EPA-600/3-76-087
August 1976
ENVIRONMENTAL AEROSOL MEASUREMENTS
USING AN AIRBORNE PARTICLE MORPHOKINETOMETER
by
W.M. Farmer and J.O. Hornkohl
Spectron Development Laboratories, Inc.
Tullahoma, Tennesee 37388
DA-6-99-2294A
Project Officer
Jack L. Durham
Aerosol Research Branch
Environmental Science and Research Laboratory
Research Triangle Park, NC 27711
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF RESEARCH AND DEVELOPMENT
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
RESEARCH TRIANGLE PARK, NC 27711
-------
DISCLAIMER
This report has been reviewed by the Environmental Science and
Research Laboratory, U.S. Environmental Protection Agency, and approved
for publication. Approval does not signify that the contents necessarily
reflect the views and policies of the U.S. Environmental Protection
Agency, nor does mention of trade names or commercial products constitute
endorsement or recommendation for use.
ii
-------
ABSTRACT
Measurements of ambient aerosols using an airborne particle morphokineto-
meter are described. The measurements were of large particles (greater than
5 micrometers in diameter) in environmental aerosols around Phoenix, Arizona,
during November 1975. Speed of the sample space on the airborne platform
and time resolved measurements of relative particle number density and particle
size distributions for horizontal and vertical flight profiles were obtained.
The results show that the large aerosol particles were predominantly 50-75
micrometers in size, that the relative number density could vary greatly in
space and time, and that the largest concentration of particles were within
28 meters of ground level.
iii
-------
CONTENTS
Page
Abstract iii
List of Figures vi
List of Tables ix
I Introduction 1
II Summary 5
III Conclusions 7
IV Recommendations 9
V PM Signal Analysis for Determination of 11
Particle Size
VI The SDL Particle Morphokinetometer 21
VII Results 35
VIII Calibration and Error Analysis 79
References 85
Appendix, Daily Log 87
v
-------
LIST OF FIGURES
Number Page
1 Generation of Free Space Interference Fringes with a
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Huygen's Diagram of Interference Fringe Generation in
PM Probe Volume
Visibility as a Function of D/6 for Two Particel Shapes . . .
Uncertainty in Spherical Particle Diameters for Various
Values of Uncertainty in the Visibility
V2700 FF 500. Channel 3 Forward and Backscatter Observation
Modes
T2700 Model B. As Flown on the EPA Helicopter
Optical System Installation on the EPA Helicopter
PMAC 554 Signal Processing Electronics
Block Schematic of the Signal Processing Electronics
DC Integrator Circuit
AC Integrator Circuit
November 19. Histogram From Instrumentation Test Flight . . .
November 19. Particle Size Distribution Obtained During
Instrument Test Operation Flight
November 22. Histogram 7, Flight 1. Particle Size
Distribution
November 22. Ground Test After Second Flight
November 24. Histogram 3, Flight 1. Particle Size
Distribution
12
15
17
22
23
25
26
28
29
31
38
39
43
44
47
17 November 24. Histogram 4, Flight 1. Particle Size
Distribution 48
vi
-------
LIST OF FIGURES
Number Page
18 November 24. Histogram 2, Flight 1. Particle Size
Distribution 49
19 Signal Time Period Distribution for Velocity Determination . . 51
20 November 24, Flight 2. Return to Airport 52
21 November 24, 0600 hours. Inversion Layer Histogram For
27.9 - 15.2 m Altitude Increment 54
22 November 24, 0600 hours. Inversion Layer Histogram For
13.9 - 1.3 m Altitude Increment 55
23 November 24. Ground Based Measurement Immediately After
First Flight 56
24 November 25. Flight 2. Signal Time Period Distribution
for Velocity Determination 58
25 November 25. Flight 1. Track 11, 152 meters AGL Altitude . . 60
26 November 25. Histogram 1, Flight 1 61
27 November 25. Histogram 2, Flight 1 62
28 November 25. Histogram 3, Flight 1 63
29 November 25. Histogram 4, Flight 1 64
30 November 25. Flight 1. Last Track East of Phoenix 65
31 November 25. Histogram 8, Flight 1 66
32 November 25. Histogram 9, Flight 1 67
33 November 25. Histogram 7, Flight 1 68
34 November 25. Histogram 8, Flight 1 69
35 November 25. Histogram 9, Flight 1 70
vii
-------
LIST OF FIGURES
Number Page
36 November 25, Flight 2 73
37 November 25. Histogram 1, Flight 2 74
38 November 25. Histogram 2, Flight 2 75
39 November 25. Histogram 5, Flight 2 76
40 November 25. Histogram 6, Flight 2 77
41 November 25, Flight 2. Track 11 (Glendale) 78
viii
-------
LIST OF TABLES
Number Page
1 Doppler Period Processor 32
2 Visibility Processor 33
3 Histogram Generator 34
4 Most Frequently Occurring Particle Size as a Function of
Altitude for 1600 Hours, November 19, 1975 37
5 November 22, Flight 1. PM System Data Summary 40
6 November 22, Flight 2. PM System Data Summary 41
!
7 November 24, Flight 1. PM System Data Summary 45
8 November 24, Flight 2. PM System Data Summary 46
9 November 25, Flight 1. PM System Data Summary 59
10 November 25, Flight 2. PM System Data Summary 71
ix
-------
SECTION I
INTRODUCTION
A number of different techniques can be used to measure large particles
(greater than about 10 micrometers in diameter) found in environmental
aerosols. Considerable difficulty exists, however, in using these
techniques accurately from an airborne platform. Large errors often
result from the fact that a mechanical probe must be inserted into the
airstream in order to sample the aerosol. These errors result from
impaction and settling in the sample line for particles greater than
about 5 micrometers. Therefore, the need exists for a device which
can measure such particles without the constraints imposed by mechanical
sampling procedures.
During the past ten years, a laser interferometer system has been
developed and successfully applied to a wide variety of applications
involving the measurement of particle speeds in fluid environments
ranging from water flows in pipes to high speed wind tunnels. 1~5
The utility of these instruments arises from the fact that they are
highly accurate, perform perturbationless measurements with exceptionally
high spatial resolution and they can function in severe environments
where mechanical measurements are impossible. One instrument design,
as illustrated in Figure 1, has found particularly wide acceptance
because of its versatility, simplicity, and insensitivity to vibration.
Light from the laser is split into two beams of equal intensity.
These beams are brought to a simultaneous cross-focus region where a
set of bright and dark interference fringes are generated. Particles
passing through the interference fringes scatter light which is observed
with a collecting lens and photodetector. The current generated in
the photodetector by the scattered light is sinusoidal with a frequency
directly proportional to particle speed.
Research has conclusively, both theoretically and experimentally,
demonstrated that in addition to a speed measurement with such
interferometer systems, it is also possible to determine the size
of the particle by measurement of the relative signal shape. °~8
Since such a measurement is obtained as a ratio of signal parameters,
the size determination is independent of absolute signal magnitude.
Hence, there is no need to calibrate either the laser for power output
or the photodetector for amplitude response in order to determine
particle size. Since particle size and speed can be simultaneously
measured, particle mass concentration can be determined when particle
density is known and the instrumental sample space is calibrated. Thus,
-------
-------
it appears that such a system should be capable of obtaining particle
size measurements where mechanical probes would be limited by losses
in the sampling line.
Spectron Development Laboratory (SDL) personnel have successfully
worked with such interferometer systems for a number of years in a
wide variety of applications. 9-12 As a result of that experience,
SDL has developed an instrument which is capable of optically mea-
suring particle size and velocity and sorts the acquired data into
histogram formats that allow convenient statistical analysis. Since
the device can measure parameters relative to shape and speed of the
particle, SDL has chosen to call its device a "particle morphokinetometer"
(PM).
A need arose for the Environmental Protection Agency (EPA) to determine
the nature of the atmospheric aerosol content in the Phoenix, Arizona
area. In order to provide measurements of the large particles which
might be present during the study, SDL was requested to install and
maintain its instrument on the EPA helicopter used for the airborne
aerosol measurements. This application is believed to be the first
time that such an instrument has been used successfully to obtain these
kinds of measurements from an airborne platform.
This report summarizes the results of the measurements so obtained
with the SDL instrument.
Section V provides a discussion of the principles utilized in the SDL
instrumentation. A detailed description of the operating characteristics
of the device and its installation on the helicopter is given in Section
VI. A summary and discussion of the data obtained is presented in
Section VII. Section VIII discusses possible instrument calibration
tests and error sources in the measurements. Appendix A provides a
daily operations log kept by SDL personnel in order to give the reader
some insight into how the instrument was operated and tested and the
problems encountered in this initial application.
-------
SECTION II
SUMMARY
The SDL particle raorphokinetometer has been successfully operated from
an airborne platform in Phoenix, Arizona. Data were obtained on six
different flights and under a variety of environmental conditions
including day and nighttime operation. Both velocity and particle
size distributions were obtained at two different positions relative
to the aircraft. Time resolved particle size and relative number den-
sity measurements were obtained for horizontal and vertical sample
directions. The horizontal sample revealed number density variations
as large as factors of 2 or 3 for sample path differentials as small
as 0.4 Km and factors as large as 100 for the overall sample. The
vertical sample, taken during the early morning hours of November 24,
revealed a very shallow inversion layer between 15 and 30 m. The
relative number density in the inversion layer was about 100 times
that measured above it. Measured mean particle sizes ranged between
50 and greater than 75 pm in diameter. The velocity measurements
yielded a value of 12.5 ± 4.9 m/sec. at a distance of approximately
66 cm from the edge of the aircraft and 9.5 ± 4 m/sec. at a distance
of 40 cm.
A primary limitation of the system was found to be signal noise resulting
from background sunlight. Due to the fact that an interference filter
was used with too broad a passband, background sunlight caused occasional
high levels of photomultiplier tube shot noise. Operation with a filter
combination having a narrower bandwidth considerably improved the
signal-to-noise ratio for bright sun operating conditions. '
Some of the particle size data obtained by the instrument is questionable
because of the smallness of the standard deviation of the size distributions
and lack of the slopes in the wings of the distribution. Also the
observed particle sizes on the ground covered a broader spectrum than
those observed airborne. Such effects suggest that either (1) the
instrument artificially biased the distributions in manners which could
not be detected during calibration or ground applications, or (2) large
fugitive dust sources such as farms or tmpaved roads produce large
particles with narrow distributions, but near paved roads (in the area
of the airport), tire action crushes the mineral, yielding smaller
particles with a greater dispersion.
