Atmospheric Science Master of Science Thesis

To the University of Wyoming:
The members of the Committee approve the thesis of Jason Sulskis presented on
01/26/2016:
Jeff French
Chairperson
Dave Leon
Co-chair
John Pierre
External Committee Member
Jefferson Snider
APPROVED:
Tom Parish, Department Chair, Department of Atmospheric Science
Steve Barret, Associate Dean, College of Engineering and Applied Science

1
Sulskis, Jason A., A Comparison and Survey of the Measured Cloud Liquid Water Content
and An Analysis Of The Bimodal Droplet Spectra Observed During COPE-MED., M.S.,
Department of Atmospheric Science, June 2016.
The primary objective of the COnvective Precipitation Experiment – Microphysics
and Entrainment Dependencies (COPE-MED) was part of a larger field campaign
undertaken during July and August 2013 with the primary goal of improving quantitative
precipitation forecasts for summertime convection over SW England, with a special
emphasis on understanding microphysical processes that impact hydrometeor
development. Understanding the interplay between the warm rain and ice processes is
necessary to lead to better parameterizations for precipitation rates in numerical
simulations so, to that end, a detailed survey of the liquid water content and total cloud
droplet number concentrations measured during COPE-MED is undertaken. Additionally,
a probe-by-probe comparison of the liquid water content was performed in order to
ascertain their relative performance and consistency during COPE-MED and under certain
conditions. These comparisons reveal generally good agreement between the in situ probes
used during COPE-MED, but also reveals that there may be potential issues with certain
probes under certain conditions.
Secondly, observations from the University of Wyoming King Air research aircraft
show occurrences of bimodal cloud droplet spectra, where there exist two distinct droplet
diameter populations. An analysis of several COPE-MED cases, based on observations
from in situ cloud microphysical probes, is presented. Several environmental factors are
examined to look for evidence of entrainment events within regions containing bimodal
spectra. Correlations between the adiabaticity and concentration in each mode are
examined. While some of these analyses indicate evidence of entrainment, others are less
clear. The theoretical super-saturation a parcel would experience when neglecting the small
mode and the updraft speed required to achieve various levels of super-saturation are also
calculated. Initial results show evidence that secondary activation could potentially explain
the observed bimodal spectra, however, further numerical modelling studies to determine
the relative importance of secondary activation on the development of the bimodal droplet
spectra observed during COPE-MED will be required.

A COMPARISON AND SURVEY OF THE MEASURED CLOUD LIQUID WATER
CONTENT AND AN ANALYSIS OF THE BIMODAL DROPLET SPECTRA
OBSERVED DURING COPE-MED.
by
Jason Alan Sulskis
A thesis submitted to the Department of Atmospheric Science
and the University of Wyoming
in partial fulfillment of the requirements
for the degree of
MASTER OF SCIENCE
in
ATMOSPHERIC SCIENCE
Laramie, Wyoming
May, 2016

iii
DEDICATION
For my Dad, who has always encouraged and motivated me to pursue my passions
in science from a young age. The moment he bought me a children’s book on Physics I
was hooked and never looked back.

iv
ACKNOWLEDEMENTS
First and foremost, I’d like to thank the members of my committee. Especially, and
specifically, Dr. Jeff French and Dr. Dave Leon for their guidance, support, and patience
with me throughout this process. I was a terrible scientific writer when this all began,
despite already having written an ad-hoc thesis for a Masters in Physics not a decade before.
After this I feel that my writing has improved by at least an order of magnitude. I’d also
like to acknowledge Dr. Bobby Jackson’s assistance in the implementation of his algorithm
used in Chapter 7, as well as his suggestions and assistance in other areas of my research.
I owe a special thanks to Dr. Jeff Snider, who has been – by far – one of the most influential
and best instructors I have had to date. He made theoretical microphysics fun to learn
through his ability to present the material in a way that was very accessible and relevant to
the real world.
I’d like to thank all of my classmates, especially Dan Welsh and Rebecca Pauly,
who helped me a great deal in both my coursework and in particular aspects of this thesis.
Not the least of which was keeping me sane and offering encouragement to stay the course
and never give up. Dan was particularly instrumental in the development of Chapter 3
having had more experience with the in situ probes described therein. Before I set out
writing that chapter I had little knowledge in their design and operation and he pointed me
in the right direction. Rebecca was particularly helpful with much of the NCL scripting and
IDL coding used in many of the figures presented herein. She is a much better coder in IDL
than I.

v
I’d be remiss if I didn’t thank my family and friends, without whose unwavering
support, I would have never made it through my second attempt at graduate school. I need
to particularly thank my roommate Anthony Stover in particular for putting up with me
during my stay here in Laramie and without whom I’d never would have been able to move
to Wyoming to begin this research in the first place. I need to particularly thank my father
who thinks I am way smarter than I actually am.
Finally, I need to acknowledge that this work was funded by the National Science
Foundation under grant #AGS-1230203. I also need to point out that COPE-MED had a
great many researchers, technicians, and flight crew involved – from multiple institutions
– that are too numerous to name, but still had an impact on this research endeavor though
their hard work and dedication. I have to particularly acknowledge Dr. Sonia Lasher-Trapp,
Dr. Alan Blyth, and Dr. Alexi Korolev who, through correspondence with my advisors,
helped me get a better handle on the direction of my research.

