 20.0
micrograms (yg) chlorophyll a/liter. Dual differential radiometers
were tested on National Eutrophication Survey (NES) lakes in
the Eastern United States to determine their applicability for
the remote sensing of chlorophyll a content in U.S. surface
waters. Output voltages obtained from the differential radiometers
showed little correlation to chlorophyll a_ values determined
from NES samples.
Although the NES tests were often compromised due to time,
space, weather, and other considerations, sufficient data were
collected to permit the following conclusions:
(1) The dual differential radiometer has limited application
in the remote sensing of surface chlorophyll
-------
INTRODUCTION
Remote sensing of water quality parameters would allow
rapid, cost-effective surveys and monitoring of surface waters.
The dual differential radiometer 1s a two-channel instrument
designed to remotely measure light in four spectral bands.
Arveson, Weaver, and Millard (1971) selected filter pairs that
allowed remote measurement of chlorophyll a in Pacific Coast
waters with fair precision. Their successled to interest in
the instrument as a possible device to rapidly survey water
bodies throughout the United States and to estimate their trophic
state on the basis of observed chlorophyll a_ values.
The National Eutrophication Survey is a research effort
investigating the threat of accelerated eutrophication in freshwater
lakes and reservoirs. Nationwide in scope, it is designed to
develop, in conjunction with State environmental agencies, information
on nutrient sources, concentrations, and impact on selected
surface water bodies. Consequently, it was decided to test
the differential radiometer as a possible vehicle to measure
chlorophyll a_ on the NES lakes.
Two dual differential radiometers were obtained and installed
aboard NES helicopters. These units were tested during the
1972 and 1973 field operations. Tests were conducted on lakes
and reservoirs which varied in size, shape, depth, and water
quality. Water colors ranged from red/brown to blue and Seechi
disc transparency from a few centimeters to over 7 meters.
Although not quantitatively measured, suspended sediments were
noted in many of the water bodies at the time of sampling. ,
These tests were conducted within the framework of the NES sampling
effort. Therefore, testing was limited largely to obtaining
spot readings when approaching or departing a lake sampling
site. Few transects were obtained due to time restrictions.
-------
THEORY OF OPERATION
The dual differential radiometer measures upwelHng sunlight
from bodies of water 1n four spectral bands. The system 1s
designed for airborne operation and real-time detection of small
changes in spectral radiance. Its theory 1s based on a simple
form of the correlation spectrometer. The Instrument 1s configured
to correlate the specific spectral characteristics unique to
chlorophyll and to reject or cancel the background.
The absorption spectra for a variety of phytoplankton have
been determined (Yentsch, 1960; Friedman & Hickman, 1972; Grew,
1973). These spectra indicate that for many species there is
a maximum absorption in the blue region at about 440 nanometers (urn)
due largely to chlorophyll, a relatively transparent region
between 530 and 650 nm9 and a secondary absorption maximum near
the red at about 680 nm. These specific absorption bands modify
upwelling sunlight from water at characteristic wavelengths
corresponding to absorption maxima and minima.
To determine chlorophyll a_, a sample filter with a maximum
transmission at 443 nm, close to one of the absorption maxima
of phytoplankton, is paralleled to a reference filter with maximum
transmission at 525 nm. This latter filter lies outside the
major absorption region of phytoplankton. The two selected
wavelengths lie near the absorption minimum for water, thus
minimizing its effect on the returning radiation. Variations
in upwelling light from a water body due to water surface roughness9
scattering, or haze should have a similar effect on both wavelengths.
Variations in the concentration of algae will primarily effect
the intensity at the selected wavelengths. The resultant differential
signal output can be calibrated by comparison with ground truth
chlorophyll a_ values to yield chlorophyll a_ concentration (NASA/AMES,
n.d.; Arveson et al., 1971).
-------
PROCEDURES
The dual differential radiometers were Installed and tested
aboard float-equipped Bell UH-1H "Huey" helicopters utilized
by NES. Voltage readings were obtained from the instruments
while either approaching or departing an NES sampling station.
At each site the helicopter would land on the water, perform
in situ measurements, and collect water samples for later analysis
(EPA, 1974 and 1975). Samples for chlorophyll ^analyses were
collected at each station from water integrated from the surface
to 4.6 meters or to the lower limit of the photic zone, whichever
was greater. In waters less than 4.5 meters deep, the lower
limit of the integration was a point just off the bottom.