-------
SECTION III
CONCLUSIONS
It has been clearly demonstrated that a PM system can function from the
EPA helicopter without any major technical difficulties. Shock mounting
for the optical system proved to be adequate, and all power inputs were
sufficient.
The instrument should be positioned on the aircraft to provide a sample
space outside any flow field perturbations. In any further applications,
a background light filter with a passband of 1 nanometer should be used.
The time-resolved histogram data clearly shows the high variability of
atmospheric number density for horizontal and vertical sample paths.
It demonstrates the need to exercise caution in estimating mass con-
centration from measurements taken over long time intervals; and that
when long sample time measurement techniques are used, numerous repe-
titive samples should be obtained. The particle size distribution
observed in the early morning inversion layer was nearly the same as
that observed on the ground. It is, therefore, to be concluded that
in that case, the particles in the first 30 m of altitude were uniformly
distributed.
The overall thrust of the data also suggests the high variability of the
day-to-day atmospheric particle content and the need to sample often to
obtain a clear picture of such.
-------
SECTION IV
RECOMMENDATIONS
The results of the PM system application in Phoenix suggest several
recommendations if the application of these measurement devices is
pursued. Specifically, it is recommended that:
(1) A second similar instrument application be attempted with
a forward scatter optical system. Such a system would not
be nearly as sensitive to background light (since the receiver
could never point directly toward the sun) and would generate
signals with much higher amplitudes. Such a system could use
a smaller laser and be more compact than the one used in
Phoenix.
(2) The system should be placed such that it is not exposed to
external flow field interferences (boundary layer effects,
struts, etc.) in order to prevent any biases which the flow
field might introduce.
(3) All electronic systems should be rack mountable. The systems
in the Phoenix application were rack mounted but were not
originally designed to be so.
(4) If possible such tests should extend for a sufficient period
of time to allow thorough data analysis between flights, and
to repeat the measurements under similar atmospheric conditions
in order to give a high confidence value to the creditability
of the data.
(5) Studies should be pursued to determine a visibility function
for randomly shaped and randomly oriented particles.
(6) In a future application, comparative measurements should be
obtained between the size determination using the mean
scattered intensity and the visibility measurement to provide
two independent determinations of particle size.
(7) In any future application, some means should be obtained of
automatically recording flight notes and data acquisition
times. A tape recorder would be profitable for such require-
ments.
-------
SECTION V
PM SIGNAL ANALYSIS FOR DETERMINATION OF PARTICLE SIZE
For completeness, a brief review is given of the basic principles
which must be utilized to determine particle size and velocity from
a measurement of signal shape and frequency.
THEORETICAL CONSIDERATIONS
Consider Figure 2. Two equal intensity, well-collimated coherent light
beams intersect at a common origin (geometric center) with an included
angle, a . A Huygen's diagram of the wavefronts show that planar inter-
ference fringes are generated which are perpendicular to the plane
defined by the beam centerlines and are parallel to the bisector be-
tween the beams. The distance between the periodic fringes, 6 , is
given by:
6 - A/[2sin(a/2)] (1)
where X is the wavelength of the coherent light. The region of inter-
ference forms an ellipsoid of revolution which is a contour of constant
intensity about the bisector between the beams. Particles which pass
through the ellipsoid generate periodic signals. The number of cycles
in the signal depends on the number of interference fringes which the
particle trajectory intercepts. A maximum number of cycles is generated
when the trajectory is in the plane of the beam centerlines and passes
through the geometric center of the ellipsoid. Particles passing
through the edge of the ellipsoid will always generate fewer cycles of
information. When a particle (assumed spherical) much less than 6
in diameter crosses the fringe pattern, it can be assumed to be uniformly
illuminated at all points along its path through the fringe pattern,
and the light which is scattered by the particle is proportional to
the observable flux illuminating it. Thus, measurement of the time
period, r , of the scattered light is related to the velocity, v, of
the particle through the relationship
v = 6/r (2)
11
-------
-2
Modulation
Contour ^--
Figure 2. Huygen's Diagram Of Interference Fringe Generation In
PM Probe Volume.
12
-------
As the size of the scattering particle increases relative to 6 ,
the illumination of the particle is no longer uniform and must be
averaged over the cross-sectional area of the particle. The non-
uniform illumination of the particle results in a reduction in
the contrast or visibility of the scattered light signal. Let
^max be the maximum value in intensity in a period of the scattered
light from a particle and Imin tne next successive minimum. The
visibility, V , can then be defined as:
-'-max 1-min
V = (3)
It is straightforward to show that V is fully equivalent to the ratio
of AC amplitude divided by the DC amplitude of the scattered light
signal. Henceforth the high frequency "Doppler" portion of the signal
will be referred to as the "AC", it usually has many cycles of informa-
tion relative to that of the DC component (the DC component refers to
the mean value of the Gaussian shaped low frequency term describing
the signal). Analytically the visibility may be written as:
f I0 cos(2Try/6)dAp
V - (4)
r
JA
where Ap is the cross-sectional area of the particle, IQ is intensity
distribution across one of the illuminating beams, and y is the
coordinate normal to the fringe planes. When I0 is a Gaussian
function (TEM OOq laser beam) it can be shown that Equation 4 is an
accurate approximation over a depth of field, & , given by:
£ ~ 0.8 b/a (5)
13
-------
f\
where b is the radius of the e L intensity point in the illumination
beam. For depths of field greater than Si , the signal visibility
is a function of particle size and position in the illumination.
In order to simplify Equation 4 for Gaussian beams and still maintain
accuracy, it is required that the particle diameter, D, satisfy the
relationship:
D £ 0.2 b (6)
and for & to satisfy:
S £ 0.2 b (7)
Under these conditions V for a sphere can be written as:
V = 2J1(irD/6)/(TTD/6) (8)
where J^ is a first order Bessel function of the first kind. For a
cylinder V can be written as
V = sin(irL/6)/(TTL/6) (9)
Equations 8 and 9 are plotted in Figure 3 to illustrate the salient
features of the visibility in particle size measurement.
Determination of particle size using a visibility measurement has a
number of advantages over other optical techniques. Among the more
important of these are:
14
-------
1.0
0.9
0.8
0.7
0.6
e °*4
H
M
1
* 0.3
0.2
0.1
0.0
Major Axis
Of Fringe
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
D/6
Figure 3. Visibility As A Function Of D/6 For Two Particle Shapes.
15
-------
(1) High quality scattered light collecting systems are not required.
(2) A visibility measurement can determine particle size to at least
ten times better than the resolution limit of the transmitting
lens.
(3) With calibration, true particle flux (i.e., mass concentration)
may be determined with a PM system.
(4) Size measurements may be made from backscattered light.
On the other hand, disadvantages in using this technique are also
apparent. For example, design and application of a PM system must
account for the facts that:
(1) Size measurement is shape dependent.
(2) High quality transmitting optical systems are required.
(3) Particle size measurement of spheres is limited to approximately
a 15:1 size variation range for a given fringe period.
(4) High quality spatially coherent light is required for good
interference fringe quality.
(5) The particle size measuring ability is limited by the frequency
response of the signal processing electronics.
Item 5 above results from the fact that the signal frequency is velocity
dependent. Hence, an instrument which measures signal visibility is
limited to the measurement of particles with velocities which are less
than those which produce a signal frequency greater than the frequency
response limit of the instrument.
SIGNAL PROCESSING CONSIDERATIONS
To minimize error in the estimate of particle size, it is of utmost
importance that signal visibility be measured with high accuracy and
precision. This is due to the fact that signal visibility error and
particle size error are not linearly related. For spherical particles,
Figure 4 illustrates the particle diameter uncertainty for several
16
-------
10
,-1
Q
ea
10
,-2
ID'3 "-
J
T 1 1 T
UNCERTAINTY IN VISIBILITY
± 0.10
UNCERTAINTY IN VISIBILITY
± 0.01
ACCURACY LIMITED BY WIDTH OF 8 BIT
DIGITAL WORD ± 1/256
n
-j
I
-i
--i
-1
_L I
A I
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
D/6
Figure 4. Uncertainty in spherical particle diameters for various
values of uncertainty in the visibility.
17
-------
different values of visibility. As the figure shows, where the ratio
of particle diameter D to fringe period 6 is near 1, the particle
diameter uncertainty is the same as the visibility uncertainty.
However, D/6 decreases, particle diameter uncertainty rapidly increases.
Thus, when D/6 is small, the error in particle size will be much greater
than that in the visibility. An instrument which measures signal
visibility must seek to minimize errors in the visibility measurements
if an accurate particle size measurement is to be obtained.
POTENTIAL SOURCES OF ERROR IN A SIGNAL VISIBILITY MEASUREMENT
In examining a PM signal, a number of potential error sources are
evident which could produce considerable error in a visibility mea-
surement. Among the most important error sources are:
(1) Signal noise
(2) Asymmetric signal shapes
(3) Variations in visibility in the signal
(4) Signals resulting from more than one particle
Signal noise may result from any number of sources but is usually either
(1) broad band Gaussian distributed noise occurring during the entire
duration of the signal, (2) single very-high-frequency spikes which
occur randomly in the signal, or (3) periodic amplitude variations
which mix with the signal. The first two forms of noise are usually
evident when signal magnitudes are low. The third form usually results
from radiative pickup from high power consuming or generating instruments
operating across the entire frequency spectrum. Asymmetric signal shapes
results from particles with trajectories that do not pass through the
optical axis of the probe volume. Such signals can also arise through
turbulence-induced distortions in the signal. Variations in signal
visibility occur when D/6 is near 1, and the AC component of the signal
has only a few cycles of information relative to that of the DC component.
Signals resulting from more than one particle usually generate aperiodic
signals of skewed shapes. Most PM optical designs account for the
possibility of multiple particle signals by generating a probe volume
which is small enough to produce a low probability of observing more than
one particle. Certain types of error logic may also be used in the
signal processing electronics to reject such signals.
18
-------
VISIBILITY FROM SIGNAL INTEGRATION
In order to minimize the errors which may result from "real world"
kinds of PM signals, it is most desirable to observe and measure as
much of the PM signal as possible. One method of satisfying this
requirement is through integration of the entire signal (within preset
limits). Consider a particle whose trajectory passes through or near
the geometric center of the probe volume. It can be shown that the
visibility obtained by (1) computing the integral of the DC component,
(2) computing the integral of the full wave rectified value of the
AC component, and (3) dividing the integrated DC into the integrated
AC, is (to within a known numerical factor) identical to that measured
for a single cycle measured at the center of the signal. Integration
of the signal is most desirable because the noise components previously
listed will contribute very little to an integral value of the signal.