vi
TABLE OF CONTENTS
CHAPTER 1: COPE-MED PROJECT BACKGROUND............................................. 1
CHAPTER 2: THE COPE-MED DATA SET............................................................... 6
2.1 RADIOSONDES ................................................................................................ 7
2.2 UWKA RESEARCH AIRCRAFT ..................................................................... 7
2.3 DATA LOCATION............................................................................................ 9
PART I.............................................................................................................................. 10
CHAPTER 3: IN SITU LWC PROBES – THEORY AND LIMITATIONS.............. 11
3.1 INTRODUCTION ............................................................................................ 11
3.2 HEATED ELEMENT PROBES....................................................................... 11
3.2.1 THE DMT LWC-100................................................................................ 15
3.2.2 THE NEVZOROV LWC/TWC PROBE.................................................. 23
3.3 OPTICAL PARTICLE COUNTER PROBES ................................................. 27
3.3.1 THE FORWARD SCATTERING SPECTROMETER PROBE.............. 28
3.3.2 CLOUD DROPLET PROBE.................................................................... 35
3.4 GERBER PVM-100A....................................................................................... 42
3.5 SUMMARY...................................................................................................... 47
CHAPTER 4: COPE-MED IN SITU LWC PROBE DATA COMPARISION .......... 50
4.1 INTRODUCTION ............................................................................................ 50
4.2 METHODOLOGY ........................................................................................... 51
4.3 RESULTS ......................................................................................................... 54
4.3.1 SEGREGATED BY RESEARCH FLIGHT NUMBER .......................... 54

vii
4.3.2 SEGREGATED BY MEASURED CONCENTRATION RANGES....... 64
4.3.3 SEGREGATED BY MEASURED MEAN DIAMETER RANGES ....... 73
4.3.4 SEGREGATED BY PRECIPITATION CONCENTRATION................ 80
4.3.5 BIFURCATED DATA POINTS .............................................................. 87
4.4 SUMMARY...................................................................................................... 88
CHAPTER 5: A SHORT COPE-MED LWC AND N SURVEY................................ 91
5.1 INTRODUCTION ............................................................................................ 91
5.2 METHODOLOGY ........................................................................................... 91
5.3 RESULTS ......................................................................................................... 93
5.4 SUMMARY...................................................................................................... 97
PART II............................................................................................................................. 98
CHAPTER 6: DROPLET SPECTRAL EVOLUTION ............................................... 99
6.1 INTRODUCTION ............................................................................................ 99
6.2 DROPLET GROWTH BY VAPOR DIFFUSION........................................... 99
6.3 MECHANISMS THAT EFFECT DSD SHAPE EVOLUTION.................... 101
6.3.1 ENTRAINMENT/MIXING.................................................................... 102
6.3.2 SECONDARY ACTIVATION .............................................................. 105
6.4 OTHER CONSIDERATIONS........................................................................ 106
CHAPTER 7: COPE-MED BIMODAL DSD ANALYSIS CASE STUDIES.......... 108
7.1 INTRODUCTION .......................................................................................... 108
7.2 METHODOLOGY ......................................................................................... 109
7.3 RESULTS ....................................................................................................... 115
7.3.1 CASE 1 – RF03 12:07:37 to 12:07:45 UTC.......................................... 117

viii
7.3.2 CASE 2 – RF03 12:07:37 to 12:07:45 UTC.......................................... 123
7.3.3 CASE 3 – RF09 13:49:09 to 13:49:18 UTC.......................................... 129
7.3.4 CASE 4 – RF09 13:52:36 to 13:52:48 UTC.......................................... 135
7.4 SUMMARY.................................................................................................... 141
CHAPTER 8: FUTURE WORK................................................................................ 143

ix
LIST OF FIGURES
Figure 1: Map of the southwest peninsula of England, where COPE-MED took place
during the summer of 2013. The red star icon marks the location of the ground-based
radar and mobile sounding unit in operation during COPE-MED located at
Davidstow, Cornwall, UK. The blue star icon marks the location of Exeter where the
UW King Air operated from during COPE-MED. The purple star icon marks the
location of the UK Met-Office operational radar. Inset: satellite image of a typical sea-
breeze convergence line over the southwest UK. (Map and inset satellite image
courtesy of UK Met Office)........................................................................................ 2
Figure 2: Schematic illustration of the microphysical processes and pathways active in a
warm-based convective cloud. While the figure depicts processes at specific locations
in the cloud, in most cases these processes are active to varying degrees throughout
the cloud, the notable exception being the HM process which is only active over a
limited temperature range as shown by the dashed lines. (Figure adapted from COPE-
MED project proposal, courtesy of French, et. al)...................................................... 3
Figure 3: Schematic of how a water droplet in air flow interacts with a hot-wire coil
evaporating as it enters the “heat field” around it. Droplets can either impact the
cylinder and form a thin film before evaporating or enter the “heat field” and evaporate
near the cylinder. Both interactions contribute to the heat load of the coil and lower
the temperature of the coil. The circuit responds by adding additional electrical power
to maintain a steady wire temperature (figure reproduced from Wendisch & Brenguier
2013). ........................................................................................................................ 12

x
Figure 4: Schematic of a simplified Whetstone Bridge circuit. The variable resistor is
adjusted to rebalance the circuit such that the voltmeter in the center reads zero. This
idea is used in determining the power dissipated by the evaporating liquid water
droplets as they interact with the hot-wire sensor coil.............................................. 13
Figure 5:Mechanical drawing of the DMT LWC-100 probe taken from the DMT user
manual. The LWC-100 has a mounting bracket designed for easy installation on
research aircraft......................................................................................................... 16
Figure 6: Modular circuit card containing the slave and sensing coils for the DMT LWC-
100. The card is designed to be plugged into the main probe housing and to be easily
replaceable (Adapted from DMT user manual)........................................................ 17
Figure 7: Pulse Width Modulated Wheatstone bridge circuit in the DMT LWC-100 that
allows for much less heat to be dissipated in the power FET’s that heat the master and
slave coils (adapted from the DMT LWC-100 user’s manual)................................. 17
Figure 8: Droplet effective diameter vs. the collection efficiency plotted using Nevzorov
and Shugaev (1992) for the LWC-100 probe, assuming do = 1.7 µm at nominal
aircraft speeds as per Korolev et. al. (1997). ............................................................ 20
Figure 9: Simplified schematic of air flow changes relative to the LWC-100 sensing
element (shown as the orange circle). The LWC is mounted in a rigid position and
can-not move relative to the incoming air flow........................................................ 21
Figure 10: Plot of a segment of the July 9, 2013 calibration flight that occurred mostly
outside of cloud. Where the sideslip angle increases in magnitude there is a noticible
jump in apperent LWC (~0.10 to 0.15 g m-3
) measured by the LWC-100............... 22