Samples were collected in unused polyethylene bottles and stored
in an icebox aboard the helicopter. At day's end, they were
removed and analyzed using a modification of the fluorometer
procedure described by Yentsch and Menzel (1963). In addit1ons
in 1973 surface samples were collected for the specific purpose
of comparing surface chlorophyll a^ values with the radiometer
data.
During the 1972 field year, use of the dual differential
radiometer was attempted utilizing an aligning yoke (located
in the helicopter rotor well) which directed the sensor bundle's
field of view 20° from the vertical in any selected quadrant.
This configuration made calibration of the instrument a tediouss
torturous process and exposed the sensor assembly to damage.
Radiometer data compared to ground truth chlorophyll a_ levels
showed little correlation. This, coupled with a high amount
of down time, inadequate instructions in Its use, and a work
schedule demanding 10- to 12-hour days, 7 days a week, resulted
in field researchers viewing the radiometer largely as a hindrance
to more important work. Consequently, 1t received minimal
attention and effort.
John Arveson of NASA/AMES Research Center visited the
Environmental Monitoring and Support Laboratory-Las Vegas during
the 1972-73 winter. He assisted in repairs and calibration,
advised as to methods of installation, and instructed NES personnel
in the use and peculiarities of the instruments. He also provided
a new calibration curve (Figure 1). Subsequently, the instrument
was tested over Lake Mead, Nevada. The remotely obtained data
compared very favorably with simultaneous ground truth chlorophyll
data (Figure 2). It was therefore decided to continue testing
the instrument during the 1973 NES field season 1n the Eastern
and Southeastern United States.
-------
Figure 1. Radiometer calibration curve
(after Arveson)
1001
50
10
5
en
105
0-
O
o:
9
U
0.1
005
/ 7
0.01 ti-Fi-;
I" :•/
J
LZ.
1Z
-10-8-6-4-2 O +2 +4 +6 +8
OUTPUT VOLTAGE
1/13/72 - Original Curve
1/13/73 - Modified Curve
-------
The instrument was relocated such that the limnologist could
aim the sensor bundle through an open window of the helicopter. The
dual differential radiometer was calibrated daily (and often before
each use) by aiming it at the sun with a Teflon diffuser over the
optic bundle. Since this was often done while airborne, concern as
to the effect of the rotor shadow was expressed. Several simple
experiments indicated that there was no appreciable effect on the
calibration although a 5% light loss was encountered. Whenever
possible, calibration was performed prior to takeoff with the rotor
still or by banking the helicopter while in flight to allow a direct
line to the sun.
Voltage readings were taken by pointing the sensor bundle at
the water surface in a direction away from the sun. Measurements
were taken at elevations from 60 to 150 meters above the lake level
during acceptable weather conditions. Care was exercised to avoid
including the helicopter's shadow, the float bag, or portions of
the shoreline within the 30° field of view. Throughout these tests
only one channel and filter pair per instrument were employed at
one time. Changes from Channel A to Channel B were made periodically
with no noticeable effect on the voltages obtained.
-------
RESULTS
In the early spring of 1973 test flights dedicated to testing
the dual differential radiometer were made on transects down Las Vegas
Bay into Lake Mead proper. Chlorophyll a^ levels decreased from the
upper end of the bay toward the main lake body. Figure 2 presents
a reproduction of the strip chart record obtained from the radiometer
on a transect flown about 150 meters above lake level. Ground truth
data collected at sites 1 and 2 compare quite favorably. Because of
rapidly deteriorating weather conditions ground truth data at sites
3 and 4 were not obtained.
During the 1973 NES field sampling in the Eastern and Southeastern
United States use of the radiometer was attempted when weather permitted.
Voltage readings were obtained by directing the sensor bundle toward
the sampling site upon approach or departure from an elevation which
varied from 60 to 150 meters above the lake. Lakes visited varied
in size, morphology, trophic state, phytoplankton assemblage, and
chlorophyll a_ level. Data collected, are presented in Figure 3. It is
readily apparent that the correlation between the output voltage and
chlorophyll a^ levels is poor.
Mr. Arveson visited the NES team in the field and checked the
calibration of the internal filters. He also performed some minor
maintenance and again instructed personnel in the use of the instruments.
Following this visit, several more tests of the differential radiometer
were conducted. The results are presented in Figure 4. Data obtained
were also reviewed on an individual lake basis (Figure 5). Again,
no consistent relationship was apparent in either case.