For example, Gaussian-distributed white noise and periodic noise have
zero or nearly zero values for an integration, and high frequency noise
spikes contribute very little to the overall integral value. Further-
more, integration of asymmetric signals tends to reduce the effects of
the asymmetry of producing an overall signal average. The primary dif-
ficulty in using a signal integration to determine the signal visibility
resides in the fact that the dynamic response of an integrator will be
affected by large frequency variations in the signal input. Thus,
without design precautions, an integrator could produce considerable
error in attempting to measure particle size in turbulent flows where
there are considerable velocity variations.
SIGNAL AMPLITUDE CONSTRAINTS
While a visibility measurement is independent of signal magnitude,
detailed consideration must be given to relative signal amplitude in
order that a visibility measuring instrument have a sufficient dynamic
range in amplitude response.
For particles of identical index of refraction and of a size greater
than about 5 urn (i.e., for particles with scattering gain coefficients
independent of particle size), the signal amplitude can be expected to
vary by a factor of 100 over the measurable size range. Probe volume
intensity variations can be expected to range over factors of 10 in
each of two orthogonal probe volume dimensions. Thus, an instrument
must be capable of responding to relative signal amplitude variations
19
-------
of at least 10 :1 and, more reasonably, 10^:1. Linear amplifiers can
cover a linear dynamic range of approximately 10^ for a single range
setting. The designer is thus faced with the difficulty of producing
an instrument, which at one range setting can cover sufficient dynamic
range to eliminate instrumental bias in particle size measurements.
One technique to overcome the dynamic range limitations is the use of
logarithmic amplifiers. For example, if an integration technique is
used to determine the signal visibility from a logarithmically amplified
signal, it can be shown that the visibility is given by:
V = sech
|Jln(iAC) - Jln(iDC)| (10)
where Jln(i^) is the integrated logarithmically amplified AC
component of the signal and CIn(i0c) is the integrated DC component.
Logarithmic amplifiers can easily cover a dynamic range of 10^:1 for
low frequency signals, making an instrument much more sensitive to
small particle signals. The primary difficulty with such amplifiers
is that their sensitivity is a strong function of signal frequency and
quickly decreases as the frequency increases making their effectiveness
questionable for most airborne applications.
Another solution to this limitation is obtained in using the integration
approach to visibility measurement. An integrator can be design to
cover a 10:1 dynamic range with three digit accuracy, and 100:1 with
two digit accuracy. The signal visibility processor designed by SDL
personnel utilizes 4 parallel integration circuits each overlapping
the other in range such that for a given signal frequency a dynamic
amplitude range of 10^:1 with three digit accuracy can be covered.
Such accuracy gives an acceptable error in particle size determined
by a visibility measurement even when the D/6 particle size is small.
20
-------
SECTION VI
THE SDL PARTICLE MORPHOKINETOMETER
The components of the particle morphokinetometers divide naturally into
two distinct subsystems - the optical system and the signal processing
system. The following sections describe these subsystems and their
mounting to the EPA helicopter.
THE SDL OPTICAL SYSTEM
The optical system used in this research was a slightly modified version
of an optical system built by SDL called the ^2700. The ₯2700 is built
in two models, A and B. The A model uses a 5mw HeNe laser while the B
model uses a 15mw laser and carries the laser and photomultiplier tube
power supplies inside the optical package. The ^2700 is designed to
cover a 2700:1 particle size range using four possible channels of
operation using different lens combinations. The instrument is designed
to cover the size range 0.18 - 500 micrometers and can be used in a
forward or backward scatter observation mode. For the helicopter appli-
cation, backscatter measurements with the probe volume approximately
50 cm from the edge of the optics cabinet were to be attempted. Hence,
it was decided that the ₯2700B in a channel 2 operating mode was to be
used. This would provide maximum signal amplitude and, therefore,
optimize the signal-to-noise ratio for the electronics system. The
channel 2 lens set was arranged to provide a probe volume approximately
51 cm from the edge of the optical cabinet. In this configuration, the
fringe period was 75 micrometers and a particle size range of 5 - 75
micrometers could be measured. Figure 5 schematically shows the optical
system light path while Figure 6 shows a photograph of the actual
instrument. The backscatter collection system utilized an F/ll lens
arrangement. A slit aperture was used on the photomultiplier tube to
reduce background light and eliminate possible multiple particle signals.
A 10 nanometer interference filter was originally used for background
light rejection. This filter was found to be insufficient for operation
on particularly bright sunny days. However, when a second 10 nanometer
filter was used with the first, sufficient background light rejection
appeared to be obtained.
A massive rack mount (400 kg) which was shock mounted and sat near the
helicopter door provided the primary support structure for the optical
system. The EPA was to furnish a small angle bracket which would allow
21
-------
J
J ;uO FF rj()0. Cli.iniu'l *. Forw.irJ and B.H ksc.it ti-r Obsi-rv.it i on
Modi's.
22
-------
Figure 6. ₯2700 Model B. As Flown On The EPA Helicopter
23
-------
mounting the optical system to the rack mount such that the optics
cabinet protruded 10 - 12 cm outside the helicopter doorway. Pre-
liminary tests of the optical system on the EPA bracket showed that
it was not strong enough for safe flight applications. Using the
EPA bracket as a pattern, SDL personnel were able to fashion one from
0.31 cm x 2.5 cm x 2.5 cm aluminum angle which was sufficient.
Figure 7 shows the optical system as it was installed on the helicopter.
This mounting system proved to be adequate during all flights. No
optical misalignment due to helicopter vibration was ever observed.
THE SIGNAL PROCESSING ELECTRONICS SYSTEM
The SDL electronics system employed during this research is designated
as the PMAC 554. This electronics system is shown in Figure 8. It
contains electronics which measures the mean amplitude of the scattered
light signal, the signal time period (for velocity), and the signal
visibility (for particle size). In addition to the above measurements,
the electronics system was also capable of sorting the acquired data
into histograms of either signal time period or visibility. The histogram
generator is capable of being programmed for 15 bins of arbitrary width.
There are four possible normalization modes for a particle count of
10, 10^, 10^, or 10^ particle measurements. The normalization modes
are:
(1) A prescribed number of measurements in the total histogram.
(2) A prescribed number of measurements in the most populated
bin in the histogram.
(3) A prescribed number of measurement attempts (events)
regardless of the number of measurements in the histogram.
(4) A prescribed measurement time interval (0.1 - 99.9 seconds).
The instrument is designed such that an operator can terminate any
histogram sequence at will or can make the instrument automatically
acquire a series of histograms until commanded to stop.
The histograms acquired during the flight were recorded by a digital
recorder. Two different recorders were used during the flights, a
24
-------
t-l
-------
Figure 8. PMAC 554 Signal Processing Electronics.
26
-------
Digitec model 6100 and a Hewlett-Packard model 5055A. The Digitec
recorder failed to operate properly during the third sampling day and
was replaced with the rented Hewlett-Packard recorder. The Digitec
recorder printed at a 2.5 line/sec rate while the Hewlett-Packard
recorder printed at 10 lines/sec. In addition to the recorder, a
Tektronix model 7633 storage oscilloscope was flown with the system
in order to directly observe the signal output from the photomultiplier
tube. These instruments were all strapped directly into the helicopter
mounting rack for inflight use.
Figure 9 shows a block schematic of the signal processing system in
front of the histogram generator. The figure illustrates how both
signal visibility and time period are obtained, and the logic tests
the signal must pass before it is considered acceptable for analysis.
Since signal visibility was the parameter of primary interest, the
following sections describe in detail how this measurement was obtained.
SDL PM VISIBILITY PROCESSOR
Spectron's approach to the measurement of fringe visibility is to use
all of the information in the entire signal for the measurement. This
is in contrast to other techniques (e.g., peak detection) in which the
visibility is determined from only sampled portions of the signal burst.
An integrator as shown in Figure 10 is a circuit which can sample a PM
signal in its entirety. The integrator is turned on (i.e., starts
integrating) at the beginning of the signal and is turned off at its
end. If the signal is integrated directly, the periodic and random
fluctuations in the signal contribute nothing to the final value of the
integral. The integrator output is given by:
1 r
Jeir
c 4
eout = ein dt
RC
in which ein is the input voltage, T is the integration period (i.e., the
signal duration time), R is the integrator resistance, C is the integrator
capacitance (RC is the integrator time constant which is of the same
27
-------
tntiitffliit
tin
ntfftnnn
!'itMir« '>. Illoi-k Sclicm.il ic Ol Tin- Signal I'rori'ss inft Klv-fl ron i i-s.
28
-------
R
Input
WV-
Integrate
Switch
Reset Switch
-o-
- Output
Figure 10. DC Integrator Circuit.
29
-------
order of magnitude as T), and t is time. In the visibility processor
under discussion, the input signal is divided into two components.
The first component is integrated directly with no further electronic
manipulation. The integral of this signal approximates the integral
of the DC portion of the signal. That it is not fully equivalent to
the integral results from the fact that the AC component of the unaltered
signal will integrate to zero only if T is an integral multiple of the
Doppler period TQ. The circuit which turns the integrator on and off
does not necessarily produce an integral time duration which is exactly
an integral multiple of Tp. Therefore, the AC component of the signal
can make a contribution to the directly integrated signal. However,
the time error in Tj) cannot exceed ± 0.5TD and the first and last
Doppler periods in the signal are small in amplitude compared to the
central periods. Hence, the error in the integrator output is typically
less than one percent for this integration to approximate that of the
DC component of the signal that can be measured.
A second integrator is used to measure the AC component but this
integrator is preceeded by two other circuits as shown in Figure 11.
The DC component is first removed by a high-pass filter where the output
contains only the AC component of the signal. The positive half of
this signal is integrated, the negative half is inverted and then
integrated, and the two integrals are summed. If the negative portions
were not inverted, the two integrals would sum to zero. This design
approach is electronically implemented by full-wave rectifying the high-
pass filter output (which has no effect on the positive signal excursions
but which inverts the negative excursions) and then integrating the rec-
tifier output with a second integrator (see Figure 11). The output of
the second integrator is a measure of the Doppler component of the
signal.
The fringe visibility is obtained from the ratio of the AC integrator
output to the DC integrator output.