xi
Figure 11:Sctter plot of the Sideslip Angle vs. LWC for the LWC-100 during the same
time period as in Figure 10 above............................................................................. 22
Figure 12: Schematic of the standard design of the Nevzorov LWC/TWC probe........... 24
Figure 13: The LWC sensor's phase discrimination capability. The solid ice particles
shatter and deflect away from the wire wrapped cylinder's convex surface with
minimal heat loss (figure reproduced from Korolev, et. al. 1997). .......................... 25
Figure 14: Nevzorov Probe control circuit (adapted from Korolev, et. al., 1997)............ 26
Figure 15: Schematic of the optical configuration of the FSSP single particle scattering
probe (figure adapted from https://www.eol.ucar.edu/instruments/forward-scattering-
spectrometer-probe-model-100). .............................................................................. 29
Figure 16: Schematic of the VAC acceptance criteria of the FSSP. The green box, which
covers 62% of the beam’s diameter, represents the acceptance range. Particle A
crosses the beam close to the beam edge and its pulse (red) is shorter than the mean,
so it is rejected. Particle B is within the acceptance region and has a pulse longer than
the mean (green).Particle C is on the mean pulse region (purple) and is also accepted.
(based upon figure 1 in Coelho, et al. 2005)............................................................. 31
Figure 17: Differences in the Mie scattering curves for glass beads (those that are typically
used in the FSSP calibration procedure) and liquid water droplets for the
manufacturer’s specified scattering angle range (figure adapated from NCAR:
https://www.eol.ucar.edu/instruments/forward-scattering-spectrometer-probe-model-
100). .......................................................................................................................... 34
Figure 18: Histograms from the May 08, 2013 glass bead sizing test of the FSSP. The FSSP
looks large for the 2 bead tests. The solid grey line represents the expected diameter

xii
reported by the FSSP. The histograms show that the mean value of the bead tests are
both larger................................................................................................................. 35
Figure 19: Representation of the CDP’s single-mode laser intensity profile. The beam is an
elliptical Gaussian, which will have a profile similar to the plot shown. However, this
plot is an idealization and the real beam profile isn’t as symmetric as shown here
because the beam diverges in the two perpendicular directions............................... 36
Figure 20: Scale representation of the mechanical and optical set-up within the housing of
the CDP (DMT CDP User’s manual). ...................................................................... 37
Figure 21: (a) Signals when the drop is in the sample area. (b) Signals when the drop is
outside the sample area. (Lance, et. al., 2010).......................................................... 37
Figure 22: Schematic of the CDP bins. In order to account for the Mie ambiguities the
UWKA research group combines certain bins, namely the 9 µm and 10 µm, 11 µm
and 12 µm, and 13 µm and 14 µm diameter bins. The bin diameters shown here are
the bin edges. ............................................................................................................ 39
Figure 23: Absolute value of the percent difference of the derived N of the un-corrected
CDP with the pin-hole qualifier mask to the N derived from the algorithmically
corrected FSSP DSD data. For low precipitation cases, like RF03, the CDP only
undercounts, on average, ~10% for N greater than 500 cm-3 (as shown by the red
line). .......................................................................................................................... 40
Figure 24: These glass bead sizing tests of the CDP, for 10 different diameters of glass
beads show that, on average, the CDP is sizing correctly. As before, the solid grey
lines indicate the expected value for that diameter of glass bead............................. 42

xiii
Figure 25: Schematic of the optical set-up of the Gerber probe. Droplets enter the field of
the laser beam and scatter the light in a near-forward direction. The light is focused
into a beam splitter by the lens and is passed through transmission filters in order to
determine LWC and PSA (figure adapted from Gerber, et. al. 1994). ..................... 45
Figure 26: PVM response vs. Mie theory. (Figure adapted from U.S. Patent #4597666).46
Figure 27: Example scatter plot of the LWC measured by the CDP vs. the LWC measured
by the First Nevzorov LWC probe. The data are from RF09. The LWC-100 shows a
roll-off effect – indicated by the blue line – at ~ 1.8 g/m3 – marked by the dashed
red line – and the Nevzorov shows roll-off at ~ 1.3 g/m3or so. The Nevzorov shows
the roll-off effect at much lower LWC, and it is more pronounced and noticeable, than
that of the LWC-100. Other research flights displayed similar behavior and the values
of LWC where a roll-off occurred, on average, was about the same........................ 52
Figure 28: Plot of the percent differences, as summarized in Table 4. The error bars for
each probe are calculated from the standard deviation of the percent differences for
that probe over all of the research flights.................................................................. 56
Figure 29: Comparison of the two Nevzorov probe LWC sensor coils measured during all
COPE-MED IOPs of interest. They agree to within ~4%. The 1:1 line is shown as the
dashed line. ............................................................................................................... 57
Figure 30: Scatter plots of the CDP LWC data vs. the LWC-100 LWC data segregated by
RF. The 1:1 line is shown as the dashed line. Regression lines, shown in solid black,
are calculated in between threshold values of LWC of 0.02 g m-3 and 1.8 g m-3
using the M-estimator method. The linear regression equations, R values, and number
of data points are also shown for each plot. NOTE: The two research flights RF04 and