-------
Figure 2. Radiometer strip chart record, Lake Mead
(December 1972.)
h3.6
5.M3
o
ct
2
o
hose
Site 4
Ui
Site 1
Ground Truth Chlorpphyll-a (mg/l).
0.90
1.14
00
TIME - DISTANCE
Horizontal distance approximately 10 kilometers
(Chlorophyll scale from 1972 calibration curve)
-------
Figure 3. Radiometer output vs. chlorophyll a_
Spring 1973 data
10O
Ralcjionhetet: A!
-2 0 2 4 ^ 8
OUTPUT VOLTAGE
(1) Output voltage vs. chlorophyll a_ content of surface dipped
water samples.
(2) Output voltage from Unit A vs. chlorophyll a_ content of
integrated water sample.
(3) Output voltage from Unit B vs. chlorophyll a_ content of
integrated water sample.
(calibration curves superimposed)
-------
10
Figure 4. August 1973 radiometer data.
•101 234
OUTPUT VOLTAGE
RADIOMETER OUTPUT vs. CHLOROPHYLL a., AUGUST 1973 DATA
(Calibration Curves Superimposed)
-------
11
Figure 5. Data for selected large reservoirs
01 2345
OUTPUT VOLTAGE
Old Hickory Reservoir, TN; 8/16/73
Cumberland Reservoir, KY; 8/21/73
Chickamauga Reservoir, TN; 8/23/73
Rend Lake, IL; 8/8/73
Berlin Reservoir, OH; 7/30/73
Lake Carlyle, IL; 8/10/73
(Lines between points for clarity only)
-------
12
DISCUSSION
Results of testing the dual differential radiometer by NES
personnel were generally poor. Little correlation between radiometer
output voltages and chlorophyll a^ was evidenced. There are several
probable reasons why this is so. Not the least of these is the fact
that the instrument had a low priority among NES objectives, and personnel
commitment to its use was correspondingly low. The extremely hectic
schedule of the NES field team left little opportunity for involved
personnel to review data in detail, perform side experiments, and
fully investigate the instrument and all its nuances.
As an example of this last consideration, the geometry of the system
was not thoroughly investigated at the time of use. Figure 6 presents
the geometry of the radiometer as it was used in 1973. Table 1 presents
the major and minor axes and area of the field of view of the sensor
bundle at several elevations and angles from the vertical.
As can be seen, the sensor bundle scans an ellipse on the ground
and integrates over this area. The area of the ellipse defining the
field of view is much larger than most of the NES team members realized.
Consequently the inclusion of shorelines, visible lake bottom, or other
objects that would affect the reflected light differentially probably
occurred occasionally. Because light intensity varies as a square
function with respect to distance, it is apparent that the portion of
the ellipse nearer to the vertical has more effect than that further
from the helicopter.
Light reflected from the water surface could mask the backscattered
signals. Since the sensor bundle was always directed away from the
sun, this effect was minimized. However, experiments by both NASA
Langley (NASA, 1973) personnel and NES indicate that a slight increase
in voltage with increasing angle is experienced utilizing the 443- and
525-micron filters.
These considerations, however, do not explain all the scatter
observed. Results obtained by researchers at NASA Langley Research
Center utilizing the dual differential radiometer were also disappointing
(NASA, 1973; Witte, 1975). They noticed that, among other problems,
atmospheric haze affected the solar standardization zero; boat wakes
could cause up to a 15% of full-scale deflection in the radiometer output;
and that the geometry of the system with respect to viewing angle, solar
orientation, and stream flow affected the results.
-------
13
Figure 6. Sensor bundle field of view geometry.
c
) F
X
/
C £
a\
L_
3 A
The distance DA = h tan (o + 15°)
The distance BA = h tan (a - 15°)
The minor radius a = (tan 15°)£C
The major radius b = (DA - BA)/2
= h
i. cos (a + 15°)
where: lc - FA * h
FA = BA + b
and the area of the. ellipse
i =
cos (o + 15°)
-------
14
Table 1. SENSOR FIELD OF VIEW FOR VARYING HEIGHT, ANGLE.*
h
meters
60
80
100
150
300
a
o
200
30
0
200
30
o
200
30
o
200
30
o
200
30
a
meters
17.3
19.0
23.0
25.4
28.8
31.7
43.2
47.6
86.4
95.2
b
meters
18.4
22.0
24.5
29.3
30.6
36.6
45.0
54.9
91.9
109.8
Area
meter;
997.8
1313.6
1774.1
2334.6
2772.2
3648.4
6236.2
8208.0
24,944.6
32,835.0
*See Appendix A for English Units
-------
15
Many of the difficulties encountered apparently stem from one
basic assumption, that "changes in light intensity, variations in water
surface roughness, or scattering within the water body have similar
effects on the intensity at both wavelengths and are automatically
corrected" (Arveson, 1971). Although not stated, the implication
is that the only significant variation in signal between the two
wavelengths is due to chlorophyll absorption. This is definitely
not the case. Clarke et al. (1970) showed that as the altitude of
the sensor above the water body increased there was a differential
increase in the percent of incident light upwelled with the larger
increase occurring at lower wavelengths. Considering the elevations
encountered during the NES study this effect would probably be negligible.