The integration techniques outlined above have the advantage of using
the entire signal for the visibility measurement. This design choice
was made with belief that an averaged measurement would be more desirable
than a sampled measurement in most applications. The integrators attenuate
periodic and random noise sources. The basic integrator circuit shown
in Figure 10 is noted for its stability and accuracy, and it is widely
used in many common electronic instruments. Tables 1-3 summarize
specific characteristics of the SDL PM signal processing system.
30
-------
Reset Switch
iHigh-Pass
Filter r
A
\
f
r- , R
Full Wave 1 ,-,
Rectifier 1" O- "6 A/sAr~^
1
Integrate .
Switch
-\j \J- - .
-11 - +
1 r T
c
!" !
--, I
.x"
x^'
*'
Figure 11. AC Integrator Circuit.
31
-------
Table 1. DOPPI.ER PERIOD PROCESSOR
Frequency Response (various channel
carrier frequencies available on
request)
Time Period Resolution:
Time Period Accuracy:
Bandpass Filter Ranges:
0-10 MHz
3 Digit,BCD
Nominal 1%
0.5 KHz, 1.0 KHz, 2 KHz,
5 KHz, 10 KHz, 20 KHz,
50 KHz, 100 KHz, 200 KHz,
500 KHz
Integrated Signal Amplitude
Dynamic Range:
Maximum Data Acquisition
Minimum Signal Input Amplitude
Data Output Code
Error Logic
20,000 sec
10 millivolts
Analog, 12 Bit BCD
and Binary. TTL Compatible
Programmable Periodicity
R.I I Lo Check
Progr.mmnbl e Aper Lodi city
Accept. ince Error
Min/Max Cycle Count
Programmable Periodic
Signal Sampling
Processor Operating Voltage
HOv/60 Hz
-------
Table 2. VISIBILITY PROCESSOR
Frequency Response (various channel
carrier frequencies available on
request)
Visibility Range
Visibility Resolution
Visibility Accuracy
Equivalent Possible Spherical
Particle Size Range
Signal Amplitude Dynamic Range
Maximum Data Acquisition Rate
Minimum Signal Input Amplitude
Data Output Code
Error Logic
-1
Processor Operating Voltage
0-2 MHz
1.0 - 0.1
± 0.1%
± 0.1%
15:1
104
20,000 sec
10 millivolts
Analog, 12 Bit BCD
and Binary. TTL Compatible
Programmable Periodicity
Ratio Check
Programmable Aperiodicity
Acceptance Error
Min/Max Cycle Count
Programmable Periodic
Signal Sampling
HOv/60 Hz
33
-------
Table 3. HISTOGRAM GENERATOR
Forms Histograms of:
Signal Visibility
Signal Time Period
15 Bins
Bin Width Programmable.
Four Data Normali/alion Modes
Total Histogram Signal Count For Preset Count
Number (10, 102, 103, 10^)
Total Signal Count For Most Frequently Occupied
Histogram Bin For Preset Count Number (10, 102, 103, 104)
Total Event Count For Preset Count Number
(10, 102, 103, 10*)
Periodic Histograms For Preset Sample Time (10/sec.- 0.01/sec.)
Total Number of Events Display
Number of Mcasureim-nts in Histogram Memory Display.
t Total Signal Count Timer
Nominal Maximum Memory Fill Rate 50,000 sec
Data Output Code: BCD TTL Compatible (Binary, Analog Optional)
-------
SECTION VII
RESULTS
The following is a presentation of the results obtained with the PM
system during the data acquisition flights and specific flight tests
for the instrument. The data is identified by the date and particular
flight on which it was recorded.
Since there was no provision made for an automatic time record of the
data acquisition times for the PM system and any other instruments used
for the inflight samples, only rough estimates of where a particular
sample was obtained during the flight can be made. The inflight operator
kept notes of facts pertinent to the condition under which a particular
data set was obtained. These notes were recorded on the digital recorder
printout, and are summarized in the tables for each day's flights.
The daily log given in Appendix A indicates why the instrument could not
acquire data on the days for which it is lacking.
NOVEMBER 19 / TEST FLIGHT
The first flight for operating the PM system was a test flight to deter-
mine its airborne operating characteristics. Particular attention was
paid to environmental effects (e.g., aircraft vibration and sunlight) on
the operating characteristics of the system. The data samples obtained
with the system were in the Chandler area near an agricultural region
that was at the edge of the desert. A number of passes were made at
different altitudes and through dust raised by agriculture equipment
operating on fields freshly prepared for planting.
Background sunlight was observed to present some problem when the zenith
angle of the sun with respect to the optical transmission axis was less
than about 45°. A technique which appeared to significantly reduce the
background light was to limit the instrument operation to times when
the door of the aircraft pointed away from the sun. With the exception
of the digital recorder, the instrument appeared to function well in
all respects. The digital recorder printed the histogram results in
such a manner that it did not consistently print the "10" column digits,
and occasionally did not print digits in the other columns. It was
usually found that the digits were not printed if they were zeros.
35
-------
Thus, the particle size distribution data obtained is subject to broad
error bands when the "10" column was uncertain and the number of mea-
surements small. Tt was possible, however, to estimate the most frequently
occurring particle size. Table 4 provides a summary of this kind of data
for flights over a series of grass covered fields which were bounded on
one side by an occasionally travelled unpaved roadway. As Table 4 shows,
the particle size mode follows the trend that might be expected for
particle size variation with altitude. Figures 12 and 13 show an example
of the particle size distribution obtained during a 30 m pass over
freshly prepared fields at approximately 1700 hours. The distribution
contained sufficient counts such that a reasonably accurate size distri-
bution could be obtained. The standard deviation for this distribution
is surprisingly small for what might have been expected under the cir-
cumstances. The skewness shown by the histogram in Figure 12 is what
might be expected for a crushed powder and is therefore reasonable.
A second interpretation of the data suggests that there are two sources
for large particles - paved and unpaved roads. The paved roads should
yield a broader size distribution with a smaller mean size due to the
crushing action of tires on the hard surface. The unpaved roads should
yield distributions with larger mean sizes due to the stirring action
of tires passing along the road bed. The deviation of the data points
about the straight line on the accumulative probability plot is believed
to be due to the uncertainty resulting from the missing "10" column
due to erratic recorder operation.
NOVEMBER 22 / DATA ACQUISITION FLIGHTS
Two missions were flown on November 22. Preflight tests on the ground
indicated potentially high noise levels due to background sunlight.
However, it was decided to fly the system with the operator choosing the
operation sequences for minimum sunlight levels. After the system was
airborne, the signal threshold level was optimized to a value of about
7 and reduced to smaller values for the late afternoon flight. Tables
5 and 6 summarize the data obtained during these flights.
The data obtained during the first flight showed the unexpected result
of a single bin being populated in 20 - 25 urn range. This occurred
with varying numbers for the first 6 histograms. Such a count configuration
was not observed thereafter. The count configuration is not indicative
of system noise (which would tend to fill the largest particle size
bins). There is no reason to suspect an instrument malfunction for this
data, although the fact that nearly all the measurements are contained
in a single bin makes the data somewhat questionable.
36
-------
Table 4. MOST FREQUENTLY OCCURRING PARTICLE SIZE AS A FUNCTION OF
ALTITUDE FOR 1600 HOURS, NOVEMBER 19, 1975
Altitude (m, AGL*) Altitude (ft. AGL) Particle Size (pro)
152.4 500 40 - 45
91.4 300 55 - 60
30.5 100 65 - 70
*Note that the term "AGL" references altitude relative to
altitude "above ground level".
37
-------
50
w
o
53
a
D
U
O
O
En
O
J
M
CQ
CQ
§
40
30
20
10
0
<5 10 15 20 25 30 35 40 45 50 55 60 65 70 75> 75
PARTICLE DIAMETER (pm)
Figure 12. November 19. Histogram From Instrumentation
Test Flight @ 30.5 m AGL In Chandler Area.
38
-------
13
o
H313WVICI
39
H
4-> X
en O
a) t-J
4J P.
P.
4J co
a) f.
e 4J
3 -H
U) 0)
C T3
H 3
j_i
60 -rl
C 4-1
H ^H
j-i rt
a iJ
H
cfl
C
o
H
H O
M VJ H
4-J CD (/)
03 r-l -~~
H 13 cn
T3 C 4-1
CO C
01 ,n 01
NUB
H O>
en CD t-i
3
05
0)
u n a)
H -H 0
4-1
S-l 4-J O
co ^r o
a. to CN
H
a)
e
-H n!
,0 4-1 VJ
6 n)
ai v-i ni
> 0) 4-1
o a, cd
Z o t3
a)
M
M
H
tM
-------
Table 5. NOVEMBER 22, FLIGHT 1. PM SYSTEM DATA SUMMARY
Histogram Number Track Altitude (meters AGL) Flight Notes
I ?* 30.5 7.4 Threshold
2 ?* 30.5 Stray Laser
Light Due to
Pull-In Strap
For Filter
Holder
3 ?* 30.5
4 ?* 30.5
5 ?* 30.5
6 ?* 30.5
7 1* 30.5
8 ?* 30.5 (2) 30 Sec. Auto
Sequences
*F] ight notes do not indicate tracks for which data was
obtained.
40
-------
Table 6. NOVEMBER 22, FLIGHT 2. PM SYSTEM DATA SUMMARY
Histogram Number Track Altitude (Meters AGL) Flight Notes
1 ?* 30.5 High Sun Level
2 ?* 30.5 High Sun Level
3 ?* 30.5 5.0 Threshold
4 ?* ?** Dust Mission
5 ?* ?** 4.3 Threshold
6 ?* ?**
7 ?* ?** 3.8 Threshold
8 ?* ?**
9 ?* ?**
10 ?* ?** Ground Test
*Flight notes do not indicate tracks for which data was
obtained.
**No altitude indication on digital recorder data.
41
-------
Figure 14 is the "reduced" particle size histogram obtained from
histogram 7. The term "reduced" is used to describe the fact that
the histogram was normalized with the counts in the 20 -25 Mm bin
removed. Even with the counts for the 20 - 25 ym bin removed, the
distribution is clearly multimodal. The data seems to show distinct
mode sizes at 20 - 25 ym, 55 - 60 |jm and somewhat greater than 75 urn.
However, the unreduced histogram has 96% of the measurements concen-
trated in the 20 - 25 pm bin.