xiv
RF15, are not included due to the probe being inoperative for all or part of those
flights. ....................................................................................................................... 59
Figure 31: Scatter plots of the CDP LWC data vs. the Nevzorov LWC data segregated by
RF. The 1:1 line is shown as the dashed line. Regression lines, shown in solid black,
are calculated in between threshold values of LWC of 0.02 g m-3 and 1.3 g m-3
using the M-estimator method described. The linear regression equations, R values,
and number of data points are also shown for each plot........................................... 61
Figure 32: Scatter plots of the CDP LWC data vs. the PVM LWC data segregated by RF.
The 1:1 line is shown as the dashed line. Regression lines, shown in solid black, are
calculated above threshold values of LWC of 0.02 g m-3 using the M-estimator
method described. The linear regression equations, R values, and number of data
points are also shown for each plot........................................................................... 63
Figure 33: Plot of the percent differences as summarized in Table 8. The error bars for each
probe are calculated from the standard deviation of the percent differences for that
probe over all of the N ranges................................................................................... 65
N ranges. The 1:1 line is shown as the dashed line. Regression lines, shown in solid
black, are calculated in between values of LWC of 0.02 g m-3 and 1.8 g m-3 using
the M-estimator method described. The linear regression equations, R values, and
number of data points are also shown for each plot.................................................. 68
Figure 35: Scatter plots of the CDP LWC data vs. the Nevzorov data segregated by N
ranges. The 1:1 line is shown as the dashed line. Regression lines, shown in solid
black, are calculated in between values of LWC of 0.02 g m-3 and 1.3 g m-3 using

xv
the M-estimator method described. The linear regression equations, R values, and
number of data points are also shown for each plot.................................................. 70
Figure 36: Scatter plots of the CDP LWC data vs. the PVM probe’s LWC data segregated
by N ranges. The 1:1 line is shown as the dashed line. Regression lines, shown in solid
black, are calculated above values of LWC of 0.02 g m-3 using the M-estimator
method described. The linear regression equations, R values, and number of data
points are also shown for each plot........................................................................... 72
Figure 37: Plot of the percent differences as summarized in Table 12 . Error bars are given
by the standard deviation of each probe’s % differences when compared to the CDP
when segregated by mean diameter range. ............................................................... 74
mean diameter ranges. The 1:1 line is shown as the dashed line. Regression lines,
shown in solid black, are calculated in between values of LWC of 0.02 g m-3 and
1.8 g m-3 using the M-estimator method described. The linear regression equations,
R values, and number of data points are also shown for each plot........................... 76
mean diameter ranges. The 1:1 line is shown as the dashed line. Regression lines,
shown in solid black, are calculated in between values of LWC of 0.02 g m-3 and
1.3 g m-3 using the M-estimator method described. The linear regression equations,
R values, and number of data points are also shown for each plot........................... 78
by mean diameter ranges. The 1:1 line is shown as the dashed line. Regression lines,
shown in solid black, are calculated above values of LWC of 0.02 g m-3 using the M-

xvi
estimator method described. The linear regression equations, R values, and number of
data points are also shown for each plot................................................................... 79
Figure 41: Plot of the percent differences as summarized in Table 17. Error bars are given
by the standard deviation of each probe’s % differences when compared to the CDP
when segregated by precipitation range.................................................................... 82
precipitation ranges. The 1:1 line is shown as the dashed line. Regression lines, shown
in solid black, are calculated in between values of LWC of 0.02 g m-3 and 1.8 g m-3
precipitation ranges. The 1:1 line is shown as the dashed line. Regression lines, shown
in solid black, are calculated in between values of LWC of 0.02 g m-3 and 1.3 g m-3
by precipitation ranges. The 1:1 line is shown as the dashed line. Regression lines,
shown in solid black, are calculated above values of LWC of 0.02 g m-3 using the M-
estimator method described. The linear regression equations, R values, and number of
data points are also shown for each plot................................................................... 86
Figure 45: Vertical profile statistics of LWC derived from CDP spectral measurements.
Statistics are from data that is averaged for 100 m vertical levels from 0 km to 6 km.
Mean values are shown by the cyan dots, median the blue dots, and maximums the

xvii
black dots. The adiabatic LWC – calculated based upon the cloud base conditions
given in Table 20 – are plotted as the solid blue curves........................................... 95
Figure 46: Vertical profile statistics of N derived from CDP spectral measurements.
Statistics are from data that is averaged for 100 m vertical levels from 0 km to 6 km.
Mean values are shown by the red circles, median the dark red stars, maximums the
black lines. N ranges for each plot are dependant on the maximum observed N for that
particular flight.......................................................................................................... 96
Figure 47: Example of how the algorithm would indicate a mode boundary for typical
bimodal DSD due to the changes of curvature and calculate a spectral ratio, . Plot (a)
is an example of a unimodal spectra with a broad tail into the smaller diameters, (b)
is an example of a DSD with a SR < 1 and more population in the smaller mode
compared to the larger, (c) is an example of a DSD with a SR > 1 and more
population in the larger mode than the smaller, and finally (d) is an example of a DSD
with a SR ≫ 1. ........................................................................................................ 111
Figure 48: Example of a typical “mixing diagram.” The total concentration is plotted
against the MVD for the penetration. When the line is horizontal, complete
inhomogeneous mixing is implied. Where the line is sloped, a more homogeneous
mixing process is implied. ...................................................................................... 115
Figure 49: WCR data images of reflectivity and vertical velocity for Case 1: RF03
penetration from 12:07:37 to 12:07:45 UTC. The actual penetration examined is
marked by the dashed lines. The UWKA flight track is depicted by the solid black
line........................................................................................................................... 118

xviii
Figure 50: CDP Spectrum, with Spectral Ratio, plotted for Case 1: RF03 penetration from
12:07:37 to 12:07:45 UTC...................................................................................... 119
Figure 51: Plots of the (a) 25 Hz Updraft velocity, (b) 10Hz Total N derived from the CDP
DSD, and (c) 10Hz LWC derived from the CDP DSD for Case 1: RF03 penetration
from 12:07:36 to 12:07:46 UTC. Sections of the penetration where the algorithm
flagged the DSD as bimodal are marked by the thick purple lines......................... 120
Figure 52: (a) Theoretical supersaturation given the total N (blue solid curve) and the N
ignoring the small-diameter mode N (red dashed curve) and (b) Theoretical required
updrafts to reach 0.05%, 0.1% and 0.15% supersaturation levels for Case 1: RF03
penetration from 12:07:36 to 12:07:46 UTC. The green band between the green
dashed lines represents the approximate range of observed in-situ updraft maximums
for this case. The purple lines on the abscissa mark locations where the DSD has been
flagged as bimodal.................................................................................................. 121
Figure 53: Mixing diagrams for Case 1: RF03 penetration from 12:07:36 to 12:07:46 UTC.
LWC, extinction coefficient, and MVD is plotted against the N from the CDP. Solid
lines represent the median and ±95%.................................................................... 122
Figure 54: Scatter plot of the large mode (blue) and the small mode (red) as a function of
cloud adiabaticity for Case 1: RF03 penetration from 12:07:36 to 12:07:46 UTC.
Outlier points below 20% adiabaticity were ignored for the regression analysis... 123
line........................................................................................................................... 124