However, since it is largely attributable to "air light" (light back-
scattered by the atmosphere) the presence or absence of haze or smogs
differences in humidity, etc., would differentially affect the two
measured spectra.
Prewett et al. (1973) found that the relative reflectances from
four ponds of differing suspended solid concentrations changed differ-
entially. Ritchie et al. (1974) were able to correlate these changes
with suspended sediments. In both studies reflectances were greater
at 525 nm than at 440 nm. This would result in an increased difference
in the signals received and give higher than true chlorophyll a^ values.
This was indeed evidenced in NES data. Although NES did not measure
suspended sediments, their presence was observed in many of the reservoirs.
particularly in the Southeast.
Another consideration is the composition of the phytoplankton
population. Spectral absorption curves for different algae can be
expected to vary as the influence of the various dominant pigments
changes. The absorption curves presented by Friedman and Hickman (1972)
demonstrate this nicely. Not only does the magnitude of the absorption
for differing genera vary, but for a terrestrial red algae, there is a
secondary maximum at about 560 nm. arid the absorption at 525 nm is about
three-fourths that at 443 nm. Similarly, in the data presented by
Grew (1973), the 525-nm filter value is significantly up the absorption
curve as it passes from its maximum to its minimum values for two of the
species. On the spectral signature Grew obtained for Clear Lake,
California, the signal strength at 525 nm is nearly one-half that at
443 nm. Clearly, one cannot expect chlorophyll a_ measurements based
on spectral characteristics of 443 and 525 nm to be the same in each
of these cases.
Grew (1973) also discusses the results of multispectral analyses
along a flight line over the New York Bight. Changes were observed
in spectral bands centered at 468 nm and at 543 nm, indicating that
the 525-nm reference band is affected by other parameters. He suggests
that one possiblity for these changes is due to particle size. In the
atmosphere and in clear water, backscattered light is predominantly
due to Rayleigh scattering which occurs when the size of the scatterers
-------
16
is much smaller than the light wavelengths. The amount of scatter is
inversely proportional to the fourth power of the wavelength. A second
type of scattering, Mie scattering, becomes important as particle size
approaches 1 microemter. Mie scattering is predominantly in the forward
direction and nonselective as to wavelength, the result is a differential
change in the radiometer output signal due to particle size alone.
Other factors could also affect the values differentially. The
presence of large amounts of pollens, dyes, or other colored substances
could be expected to result in a differential output. Surface films
of oils would greatly reduce the upwelled signal from the water and
probably increase the reflected light. Since none of these factors which
may differentially affect the sensor channels were allowed for during
the testing, the poor results are not surprising.
-------
17
SUMMARY
The concept of the dual differential radiometer is basically
straightforward. The instrument is small, relatively inexpensive,
and electronically simple. It may be readily installed in light
aircraft without difficulty. Because of these considerations and
the capabilities demonstrated by both Arveson et al. (1971) and the
NES tests on Lake Mead, it is felt that the instrument can be
successfully utilized in some situations.
The radiometer could be of great value in contouring chlorophyll a_
content in a lake of relatively low turbidity. However, it cannot
be effectively employed without collection of simultaneous ground
truth data, at least until the presence or absence of interferences
is established. For successful utilization it should be operated
by a well trained individual and receive a high operations priority.
Possible interferences should be known, considered, and allowed for.
The dual differential radiometer should not be considered a
proven field instrument and used blindly without regard to the various
possible interferences. As presently configured, it cannot be
successfully utilized to measure chlorophyll a_ on areas of high turbidity.