During the second flight, a very low percentage of measurements were
recorded in the 20 - 25 ym bin (typically less than 4%). The large
majority of the measurements were recorded in the greater than 75 ym
bin. However, immediately upon landing (1800 hours) a particle size
histogram was obtained for the ground level distribution. Figure 15
shows the results of that set of measurements. Three mode sizes
25 - 30, 50 - 55, and greater than 75 ym are apparent.
NOVEMBER 23, 24 / DATA ACQUISITION FLIGHTS
Two flights were flown during the evening of November 23 and early
morning of November 24 before sunrise. The first flight began at about
2230 hours on November 23 and terminated at about 0030 hours November 24.
The second flight began at 0400 hours November 24 and terminated at
0600 hours. The atmosphere was exceptionally clear and the data acqui-
sition rate low. For example, on November 19, the data rate had been
between 10 and 200 measurements per second. During the track measurements
on the 23 and 24, the data acquisition rate varied from 0.25 per second
to less than 0.1 per second. Tables 7 and 8 summarize the data acquired
during the flights.
The first two histogram sequences during the first flight were for 30
second automatic sequence times. It became apparent that at the 366 m
altitude, the particle density was too low to record any measurements
in the sample time. Hence, the instrument was placed in a manual opera-
tional mode and only one histogram attempted per flight track. Nearly
all the particle sizes observed were greater than 60 micrometers in
diameter. Figures 16 - 18 are histograms for the particle sizes observed
during these flights. The data was deemed to be too sparse for accumu-
lative probability plots which could provide statistical information on
the data so obtained. However, it is interesting to observe that the
distribution mode appears to shift toward smaller sizes as the altitude
is increased.
42
-------
U
D
U
U
O
H
(£)
8
50
40
30
20
10
0
1 1 1 1 T
r I
< 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75> 75
PARTICLE DIAMETER (pm)
Figure 14. November 22. Histogram 7, Flight 1. Particle Size Dis-
tribution.
43
-------
o
D
L)
CJ
O
H
H
^
M
CQ
§
50
40
30
20
10
T 1 1 1 1 T
I 1 I I
1
< 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75> 75
PARTICLE DIAMETER (ym)
Figure 15. November 22. Ground Test After Second Flight. Particle
Size Distribution.
-------
Table 7. NOVEMBER 24, FLIGHT 1. PM SYSTEM SUMMARY
Histogram Number Track Altitude (Meters AGL) Flight Notes
1 10 366 (1) 30 Sec. Histogram
2 10 366 30 Sec. Autosequence
3 10 366 Histogram Acquisition
Time: 2318 - 2335
4 4 152 Histogram Acquisition
Time: 2348 - 0005
5 4 366 East Bound. Histogram
Acquisition Time:
0007 - 0025
6 4 366 Velocity Histogram
45
-------
Table 8. NOVEMBER 24, FLIGHT 2. PM SYSTEM DATA SUMMARY
Histogram Number
Track
Altitude (Meters AGL) Flight Notes
4
5
6
7
Flight To
Track 3
3
3
Return To
Airport From
366 meters
366
152
152
366
366
152
76
Threshold Test
@ 4.5
Threshold Test
@ 5.0
Threshold Test
@ 4.5 (very low
data rate)
10 Second
Autosequence
46
-------
u
PS
a
OS
D
CJ
u
o
fn
o
>l
EH
M
^
M
§
CQ
50
40
30
20
10
I I i I I I I I T
I I I
I I
< 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75> 75
PARTICLE DIAMETER (ym)
Figure 16. November 24. Histogram 3, Flight 1. Particle Size Dis-
tribution.
47
-------
8-*
u
OS
D
0
O
O
O
M
CQ
O
50
40
30
20
10
iI I I r
T r
.J I L
1 L_JL_
rn
<5 10 35 20 25 30 35 40 45 50 55 GO 65 70 75> 75
PAKTICI,K D1AXKTKR (pm)
Figure 17. November 24. Histogram 4, Flight 1. Particle Size Dis-
tribution.
48
-------
W
o
U
u
o
CM
O
M
ra
<
m
o
a:
50
40
30
20
10
0
ii iir
<5 10 15 20 25 30 35 40 45 50 55 60 65 70 75> 75
PART 1 C1,E D I AMKTKR (vim)
Figure 18. November 24. Histogram 2, Flight 1. Particle Size Dis-
tribution.
49
-------
Near the end of the flight on track 4, a velocity histogram was obtained.
The measurements were for a sample space which set 51 cm (20.25") from
the edge of the optics cabinet. The results are shown in Figure 19 as
an accumulative probability plot of the signal time period. The figure
shows that the mean velocity was 12.5 ± 4.9 m/sec., which is considerably
less than the helicopter air speed of approximately 41 m/sec. The signal
time period distribution shows that the flow through the sample space
was quite turbulent. It should be borne in mind that the velocity com-
ponent measured is that parallel to the centerline of the aircraft, and
that the sample space of the instrument could still be well within the
boundary layer of the aircraft, or shadowed by the wheel strut.
Due to the exceptionally low data rate, it was decided to vary the signal
level threshold during the second flight in order to determine if the
instrument was possibly rejecting signals with too low an amplitude.
There was no measurable difference between the low and high values of the
instrument threshold settings. This indicated that the particle number
density was extremely low as the first flight had indicated. Over 95% of
the measurements obtained in the histograms during this flight were
predominately located in the bin containing sizes greater than 75
micrometers. The data rate was typically less than 0.8/sec.
Upon return to the airport, the PM was set to automatically obtain a
histogram sample once every 10 seconds as the aircraft descended from an
altitude of 366 m. Descent rate was 1.27 m/sec, so that a sample of the
relative number density over 12.7 m intervals was obtained. The PM did
not operate while the digital recorder recorded the histogram. Thus,
there are 2 m intervals where no sample was obtained while the recorder
functioned. Figure 20 shows that measured relative number density as
a function of altitude. It is clear that the number density is not
isotropic with altitude. The figure shows variations of factors of 2
or 3 in 25 m altitude changes. The figure shows an exceptionally high
number density below an altitude of about 28 m which is believed to be
indicative of the inversion layer. The data for the last 14 m of altitude
suggests that the PM is observing particles swept up from the ground
by the aircraft downwash. This conclusion follows by examining the
particle size distributions for these altitudes and comparing them with
a distribution obtained approximately 10 minutes after landing. An
accurate estimate of absolute number density is not possible since the
size of the sample space cross-section was not calibrated. However,
a rough estimate of the cross-sectional area can be made predicated on
previous water spray calibration measurements. For a 500 ym slit
aperture width, the absolute number density p is roughly approximated as:
50
-------
7
ON
ON
ON
ON
0>
CO
0>
ON
ON
ON
oo
ON
tri
ON
O
ON
O
OO
O
f
o
o
o
O
CM
3
o
CM
o
o
OJ
H
nJ
c
o
H
U
n)
C
H
B
y<
0)
J-)
0)
H
U
O
O hJ
<4-4 O
<1
c
o B
4-1 T3
cn c
H 3
T3 O
O 4->
H W
VJ CO
a) w
o.
a;
H O
ON
iH
(U
50
o
d
(Klt) (I01H5I.I MW1.I. 'IVND1S
51
-------
J
o
-------
p = N/vA (12)
p = 2.9.10 3N cc l (13)
where v is the sample space velocity, A is the sample space cross-sectional
area and N is the number of observed signals per second. The assumed
numerical value of A is purely a guess predicted on the observed length-
to-width ratio observed in water spray calibrations. It is expected to
be within a factor of 2 or 3 of the true value.
Figure 21 shows that the mode for the size distribution is between 70 and
75 micrometers for the altitude increment lying between 27.9 and 15.2 m.
However, Figure 22 shows that particle size mode is greater than 75 ym
in the altitude increment nearest the ground. Comparison of these two
distributions with that of Figure 23 indicates the 27.9 - 15.2 m altitude
distribution is very similar to that of the undisturbed atmosphere at
ground level. Since the previous flight data showed that the PM sample
space velocity is approximately 1/3 that of the aircraft, it would appear
entirely possible that the sample at the 14 to 1.3 meter altitude reflects
the ground level dust stirred into the aircraft airstream by the rotor
downwash. The above conclusions, however, are not definitive since there
is an approximate +12.7 m, -0, uncertainty in altitude which resulted from
the way the automatic sequence was terminated. Figure 20 was constructed
under the assumption that the automatic sequence was terminated immediately
after the last histogram was recorded.
NOVEMBER 25 / DATA ACQUISITION FLIGHTS
The flights for November 25 began at approximately 0700 and 1030 hours
respectively and were about two hours in duration. During these flights,
a second background light filter was used giving an estimated passband
of 70 nanometers. This reduced the scattered light amplitude to the
photodetector but allowed the use of increased amplifier gain. To
further compensate for the reduced instrument transmission, the lenses
in the optical system were adjusted to move the sample to a position
28.5 cm (11.25 inches) from the optics cabinet. The same fringe period
used at the 51.4 cm (20.25 inches) position was maintained. At this
position, the sample space appeared to be nearly shadowed by the aircraft
wheel strut. Furthermore, since the sample space was moved closer to the
53
-------
H
U
D
U
75
Figure 21.
27.9
PARTICLE DIAMETER (ym)
November 24, 0600 hours. Inversion Layer Histogram For
15.2 m Altitude Increment. Particle Size Distribution.
54
-------
w
o
(J
U
O
H
ffl
<
rQ
O
100
90
80
70
60
50
40
30
20
10
r~~r
irr
'j iO ^5 -10 45 50 55 60 65 70 75> 75
22.
l'J.9
I'AK'I'ICI.H 1)1 Av.K'I'KR (pin)
Novembor 24, 0600 hours. Inversion Layer Histogram For
1 .') m AH illicit* Increment. Particle Size Distribution.
55
-------
M
U
U
0
O
Cn
O
M
^
H
CQ
75
PARTI rijK DIAV.KTKR (ym)
Figure 23. November 24. Ground Based Measurement Immediately After
First Flight. Particle Size Distribution.
56
-------
aircraft, it is to be expected that the air velocity should be less than
at the previous sample space position. As the data shows, this expecta-
tion was confirmed. Figure 24 is an accumulative probability plot of
the signal time period distribution obtained during the second flight.
The mean time period indicates a velocity of 9.6 ± 4 m/sec. Comparison
with the velocity measurement at the previous sample space indicates that
the turbulence level was increased (probably because of the wheel strut).