xix
13:46:44 to 13:46:52 UTC...................................................................................... 125
LWC, extinction coefficient, and MVD is plotted against the N from the CDP.... 128
cloud adiabaticity for Case 2: RF03 penetration from 13:46:44 to 13:46:52 UTC .
................................................................................................................................. 129
line........................................................................................................................... 130

xx
13:49:09 to 13:49:18 UTC...................................................................................... 131
cloud adiabaticity for Case 3: RF09 penetration from 13:49:09 to 13:49:18 UTC.135
line........................................................................................................................... 136
13:52:36 to 13:52:49 UTC...................................................................................... 137

xxi
cloud adiabaticity for Case 4: RF09 penetration from 13:52:36 to 13:52:49 UTC.141

xxii
LIST OF TABLES
Table 1: Table of instruments and platforms in use during the COPE-MED research flights
of interest relevant to this work. ................................................................................. 6
Table 2: Magnitudes of baseline drift of the LWC in the Nevzorov LWC/TWC probe due
to various sources...................................................................................................... 27
Table 3: Summary of in situ LWC probes in use during COPE-MED and their
methodologies, LWC ranges, and limitations........................................................... 49
Table 4: Table of comparisons of the percent differences between the LWC-100 LWC data
vs. the CDP LWC segregated by a given UWKA research flight number during
COPE-MED. The error for each probe is calculated from the standard deviation of all
of the percent differences for that probe when binned by research flight. The FSSP is
not included due to the oversizing issues discussed in Chapter 3............................. 55
Table 5: Table of comparisons of the regression line slope for measured CDP probe LWC
data vs. the LWC-100 hot wire probe’s measured LWC data for a given UWKA
research flight number during COPE-MED. Here n is the number of points, and R is
the Pearson product-moment correlation coefficient................................................ 58
data vs. the Nevzorov LWC probe’s measured LWC data for a given UWKA research
flight number during COPE-MED. Here n is the number of points, and R is the
Pearson product-moment correlation coefficient...................................................... 60
data vs. the PVM probe’s measured LWC data for a given UWKA research flight

xxiii
number during COPE-MED. Here n is the number of points, R is the Pearson product-
moment correlation coefficient, and σPVM is the standard deviation of the PVM LWC
data............................................................................................................................ 62
data vs. each of the other probe’s measured LWC data for a given N range during
COPE-MED. The error for each probe is calculated from the standard deviation of all
of the percent differences for that probe when binned by N ranges. The FSSP is not
included due to the oversizing issues discussed in Chapter 3................................... 65
data vs. the LWC-100 hot wire probe’s measured LWC data for a given N range
during COPE-MED. Here n is the number of points, and R is the Pearson product-
moment correlation coefficient................................................................................. 67
data vs. the Nevzorov probe’s measured LWC data for a given N range during COPE-
MED. Here n is the number of points, and R is the Pearson product-moment
correlation coefficient............................................................................................... 69
data vs. PVM probe’s probe’s measured LWC data for a given N range during COPE-
MED. Here n is the number of points, and R is the Pearson product-moment
data vs. each of the other probe’s measured LWC data for a given mean diameter
range during COPE-MED. The error for each probe is calculated from the standard

xxiv
deviation of all of the percent differences for that probe when binned by mean
diameter ranges. The FSSP is not included due to the oversizing issues discussed in
Chapter 3................................................................................................................... 73
data vs. the LWC-100 hot wire probe’s measured LWC data for a given mean diameter
range during COPE-MED. Here n is the number of points, and R is the Pearson
product-moment correlation coefficient. .................................................................. 75
data vs. the Nevzorov’s measured LWC data for a given mean diameter range during
COPE-MED. Here n is the number of points, and R is the Pearson product-moment
data vs. the PVM probe’s measured LWC data for a given mean diameter range during
data vs. each of the other probe’s measured LWC data for a given precipitation
concentration range during COPE-MED. The error for each probe is calculated from
the standard deviation of all of the percent differences for that probe when binned by
precipitation size drop concentration ranges. The FSSP is not included due to the
oversizing issues discussed in Chapter 3. ................................................................. 81
data vs. the LWC-100 hotwire probe’s LWC data for a given precipitation range

xxv
during COPE-MED. Here n is the number of points, and R is the Pearson product-
moment correlation coefficient................................................................................. 83
data vs. the Nevzorov’s LWC data for a given precipitation concentration range during
data vs. the PVM probe’s measured LWC data for a given precipitation concentration
range during COPE-MED. Here n is the number of points, R is the Pearson product-
moment correlation coefficient, and σPVM is the standard deviation of the PVM LWC
data............................................................................................................................ 86
Table 20: Range of cloud base conditions calculated using soundings and UWKA aircraft
data............................................................................................................................ 93
Table 21: Summary of field note highlights and results of the vertical profiles of LWC and
N plotted in Figure 45 and Figure 46........................................................................ 94
Table 22: UWKA conditions for each of the following penetrations. Distance is dependent
on the cloud penetrated since it is calculated from the average UWKA true airspeed
and time of penetration. The temperature and altitude are also all averages over the
time for the entire penetration and are rounded. Distance from cloud top is estimated
from WCR data....................................................................................................... 116
Table 23: Summary of results for the four penetrations examined................................. 142