It is weather limited, requires a water body of substantial size and
depth, cannot tolerate excessive boat wakes, and is subject to other
interferences. From the NES data and literature review, it is felt
that calibration curves should be constructed for each different
water body, and perhaps for the various seasons. (Typically, lacustrine
phytoplankton populations shift from diatom domination in the spring
to blue-green or green algae predominance in the summer. Suspended
sediments would also typically be greater in the spring months than
later in the year.) Also, it cannot be successfully used without
considerable attention as to its operation. Frequent checks on solar
zero, angle of view with respect to solar azimuth elevation and
atmospheric conditions must be made.
-------
18
REFERENCES CITED
Arveson, J. C., E. C. Weaver, and J. P. Millard. 1971. Rapid
Assessment of Water Pollution by Airborne Measurement of
Chlorophyll Content. Paper presented at the Joint Conference
on Sensing of Environmental Pollutants, Palo Alto, California,
November 1971. AIAA Paper #71-1097. 7 p.
Clarke, G. L., C. C. Ewing, and C. J. Lorenzen. 1970. Spectra of
backscattered light from the sea obtained from aircraft as a
measure of chlorophyll concentration. Science 167:1118-1121.
Friedman, E. J. and G. D. Hickman. 1972. Laser Induced Fluorescence
in Algae: A New Technique for Remote Detection. NASA Contract
#NAS6-2081 Final Report, October 1972. 103 p.
Grew, G. W. 1973. Remote Detection of Water Pollution with MOCS:
An Imaging Multispectral Scanner. Proceedings, Second Con-
ference on Environmental Quality Sensors, Las Vegas, Nevada,
October 1973. U.S. EPA. pp. 11-17 through 11-40.
NASA. 1973. EPA-NASA Water Pollution Detection Sensor Evaluation
(Status Review, September 1973). Langley Research Center.
NASA/AMES Research Center, n.d. Operating Manual for Dual Differ-
ential Radiometer. 19 p.
Prewett, 0. E., D. R. Lyzenga, F. C. Polcyn, and W. L. Brown. 1973.
Techniques for Measuring Light Absorption, Scattering, and
Particle Concentrations in Water. Environmental Research
Institute of Michigan. (NTIS AD-759 668). 32 p.
Ritchie, J. C., J. R. McHenry, F. R. Schiebe, and Wilson. 1974.
The Relationship of Reflected Solar Radiation and the Concen-
tration of Sediment in the Surface Waters of Sediments. Pro-
ceedings of the Third Remote Sensing of the Earth Resources
Conference (unpublished manuscript).
-------
19
U.S. Environmental Protection Agency. 1974. National Eutrophication
Survey Methods for Lakes Sampled in 1972. National Eutrophication
Survey Working Paper No. 1. Environmental Monitoring and Support
Laboratory, Las Vegas, Nevada, and Environmental Research Labora-
tory, Corvallis, Oregon. 40 p.
1975. National Eutrophication Survey Methods 1973-1976.
National Eutrophication Survey Working Paper No. 175. Environ-
mental Monitoring and Support Laboratory, Las Vegas, Nevada,
and Environmental Research Laboratory, Corvallis, Oregon. 91 p.
r
Witte, W. G. 1975. Evaluation of the Dual Differential Radiometer
for Remote Sensing of Sediment and Chlorophyll 1n Turbid Waters.
Paper presented at the Fourth Annual Remote Sensing of Earth
Resources Conference, Tullahoma, Tennessee, March 1975.
Yentsch, C. S. 1960. The influence of phytoplankton pigments on
the color of sea water. Deep-Sea Research 7:1-9.
and D. W. Menzel. 1963. A method for the determination of
phytoplankton chlorophyll and phaeophytin by fluorescence.
Deep-Sea Research 10:221-231.
-------
APPENDIX
SENSOR BUNDLE FIELD OF
VIEW GEOMETRIC VALUES FOR VARYING h, a.
-------
Sensor Bundle Field of View Geometric Values for Varying h9 a.
h
ft
200
300
400
500
1,000
a
o
o
200
30
o
200
30
o
200
30
0
200
30
o
200
30
a
ft
58
64
86
95
115
126
144
159
288
317
ft
62
73
92
110
122
146
153
183
306
366
0
ft
201
207
301
311
402
414
502
518
1,004
1,035
j,
f?
244
283
366
424
488
566
610
707
1,220
1,414
I
f!
215
237
322
355
430
473
537
592
1,075
1,184
Area
ft2
11,206
14,678
249856
329830
44,257
57,991
69,215
9194iO
276,862
364 , 494
------- |