Table 9 summarizes the status of the histograms obtained during the first
flight. A programming error was made in the PM preflight setup which
caused all measurements of particles greater than 75 micrometers not to
be entered into memory. As a result, the distributions are somewhat
skewed. This error was corrected for the second flight. During the
second flight, it was found that a relatively small percentage of the
size distribution occurred for sizes greater than 75 ym, so that this
error for the first flight had little effect on the overall results.
Figure 25 presents the results of histograms 1 - 4 in terms of accumulative
probability plots. Figures 26-29 are the corresponding histograms.
The distributions are characterized by mean sizes between 68 and 72 ym,
and relatively small standard deviations. The data for histograms 5-7
was obtained for the most part in an auto-sequence mode, and predominately
contains measurements in the 70 - 75 ym size interval. Data from the
last track east of Phoenix for the first flight is given in Figures 30 -32.
These data suggest that while the mean particle size remains approximately
constant, the standard deviation of the distribution has increased. In
fact, the data indicated measurements across the entire size range although
the frequency of occurrence was very small compared to that for sizes
generally greater than 60 ym. It was interesting, therefore, to examine
the relative distribution of these measurements with the larger sizes
removed. Figures 33 - 35 show the results of artificially modifying
histograms 7 - 9 by removing the populous larger size bins. Since these
distributions represent only about 100 total measurements each, very little
can be said about the distributions they might represent. It is interesting
to note, however, that there appear to be distinct size modes occurring
between 5 - 10 ym and 15 - 30 pm as shown in Figures 34 and 35 respectively.
Data obtained during the second flight is summarized in Table 10. Data
obtained in the first two histograms was obtained during signal level
threshold adjustments. The originally set threshold was adjusted down-
ward to give the instrument increased sensitivity after it was observed
that background light did not appear to interfere with the measurement
at the lower threshold setting. The data from the first two histograms
57
-------
c> o> at
/
o.
00
o.
Ov
o
a\
o
OJ
C)
m
o
CM
Oa\cor-vou~i
o
o
w
O
8~S
"3
>.
H
O
,-t
>
O
o
>-.
u
W
H
T3
O
O
-------
Table 9. NOVEMBER 25, FLIGHT 1. PM SYSTEM DATA SUMMARY
Histogram Number
Track Altitude (Meters AGL) Flight Notes
1
2
3
4
5
6
7
8
9
11
*
152
**
A*
**
30.5
30.5
30.5
30.5
Flight Notes Are
Sparse. Data
Sequence Suggests
That Histograms 2 & 3
Are Also At 152 meters
Track 11.
11.5 Miles South Of
Phoenix VOR
30 Sec. Autosequence
8 & 9 Are Last Run
On East Side of City
*No Track Indication On Digital Recorder Data.
**No Altitude Indication On Digital Recorda Data.
59
-------
i
E-i
0)
N
H
a)
iH
O
4-1
M
01
T3 -tf
3
4-J t3
H q
O
< ^
CO ro
)-i
a) "
O
n)
60
O
4J
en
DO
H SJ
rH O
" 0
m o
CM -H
-2
B
0)
>
o
55
00
H
60
-------
1UU
90
80
8 70
53
g 60
u
u
o
fc 50
O
H 40
H
CQ
n 30
o
a.
20
10
n
III 1 1 1 l ll II
__
Ill 1 . 1 1 1 1 1 1
--
< 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75> 75
PAKTFCLE DIAy.KTKR (\im)
Figure 26. November 25. Histogram 1, Flight 1. Particle Size Dis-
tribution.
61
-------
CJ
CJ
CJ
o
O
>H
EH
M
CQ
<
CQ
O
100
90
70
60
40
30
20
10
I I
~F rrr
T~ r
I I I
I I
< 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75> 75
PARTICLE DIAMKTER (ym)
Figure 27. November 25. Histogram 2, Flight 1. Particle Size Dis-
tribution.
62
-------
M
a
u
o
o
EH
H
J
H
PQ
§
TOO
90
80
70
60
50
40
30
20
10
ir
i i i i r
1
< 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75> 75
PARTICLE DIAXKTER (ym)
Figure 28. November 25. Histogram 3, Flight 1. Particle Size Dis-
tribution.
63
-------
CJ
r-*
s
u
O
O
fn
O
EH
H
H
O
100
90
80
70
60
50
40
30
20
10
T T 1 1 I f T -r-T --,- r
1 1 1 1
J L_J.
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75> 75
PARTICLE DIAMETER (ym)
Figure 29. November 25. Histogram A, Flight 1. Particle Size Dis-
tribution. .
64
-------
oo r-»
z
CTi
00
oo
0\
in
CT\
O
00
o
CM
O
iH
m
O
Q)
T3
3
4-1
H
hJ
a
o
ro
d
X «
H
01 A
o *
PH
00
N
)-i cu
Q) iH
XI O
S -H
CD 4-1
> VJ
O cfl
O
CO
01
VJ
040
H
aaiawvia
65
-------
U
CJ
u
o
EH
H
J
M
CQ
rt!
en
§
1 00
90
80
70
60
50
40
30
20
10
"i i T "i" r~ ~\r
< 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75> 75
PARTICLE DIAMETER (pm)
Figure 31. November 25. Histogram 8, Flight 1. Particle Size Dis-
tribution.
66
-------
W
u
O
U
O
EH
H
A
H
CQ
CQ
O
100
90
80
70
60
50
40
30
20
10
T T
J I I I I I I I
1 L
< 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75> 75
PARTICLE DIAMETER (pm)
Figure 32. November 25. Histogram 9, Flight 1. Particle Size Dis-
tribution..
67
-------
U
U
U
o
o
EH
H
^
M
CQ
§
50
40
30
20
10
1 I T
1 I 1 1 1 T~T
<5 30 15 20 25 30 35 40 45 50 55 60 ' 65 70 75> 75
PARTICLE DIAMETER (pm)
Figure 33. Novenber 25. Histogram 7, Flight 1. Reduced Particle Size
Distribution.
-------
75
PARTICLE DIAMETER (pirn)
Figure 34. November 25. Histogram 8, Flight 1. Reduced Particle
Size Distribution.
69
-------
50
40
w
u
75
PARTICLE DIAMETER (pm)
Figure 35. November 25. Histogram 9, Flight 1. Reduced Particle
Size Distribution.
70
-------
Table 10. NOVEMBER 25, FLIGHT 2. PM SYSTEM DATA SUMMARY
Histogram Number Track Altitude (Meters AGL) Flight Notes
1 11 30.5 Threshold: 9.2
2 11 30.5 Repeat of Histogram
1 (3 8.0 Threshold.
519 Second Requisition
Time.
3 11 30.5
4 11 30.5 Velocity Histogram
5 11 229 Possible Sun Noise.
Passing Through Brown
Cloud. 1720 Sec. Data
Acquisition Time.
6 11 457 Heading South Above
Brown Cloud.
7 2? 152 Glendale Run. N. Bound
803 Sec. Autosequence.
8 Descent Into Airport
@ 152 m/roin.
71
-------
are presented in Figures 36 - 38. The mean size is approximately that
observed during the first flight while the standard deviations appear
to be slightly broader, although surprisingly narrow compared to what
might be expected. Histogram 3 showed results identical to that of
Histogram 2. Histogram 4 was a velocity measurement and its results
have already been indicated. Histograms 5 and 6 were obtained in the
vicinity of an operator-observed "brown cloud". The data from these
histograms is shown in Figures 39 and 40 and show a remarkedly narrow
distribution with the mode centered between 70 and 75 ym. Histogram 7
represents a series of histograms obtained in continuous 30 sec. inter-
vals over nearly 10 km of horizontal flight path. The modal value
observed on nearly all of the histogram sequences was between 70 - 75 ym.
Figure 41 shows the relative number density observed along the flight
path. The figure clearly shows the high variability of the particle
concentration. Two dashed lines are shown on Figure 41. These represent
the mean number density observed: (1) with the first 900 m of flight
path excluded from the computation and (2) for the entire flight path.
The first 900 m of flight path were through an exceptionally dense
particle concentration which increased the entire path average concen-
tration. If the first 900 m of path is excluded, the mean number density
falls to half that observed for the entire path. These data clearly
demonstrate the care that must be exercised in estimating atmospheric
particle concentration when averaging over long time intervals or long
flight paths.
72
-------
O
o
Ul
0)
I
o
z
vC
en
SO
H
h.
(uirt) ^_OT.
73
-------
w
u
S3
u
o
fn
O
H
H
CQ
O
100
90
80
70
60
50
40
30
20
10
< 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75> 75
PARTICLE DIAMETER (ym)
Figure 37. November 25. Histogram 1, Flight 2. Particle Size Dis-
tribution.
74
-------
w
u
o
u
o
EH
H
A
H
CQ
§
100
90
80
70
60
50
40
30
20
10
iir
JL_J I I I I I
< 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75> 75
PARTICLE DIAMETER (ym)
Figure 38. November 25. Histogram 2, Flight 2. Particle Size Dis-
tribution.
75
-------
o
P3
a
o
o
9
PQ
O
100 i
90
80
70
60
50
40
30
20
10
I
1 I I I J II
< 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75> 75
PARTICLE DIAXETER (ym)
Figure 39. Novenfcer 25. Histogram 5, Flight 2. Particle Size Dis-
tribution.
76
-------
u
u
u
o
EH
H
^
PQ
§
100
90
80
70
60
50
40
30
20
10
0
T 1 i 1I 1 T
< 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75> 75
PARTICLE DIAMETER (pm)
Figure 40. November 25. Histogram 6, Flight 2. Particle Size Dis-
tribution.
77
-------
w
H
S
0
OJ O
P4 N
<£ <}
O
CNl pM
U~>
rH 01
4-1
cn c
M O>
01 cu
T3 H
4-> C
H M
rH
CN -H
,C cn
60 -H
H O
> -H
LO CO ft
CN C 4J
CU CO
y> Q PH
B 0) rH
qt n p.
> B B
co
60
rl
.UISN'JC
3AILV'I3a
-------
SECTION VIII
CALIBRATION AND ERROR ANALYSIS
The following discussion describes the calibration followed to characterize
the response characteristics of the instrument. Potential error sources
during its operation in the flight tests and their effects on the resulting
measurements are indicated.