1
CHAPTER 1: COPE-MED PROJECT BACKGROUND
The duration and intensity of precipitation, and the spatial and temporal
organization of such convective systems, is controlled by a wide range of physical
processes and their intricate interactions. These processes operate on multiple scales, from
the microscale to the mesoscale, and include processes such as the boundary-layer
transports of heat and moisture, cloud microphysical interactions with both aerosols and
moisture, and convective dynamics. If these convective clouds are able to develop
vertically with greater strength and speed they will tend to initiate locally heavy rain, often
exceeding several tens of millimeters per hour. In some circumstances these systems can
become well organized into convergence lines that, especially when slow moving, can
further increase the local accumulation of rainfall and lead to flash flooding (Leon et. al.
2015).
The COnvective Precipitation Experiment (COPE) was an international, UK led
field campaign undertaken during July and August 2013. COPE was commissioned to
improve Quantitative Precipitation Forecasts (QPF) for summertime convection over SW
England. COPE was designed to study the complete storm evolution, including the details
of convergence lines, and storm growth and persistence. Prior to COPE, there had never
been a study in this region that encompassed all aspects of storm evolution. Past studies
emphasized convective initiation but did not examine their entire life cycle, particularly the
more detailed cloud microphysical and dynamic processes of those convective clouds
(Leon et. al. 2015).

2
The goal of COPE is, therefore, that careful analysis of observational data for future
incorporation into numerical model simulations will lead to more accurate predictions. This
goal was to be achieved by utilizing a wide range of observational facilities to study
convective cloud systems in the region (Leon et. al. 2015). Figure 1 shows a map of the
southwest England, where COPE took place during the summer of 2013.
Figure 1: Map of the southwest peninsula of England, where COPE-MED took place during the summer of 2013. The
red star icon marks the location of the ground-based radar and mobile sounding unit in operation during COPE-MED
located at Davidstow, Cornwall, UK. The blue star icon marks the location of Exeter where the UW King Air operated
from during COPE-MED. The purple star icon marks the location of the UK Met-Office operational radar. Inset: satellite
image of a typical sea-breeze convergence line over the southwest UK. (Map and inset satellite image courtesy of UK
Met Office).
Mixed-phase clouds are extremely challenging to study observationally. They also
pose especially large challenges in numerical simulation work (e.g. parameterization

3
schemes). This is because mixed phase clouds have all the complexity of both warm clouds
and ice clouds individually. Additionally, precipitation formation can occur via multiple
microphysical pathways. Some of these Microphysical pathways are already well
understood, like the Bergeron-Findeisen process (Bergeron 1935; Findeisen 1938), yet
others are only beginning to be fully quantified. Figure 2 illustrates the various
microphysical processes at work in warm-based convective clouds. It shows just how
complex these processes can be.
Figure 2: Schematic illustration of the microphysical processes and pathways active in a warm-based convective cloud.
While the figure depicts processes at specific locations in the cloud, in most cases these processes are active to varying
degrees throughout the cloud, the notable exception being the HM process which is only active over a limited temperature
range as shown by the dashed lines. (Figure adapted from COPE-MED project proposal, courtesy of French, et. al).
The COnvective Precipitation Experiment – Microphysics and Entrainment
Dependencies (COPE-MED) project is a key component of COPE. The primary motivation
of COPE-MED is to investigate those microphysical pathways, and the dynamical
interactions, involved in convective precipitation formation, especially those of storm
clouds in the mid-latitudes. Each microphysical pathway (like shown in Figure 2) will

4
typically feed into, and simultaneously compete with, each other for condensate. Of
particular interest is how the relative strength of the warm rain process (the processes
involving only the liquid phase, such as collision-coalescence) directly – and indirectly
through the ice multiplication processes – impacts precipitation development. The warm
rain process is well known to have a primary role in the tropics. Part of what COPE-MED
hopes to help answer is how important the warm rain process is in heavy convective
precipitation events in mid-latitudes (Leon et. al. 2015).
Additionally, entrainment of dry air into these clouds will strongly influence their
microphysical properties and thus limit the effectiveness of the various microphysical
pathways to precipitation formation. Entrainment depletes condensate and modifies the
droplet diameters and number concentrations through either a homogeneous or
inhomogeneous mixing processes (which are explained in more detail in Chapter 6) or
both. Thus, the process of entrainment must also be considered here.
The objectives of COPE-MED are to (a) investigate and understand the interaction
between the different microphysical pathways that affect heavy convective precipitation
formation and to (b) investigate the relative sensitivity of those pathways to changes in
environmental conditions. The two main hypotheses of COPE-MED are that:
I. The formation of raindrops through the warm rain process is critical to the
development of heavy precipitation at the surface, even when ice processes are active.
II. The effects of entrainment must be mitigated by some factors in order to produce
heavy precipitation at the surface.

5
This thesis is meant to help lay the foundation for the broader objectives of COPE-
MED to be realized through continuing and future work. It is broken down into two major
but indirectly connected parts. For the first part, a descriptive statistical analysis was done
on the in situ probe data. Both probe-by-probe comparisons of each type that measures
liquid water content (LWC) – either directly or as a derived quantity – will be described to
provide a general probe performance review of the COPE-MED data set. Additionally, a
general survey of the values of LWC, N, and the environmental conditions encountered
during COPE-MED will be presented. Particularly, information about the vertical LWC
profiles will aid in improving our understanding of the microphysical processes acting to
form and maintain the convective systems observed in COPE-MED. This is an integral part
of the overarching goal of improving the representation of clouds in numerical models in
order to improve QPF.
The second part of this thesis is dedicated to an investigation and quantification of
the characteristics of the bimodal droplet spectra distributions (DSD) measured during
several, non-precipitating (or negligible precipitation) cloud penetrations. Additionally,
the environmental conditions that lead to those characteristics is analyzed in detail. How
these characteristics connect to other bulk quantities is also discussed. These connections
provide additional information about the underlying processes responsible for droplet
growth (and evaporation) and their role in the formation and production of precipitation.
This analysis shows some evidence that is consistent with entrainment/mixing and
potentially secondary activation in the penetrations discussed.

6
CHAPTER 2: THE COPE-MED DATA SET
COPE-MED resulted in a robust data set that contains observations from multiple
platforms and locations. Detailed descriptions of all of the platforms in operation, and the
measurements taken during all of the COPE-MED intense operation periods (IOP), which
are defined as when one or more of the data collection platforms were in operation, can be
found in Leon et. al. (2015).
The data relevant to this work focuses primarily on the in situ measurements
obtained from the 14 missions flown by the University of Wyoming King Air (UWKA).
Table 1 shows a breakdown of the instruments and platforms that were operational during
those flights. COPE-MED had 17 IOP, therefore, the data used herein represents a mere
subset of what is available to research scientists.
Table 1: Table of instruments and platforms in use during the COPE-MED research flights of interest relevant to this
work.
Flight Date Location UWKA GB Soundings
RF03 2013/07/10 NW Wales ü û
RF04 2013/07/18 SW England ü ü
RF06 2013/07/27 SW England ü û
RF09 2013/08/02 N. Cornwall ü ü
RF10 2013/08/03 N. Cornwall ü ü
RF12 2013/08/07 NW Wales ü ü

7
2.1 RADIOSONDES
Radiosondes are used to provide a broad thermodynamic profile of the pre-storm
environmental conditions. They are particularly useful for the estimation of cloud base
conditions and for the calculation of the adiabatic LWC used in Chapter 5 and Chapter 7.
Radiosondes were launched at regular intervals from the Cardington mobile sounding unit,
which was deployed at Davidstow (see Figure 1), during most of the UWKA missions.
Additional radiosondes were launched at the standard UK Met Office sites of Camborne
and Larkhill – which are co-located on the peninsula near the COPE-MED field campaign
domain – at the standard times of 0000 and 1200 UTC daily (Leon et. al. 2015). Where
ground based radiosondes are unavailable, sounding profiles taken by the UWKA aircraft
are used instead.
2.2 UWKA RESEARCH AIRCRAFT
The primary data source was the UWKA, a specially modified Beechcraft Super
King Air 200T. In addition to a suite of in situ probes, it carries a W-band, 95 GHz (~ 3mm
wavelength) polarimetric Doppler radar. The WCR can have up to four antennas
transmitting and receiving, providing information on the vertical distribution of both cloud
and precipitation. The radar configurations allow for information including pulse-pair,
polarimetric parameters, and full Doppler spectra to be gathered from both above and
below the aircraft with a dynamic range of more than 65 dB. During COPE-MED the WCR

8
operated with 3 antennas – near zenith, near nadir, and one forward of nadir. It collected
reflectivity and Doppler velocity data from these 3 antennas nearly simultaneously to
collect a 2-D curtain of reflectivity and near-vertical particle velocities above and below
the aircraft.
The LWC in situ measurement capabilities aboard the UWKA include
measurements from 4 in situ probes that utilize 3 different methodologies. Bulk LWC was
measured directly by the Droplet Measurement Technologies (DMT) LWC-100 hot-wire
probe and the Nevzorov LWC probe. The Nevzorov also has the capability to measure total
condensed water content (TWC), however that capability is not considered here. LWC was
also measured directly, for droplets up to ~ 50 µm, by bulk optical scattering using the
Gerber Scientific PVM probe.
Droplet diameter spectra, in the range of droplet diameters of 1.5-50 µm, were
measured using two single optical scattering probes. These included a Particle Measuring
Systems (PMS) Forward Scattering Spectrometer Probe (FSSP) and a DMT Cloud Droplet
Probe (CDP). These probes allowed for the direct measurement of droplet diameter and the
derivation of the LWC through integration of their droplet diameter spectra.
There were also probes that measured environmental conditions such as
temperature, dew point, pressure, and other standard parameters of interest installed aboard
the UWKA. Chapter 3 gives a more detailed description of the probes that directly measure
LWC or derive it from droplet diameter distributions. For more information on other in-
situ probes installed aboard the UWKA during COPE-MED, and the Wyoming Cloud
Lidar (WCL), the reader is again referred to Leon et. al. (2015).

9
2.3 DATA LOCATION
All of the COPE-MED research data are stored in the University of Wyoming
Department of Atmospheric Sciences on a Linux computer system which has been
published under DOI numbers 10.15786/M2MW2S, 10.15786/M2H598, and
10.15786/M2CC7B for the flight level data, WCR data, and WCL data respectively. As of
this writing the data may be accessed via the world wide web at
http://flights.uwyo.edu/projects/copemed13/.
All of the computer code used in analyzing the data, and creating the figures, within
this thesis are also located in the same location as the official data set, as well as on a
GitHub repository freely available on the World Wide Web via the url
https://github.com/jaysulk/COPE-MED/. All of the code used herein is released under the
GNU General Public License v3.0 for the reader to freely use and improve upon. Details
of the tools required to run the code as well as details of the license are provided within the
repository.

11
CHAPTER 3: IN SITU LWC PROBES – THEORY AND LIMITATIONS
3.1 INTRODUCTION
In situ measurements provide us with the most accurate information about cloud
characteristics, mostly on scales from less than a meter to hundreds of meters. For in cloud
measurements these instruments are often mounted on an aircraft, usually under the wings
or near the front of the fuselage in such a way that they are kept isolated from major sources
of flow distortion. In situ probes were used during COPE-MED to evaluate droplet size
distributions and LWC. They measure the LWC directly (or it is derived from a DSD),
rather than remotely as in radar, lidar, and satellite imagery, employing various methods.
The first is to use a technique to directly measure the LWC by measuring the power needed
to evaporate liquid impacting the heated element. The other common methodology is to
derive the LWC from the DSD measured by optical scattering probes. This chapter
introduces the concepts and the details of the theory of operation of these probes.
3.2 HEATED ELEMENT PROBES
Heated element, or “hot-wire” probes utilize a heated wire coil maintained at a
constant current or constant temperature (Wendisch & Brenguier 2013). Hot-wire liquid
water probes are an adaptation of the hot-wire anemometer, an instrument that measures a
fluid’s velocity by measuring the amount of heat taken away by the fluid by convection.