CALIBRATION PROCEDURES
It has been well established that for particles where the shape is known
that a measurement of the signal visibility gives an accurate measure of
the particle diameter. Since a visibility measurement does-not depend on
the absolute value of the scattered light intensity, there is no need
to calibrate the light detector response or the laser power output of the
optical system. The primary optical uncertainty in the particle size
measurement results from the accuracy with which the fringe period is
determined. For beam angles less than 5°, the fringe period may be
obtained by using the approximation:
6 - A/a (14)
Since the laser wavelength is known to a high degree of accuracy, the
only parameter to be measured is the angle between the transmitted beams,
a . The measurement uncertainty in this angle is less than 2%. This
uncertainty results in the accuracy with which the beam separation can
be determined at the transmitting lens. Experience using this measurement
technique and then comparing the measurement against that obtained with
a direct microscope measurement has shown that such 2% errors are commonly
to be expected for fringe periods of the order used in this work.
The electronic system was calibrated by determining how accurately the
system could measure a signal of known visibility. The signal calibration
input was an electronically synthesized signal which could synthesize
the Gaussian shape of real signals to within 5% for the entire duration
of the signal. For the signal frequencies encountered during flight
operation, it was found that the uncertainty in the visibility was approx-
imately ±1% across the full range of the measurement system. Reference
to Figure 4 will show that a ±1% uncertainty in thp signal visibility will
79
-------
correspond to a ±1% relative uncertainty in particle diameter when the
particle size is near a fringe period in size. The size uncertainty
increases to greater than ±20% when the particle diameter is near 0.1
fringe periods in diameter. Thus, when the particle is comparable to
a fringe period in size, the predominate error results from fringe period
uncertainty. For sizes near 0.1 fringe periods, the predominate error
results from uncertainty in the visibility measurement.
Two flow calibration tests were made with the system. The first cali-
bration test was made with the instrument focused near the center of the
inlet to the small wind tunnel at Spectron's Tullahoma facility. The
rectangular test section of the tunnel is 10 x 15 cm and is capable
of speeds up to 55 m/sec. The airflow was seeded with a water fog and
chalk dust combination to simulate flight conditions on the aircraft
and to provide a wide variation in signal amplitudes. The PM system
clearly demonstrated its ability to respond to the seeded airflow across
the entire spectrum of particle sizes. The observed mode sizes were
between 25 - 30 ym. The mean velocity was 28 m/sec.
The second flow test occurred during the week following the conclusion
of the Phoenix flight test program. In this test, the PM system was
used to measure the small particle size distribution of a cloud simulation
wind tunnel at UCLA. The tunnel is designed to produce a 2-3 m/sec.
laminar airflow under highly controlled temperature conditions. The
airflow seeded with a fog having an estimated size distribution ranging
between 2 - 30 ym is used to support a 100 ym water droplet. Droplet
growth is observed as the small particle distribution droplets collide
with the larger drop. In order to cover the required particle size range,
the fringe period was changed to 38.7 ym. Impact measurements from gelatin
covered glass rods of the small particle distribution were periodically
compared with those of the PM system. Typically measured mean sizes
were 10 - 12 ym. Excellent agreement was obtained between the two sets
of measurements. It was observed, however, that too high a threshold
setting could bias the distribution toward the larger sizes. A suitable
threshold setting was found to be less than 6.8, this is not unrealistic
in view of the particle size range being examined.
80
-------
ERROR ANALYSIS
The errors associated with these measurements may be divided into two
categories - those inherent in the' PM measurement technique, and those
which arise from the way in which the PM system is applied. The
following discussion is centered around these two classifications.
INHERENT MEASUREMENT ERRORS
A visibility measurement can only be directly associated with a measure
of particle size when the shape of the particle is known. All data
evaluation presented in this paper has assumed a spherical particle shape.
Clearly, atmospheric dust particles are not spheres. Hence, the measured
visibilities represent some average particle dimension for particles with
presumably random shapes. Even if the particles have reasonably well
defined shapes (e.g. cylinders), it is to be expected that they will be
observed in all orientations so that an averaging estimate must be used
for all possible particle orientations. Such averages have not been
computed. Thus, valid estimates between the assumed diameter using a
spherical particle visibility function versus size estimates for a
randomly shaped or randomly oriented particle visibility function cannot
be made. It is clear, however, that the visibility function for a sphere
should represent the asymptotic limit for a large number of measurements
of a randomly shaped particle observed for a continuum of orientation
angles.
The equivalent diameter of the estimate should be expected to approximate
the maximum diameter of the particle. Thus, the spherical particle
visibility function is assumed to be a good approximation when a large
number of measurements occur in any one size increment in the histogram
distribution.
The calibration procedures discussed in Section VIII ensure that the PM
system accurately measures a particle even when the signal-to-noise
ratio is low. The PM system electronics are designed with logic circuits
which cause the instrument to reject erroneous or overly noisy signals.
When the signals become too noisy, the data rejection rate becomes high
with subsequently few measurements obtained by the instrument. A primary
source of noise is the shot noise generated by background light incident
on the photomultiplier tube. The background light may result from stray
laser light not associated with the signal or from broadband environmental
sources such as the sun. Laboratory tests of the PM system showed that
81
-------
the only stray laser light sources are those which result from light
scattered out of the beams transmitted to the sample space. This light
is only observable when the particle number density becomes several
hundred per cc. For the Phoenix flights, this condition did not exist.
Sunlight, however, did present a problem. A 10 nanometer interference
filter was originally used in the instrument and was found to be adequate
in laboratory simulations with bright lights. For example, a standard
3 cell flashlight could be pointed almost directly into the PM receiver
with no noticeable effect. A measure of the flashlight power output
indicated that it should produce a mean power output comparable to that
of commonly accepted sunlight figures. However, it was found that between
the hours of about 9 AM and 3 PM on clear sunny days in Phoenix, the
10 nanometer filter required instrument operation at very high threshold
levels. Such levels were found to bias the system towards large particle
measurements in the UCLA flow tests. Comparison of night flight data
where the threshold was set to a low value (A.5) to those of day flights
did not reveal significant changes in the distribution functions. From
this field application it is estimated that a filter with a 1 nanometer
bandpass should be more than adequate for nearly all sun conditions that
would be encountered with an instrument of this type. This type of
filter is a commonly produced item but could not be acquired in time
for use in the test program. A second 10 nanometer filter was obtained,
however, and used with the filter originally in the system. It is esti-
mated that the passband of the two filters was about 70 nanometers.
This filter combination proved to be sufficient for a relatively low thres-
hold during the high sun hours.
Errors were also found to result due to radiative pickup during radio
transmission while the PM was in flight. However, the system was equipped
with an amplifier to preserve signal shape fidelity between the electronics
system and the photomultiplier tube. By adjusting the capacitance in the
amplifier this source of error was eliminated.
In order for accurate measurements to be obtained, it is necessary that
the optical system maintain paraxial alignment. Since the optical system
was attached to the instrument rack which is shock mounted, no alignment
difficulties were experienced during the entire test program.
APPLICATION MEASUREMENT ERRORS
It is difficult to evaluate application measurement errors. They resulted,
in this application, from the data bias introduced by the flow field
pattern around the aircraft. It is clear from the velocity measurements
82
-------
obtained while the PM system was in flight that the sample space of the
instrument either was well within the boundary layer of the aircraft
or was shadowed by the wheel strut. The particle size distributions
obtained with the PM system show indications of being artificially sepa-
rated or sieved before they were measured. Small values of the distri-
butions' standard deviations and the relatively large values of the
mean particle size seem to make the data suspect. Furthermore, PM
measurements obtained on the ground prior to takeoff seem to indicate
about the same mean particle size as the airborne measurements but the
ground based measurements reflect higher concentration of smaller
particles than the airborne measurements. This effect is opposite to
that generally thought to exist, i.e. the airborne measurements should
show an increased number of small particles, since the larger particles
would be expected to fall out while the smaller ones would remain aloft.
This was generally observed to be the case on the test flight flown
November 19 and in histograms obtained on November 22. However, on
subsequent flights, it was found that the particles were distributed
uniformly in size except for the very shallow inversion layers which
were observed. While it is possible that a high threshold setting to
avoid background light noise biased the distribution towards the larger
particle sizes, the possible sieving effects of the aircraft flow field
should not be discounted. U. S. Air Force studies have shown that an
aircraft boundary layer tends to cause particle size separation to
distances as large as 2 - 3 fuselage diameters. These studies were
for particles greater than 50 ym in diameter and for aircraft flying
at speeds of about 120 knots at relatively high altitudes. While the
aircraft in this case flew at significantly slower speeds and altitudes,
the data obtained by the PM system strongly suggests that a careful
evaluation of such effects is in order.
An alternative explanation of the data differences in the inflight and
ground level measurements is suggested by assuming that a general sus-
pension of larger particles exists over the county from farms and dirt
roads. The particle distribution at near ground level altitudes is then
probably governed by auto traffic on paved roads. The grinding action
of tires reduces the mean value of the general suspension and broadens
the distribution.
83
-------
REFERENCES
1. Y. Yeh and H. Z. Cummins, Appl. Phys. Letters 4^ 126 (1964).
2. D. B. Brayton and W. H. Goethert, Trans. Inst. Soc. Am.,
10, 41 (1971).
3. R. S. Goldstein and D. K. Kried, J. Appl. Mech. , (E) 4L,
813 (1967).
4. F. Durst and J. H. Whitelaw, Proc. Roy. Soc. Londa A.,
_324, 157 (1971).
5. W. M. Farmer and J. 0. Hornkohl, Appl. Optics 12, 2636
(1973).
6. W. M. Farmer, "The Interferometric Observation of Dynamic
Particle Size, Velocity and Number Density" Ph.D. Thesis,
(College of Liberal Arts, The University of Tennessee, 1973).
7. F. Durst, "Development and Application of Optical Anemometers,"
Ph.D. Thesis (Heat Transfer Section, Dept. of Mech. Engr.,
Imperial College London, S.W.I., 1972).
8. R. M. Fristrom, et.al., Faraday Symposia Of The Chemical
Soc., No. 7, 183 (1973).
9. W. M. Farmer and D. B. Brayton, Appl. Opt., 10, 2319
(1971).
10. F. L. Crosswy and J. 0. Hornkohl, Rev. Sci. Instrum., 44,
1324, (1973).
11. W. M. Farmer, Appl. Opt. 13^ 610 (1974).
12. J. D. Trolinger, "Laser Instrumentation For Flow Field
Diagnostics" AGARDograph No. 186. (1974).
85
-------
APPENDIX. Daily Log
The following is a daily log kept by Spectron personnel of the day to
day operation during the EPA particle morphokinetometer test program.
NOVEMBER 17
The system equipment arrived in Phoenix and was picked up at 1500 hours.
Jeff van Ee indicated in a prior discussion that he would not want to
have the equipment placed on the helicopter that day since a mission was
being flown during the afternoon. The equipment was uncrated and set up
at the motel and appeared to work properly and to generate signals from
dust in the air.
NOVEMBER 18
The equipment was taken to the helicopter at 0800 hours. An attempt
was made to fit the optical system to the helicopter with the bracket
built for it by the EPA/Las Vegas. The bracket material was found to
be too flimsy to support the optical system. The helicopter was sche-
duled to fly a mission beginning at 1100 hours and returning at 1300
hours. During this time, 2.5cm x 2.5cm x 0.31cm aluminum angle was
purchased at a hardware store and a more substantial bracket was built
by Spectron using the EPA bracket as a pattern. At 1330 hours, the
optical system was mounted on the helicopter using the new bracket. It
was found to be sufficient. The electronics system (except for the
oscilloscope) was strapped in place on the instrument rack and connected,
Jeff van Ee was instructed an hour to program the histogram generator
of the PMAC 554 and the system was given an initial shakedown flight
to discover any mechanical or electronic defects. This flight lasted
from 1515 hours until 1800 hours. Upon return, all systems appeared
to be functioning normally. Some data printout was obtained. However,
it was not considered reliable since no oscilloscope observations were
obtained to determine the threshold necessary to limit noise from
background light.
87
-------
NOVEMBER 19
The optical system was reinstalled on the aircraft and optical alignment
checked. The oscilloscope was rack-mounted at this time. A system
check indicated a high level of noise due to background light. A 10
nanometer interference filter was installed on the front of the PMT
slit aperture. The filter appeared to be sufficient for high threshold
operation of the PMAC 554. The digital recorder began to print in
a sporadic manner. The Costa Mesa office was contacted in an attempt
to locate a replacement recorder after all attempts to find one in
Phoenix failed. In order to verify that all the other systems functioned
properly, Hornkohl and Farmer took a test flight with the pilot, Don
Nail, around the edge of Phoenix in the Chandler area. The PMAC 554
appeared to work well. Hornkohl set a maximum threshold level and was
still able to obtain very high quality signals. Direct sunlight shining
towards the optical system provided too great a background for operation.
However, with the helicopter oriented such that at least a 45 degree
zenith angle relative to the optical axis was maintained, the system
functioned acceptably well. During this flight, typical data rates were
10 to 200 acquired measurements per second. Passes were made across
an agricultural area next to the desert. There appeared to be a consid-
erable amount of dust above some of the agricultural fields which were
flown over. Visual observation of the digital displays indicated expected
data trends (e.g. smaller particle sizes were observed at a 152 m. altitude
than at a 30.5 m. altitude). In the area next to the desert, three
passes were made - one at 152 m., one at 91.5 m., and one at 30.5 m.
Data was also generated on the return leg of the test flight. The altitude
at this time was 152 m. and flight speed was 42 m/sec (80 knots). Even
though the printer output was sporadic some usable data was obtained.
NOVEMBER 20
A Hewlett Packard Model 1055A recorder was received from the Costa Mesa
office. It was tested with the electronics and found to function properly.
NOVEMBER 21
The optical system was reinstalled on the helicopter at 0615 hours.
Temperature was 7.2°C. The optical system had been left in the trunk
of the rental car all night and was too cold to start (laser would not lase)
88
-------
Also it was found that the digital recorder printed incorrect digits
from the histogram generator. The laser power supply was left on and
the system flown as was in the hope that the laser supply would suffi-
ciently warm the optics box to make the laser start. Upon return from
the early morning flight, both the laser and the printer functioned.
A second test flight was flown during the afternoon beginning at 1500
hours with Hornkohl and Farmer on board the helicopter. Immediately
after the system became airborne, Hornkohl could not observe signals
on the oscilloscope. The helicopter was returned to the airport and
system alignment checked. It was found to be satisfactory. Hornkohl
changed the capacitance in the PMT tube signal output amplifier to
eliminate occasional circuit oscillations induced by radio transmission.
Flight resumed at 1540 hours. Some initial passes were made over fields
in the Chandler area which were being cultivated. The observed data
rate was very low. Approximately 1/3 of the flight path 11 was flown
in a 10 second autosequence mode. The data rate was so low that the
autosequence mode was changed to a 30 second sample time. Additional
passes were made over other agricultural fields in this autosequence
mode. The observed data rate was at least a factor of 10 less than
that observed on the 19th. However, the atmospheric visibility was
considerably higher this date than on the 19th. Flight returned to
the airport at 1630 hours. The optical system was returned to the
motel and the transmission alignment was checked since it could possibly
have been changed in attempting to start the laser in the early morning
hours. Paraxial alignment was found to be about 5 degrees off. It is
also possible that the position of the light filter was such that the
bandpass of the filter greatly attenuated the amplitude from the small
particles. However, rotational tests of the filter in the laser beam
indicated that this was probably not the case.
NOVEMBER 22
The optical system was installed on the aircraft at 1100 hours. By
AC coupling the oscilloscope, it was determined that the system was
responding to airborne particles but that background light induced by
the high sun angle was inducing an exceptionally high noise level. The
only signals that could be observed were those from exceptionally large
particles. An attempt was made to reduce the background light by using
additional red celluloid with the 10 nanometer interference filter.
The mission lasted from 1200 hours to 1500 hours. The data recorded in
almost all cases fell in the greater than 20 - 25 urn histogram bin.
89
-------
One histogram obtained at a 305 m. altitude appeared to have a broad
distribution. A second flight was begun at 1600 hours and returned at
1800 hours. During operation with a low sun angle, Jeff van Ee was able
to adjust the system to a much lower threshold and a high data rate
was obtained. The measurements, however, consistently fell in the
particle size bins greater than 65 ym.
NOVEMBER 23
Instruments in the aircraft were turned on at 1800 hours for an early
morning mission on November 24. The optical system was installed at
2100 hours and left on. The atmosphere was exceptionally clear (local
radio reports indicated that atmospheric visibility was 113 Km). There
appeared to be a shallow inversion layer close to the ground. The signal
threshold level was set to 8 for the initial flight. Flight return was
at approximately 0100 hours, November 24. The system recorded very little
data although it seemed to be functioning normally. Histograms taken
on the ground both before and after the flight showed a particle size
distribution usually observed on a day-to-day basis. The system was
left on for the 0400 November 24 mission.
NOVEMBER 24
Flight takeoff was at 0400 hours. Farmer rode as co-pilot to ask van Ee
to vary different system parameters tn order to discover any possible
system malfunctions. During the first data pass, various thresholds
were attempted with essentially identical results. A threshold level
4.5 was chosen because such a level was found to be acceptable in all
calibration tests and did not appear to induce any data bias. During
the flight, the observed data rate was again exceptionally low compared
to that of November 19. On the return flight to the airport, the pilot
was requested to descend at a rate of 76 m. per minute from an altitude
of 366 m. van Ee was requested to set the PMAC 554 at autosequence
mode and to sample every 10 seconds. Visual observation of the acquired
data rate indicated an inversion layer somewhere between 30.5 and 15 m.
90
-------
NOVEMBER 25
An additional 10 nanometer interference filter was received from the
Costa Mesa office. With both filters used together it was estimated
that a passband of 70 nanometers could be received with about 25%
transmission. Both of these filters were used in the system. To
compensate for the reduced transmission losses, the sample space of
the optical system was moved closer to the receiver window (28.6 cm).
The sample space appeared to sit immediately behind the wheel strut.
Takeoff was at 0700 hours and return was at 0930 hours. The threshold
was set to about 9.2. van Ee reported observing good signals on the
oscilloscope and numerous counts (typically 10,000) were recorded in
the histogram. The second mission takeoff was at 1030 hours, van Ee
was requested to obtain a velocity histogram and to record some data
in a 10 second autosequence upon descent. Upon return, the aircraft
had to abort the straight-in descent and make a circular pass about
the airport. Aircraft return was at approximately 1230 hours. The
PM system appeared to generate considerable data. There was a slight
overcast due to high cirrus clouds which also probably significantly
contributed to the reduction of background sunlight. The equipment
was packed at 1300 hours and sent to shipping.
91
-------
TECHNICAL REPORT DATA
(/'lease read lu^inictions on the reverse before completing)
i HI PORT NO
EPA-600/3-76-087
4 Til LL AND SUBTITLE
ENVIRONMENTAL AEROSOL MEASUREMENTS USING AN
AIRBORNE PARTICLE MORPHOKINETOMETER
6. PERFORMING ORGANIZATION CODE
7 AUTHOR(S)
W.M. Farmer and ,T.O. Hornkohl
9 PL R> OHMINti ORGANIZATION NAME AND ADDRESS
Spectron Development Laboratories, Inc.
P.O. Box 861
Tullahoma, TN 37388
SPONSORING AC,I NCY NAME AND ADDHI SS
Environmental Sciences Research Laboratory
(Iffice of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
8. PERFORMING ORGANIZATION REPORT NO
3 RECIPIE NT'S ACCESSI Ol> NO.
5 REPORT DATE
August 1976
10. PROGRAM ELEMENT NO.
1AA603
11. CONTRACT/GRANT NO.
DA-6-99-2294A
13. TYPE OF REPORT AND PERIOD COVERED
...Final 11/75_ - 2/76
14. SPONSORING AGENCY CODE
EPA-ORD
I Ml NT AM Y NOTES
16 ABSTRACT
Measurements of ambient aerosols using an airborne particle morphokineto-
meter are described. The measurements were of large particles (greater than 5
micrometers in diameter) in environmental aerosols around Phoenix, Arizona, during
November 1975. Speed of the sample space on the airborne platform and time resolved
measurements of relative particle number density and particle size distributions
for horizontal and vertical flight profiles were obtained. The results show that
the large aerosol particles were predominantly 50-75 micrometers in size, that
the relative number density could vary greatly in space and time, and that the
largest concentration of particles were within 28 meters of ground level.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
*Air Pollution
*Aerosols
*Particle size distribution
Velocity measurement
^Interferometers
*Lasers
b.IDENTIFIERS/OPEN ENDED TERMS
Phoenix, AZ
c. COSATI l-'icld/Group
13B
07D
14B
2 OF
20E
1'i DISTRIBUTION STATLME NT
RELEASE TO PUBLIC
19 SECURITY CLASS ('I'llix He/ton)
UNCLASSIFIED
i>O SE CURIT Y CLASS ("/'/in /<«.V
------- |