12
These probes respond both to liquid water and to the airflow past the sensing element
(Wendisch & Brenguier 2013).
LWC is calculated from first principles from the energy required to heat and
evaporate liquid droplets that interact with it (King et. al. 1978). Droplets will either pass
through the sensor wire’s “heat field,” or directly impinge upon it, and thus contribute to
the wire’s total heat load. Figure 3 shows an idealization of how droplets may interact with
the hot-wire coil. These interacting droplets change the wire’s temperature. The control
circuit then responds with the addition of more electrical power in order to maintain the
wire’s set operating temperature.
Figure 3: Schematic of how a water droplet in air flow interacts with a hot-wire coil evaporating as it enters the “heat
field” around it. Droplets can either impact the cylinder and form a thin film before evaporating or enter the “heat field”
and evaporate near the cylinder. Both interactions contribute to the heat load of the coil and lower the temperature of the
coil. The circuit responds by adding additional electrical power to maintain a steady wire temperature (figure reproduced
from Wendisch & Brenguier 2013).
The power dissipated by the probe wire is given by the power lost by the heat losses
due to convection, radiation, and the latent heat of vaporization:

13
P = P?@AB@CBDE + PGHIDEGCBDE + PED@JH?@CBDE. ( 1 )
The radiative losses are usually negligible so only the convective and evaporative power
losses are considered:
P = PGHIDEGCBDE + PED@JH?@CBDE. ( 2 )
A typical Wheatstone bridge type circuit (Figure 4) is used to determine the change in
resistance of the wire that occurs from the temperature change from these two processes.
This circuit – in its simplest form – consists of two known resistances, one unknown
resistance – that of the hot-wire sensor coil, and a fourth variable resistance. While the
circuit is in balance, e.g. when the variable resistance and the unknown resistance – e.g.
the hot-wire coil – is at its ambient value, the voltage drop between resistor pairs will be
zero. When this condition holds, the ratio of the two known resistors is exactly equal to the
ratio of adjusted value of variable resistance and the value of unknown resistance.
Figure 4: Schematic of a simplified Whetstone Bridge circuit. The variable resistor is adjusted to rebalance the circuit
such that the voltmeter in the center reads zero. This idea is used in determining the power dissipated by the evaporating
liquid water droplets as they interact with the hot-wire sensor coil.

14
When the wire changes its resistance due to evaporation, the circuit will go out of
balance and there will be a voltage drop between the nodes. The control circuit maintains
the hot-wire, RK, at a constant temperature by maintaining it at a constant resistance. The
resistance of the sensing coil decreases as the wire temperature decreases. This then allows
one to determine the LWC directly viz:
LWC =
OPOQ
R D STUGV WVPWX
( 3 )
where P is the measured power, PA is the convective power term (also called the dry-air
power term), and A is the sensor area. The convective heat losses are calculated by:
PA = π d l κ T^B?E α Rea
( 4 )
where κ is the thermal conductivity of the wire and is a function of the wire temperature, d
is the diameter of the element, and l is the length of the element. The quantity α times the
Reynold’s number raised to the power of β is a parameterization of the Nusselt number.
The magnitude of the power dissipated, measured by this voltage change, is thus a function
of both the LWC in the airstream and the temperature and velocity of the air that is passing
over the coil.
The earliest forms of hot-wire probes used on aircraft date back to the 1950’s. The
Johnson-Williams Liquid Water Content Probe (J-W) determines LWC by the change in
electrical current in the hot-wire coil, which is maintained at a constant voltage, as it cooled
due to the evaporation of cloud droplets (Neel 1955). However, the J-W probe also suffered
from several disadvantages, including that it required a “wet” calibration box and could
not be calibrated in dry conditions. In the late 1970’s King et. al. improved upon the design
of the J-W LWC Probe. Their design used a wire maintained at a constant temperature, set

15
by the electronics, rather than at constant voltage. Their design both allowed for the
calculation of its response characteristics from first principles but more importantly for
calibration in dry conditions. It also addressed issues with axial heat losses, due to
conduction, by adding a slave coil at each end of the master coil. The slave coils are
maintained at the same temperature as the master, sensing coil. The temperature of the wire
must be high enough to efficiently evaporate water quickly, but low enough to avoid the
formation of a vapor barrier between droplets and the wire, which would result in lower
collection efficiency (King et. al. 1978).
King et al. (1978) described the energy balance of their cylindrical heated element
with the following relationship:
P = l d v LWC LD T + c^ T^ − T@ + π l κ T^ − T@ Nu. ( 5 )
Here κ is the thermal conductivity of the wire, v is the velocity of the cloud relative to the
element, LD T is the latent heat of vaporization at the temperature of the hot wire, c^ is
the specific heat of water, T@ is the ambient air temperature, κ is the thermal conductivity
of air and Nu is the Nusselt number. Nu can be parameterized in terms of the Reynolds’s
number, airspeed, and other measured properties (King et al., 1978).
3.2.1 THE DMT LWC-100
The LWC-100 probe, manufactured by Droplet Measurement Technologies
(DMT), is a modern implementation of the design of the King probe. The basic design
remains relatively unchanged. The greatest difference between the original King probe

16
and the commercially available LWC-100 is improved electronic control systems and
software. It also has an anti-icing heating element built into the sensor strut. The LWC-
100’s sensing element is 1.8 mm in diameter and 20 mm in length, operates at ~125℃ as
set at the factory*
, and comes as a modular circuit card that can be easily swapped in and
out. The LWC-100 is mounted on a specially designed strut that can be affixed either to
the fuselage or the wing of the aircraft with the sensing element directed toward the
incoming flow. Figure 5 shows the mechanical drawing of the LWC-100 probe taken from
DMT’s user manual and Figure 6 shows a photograph of the modular circuit card for the
sensor and slave coils.
Figure 5:Mechanical drawing of the DMT LWC-100 probe taken from the DMT user manual. The LWC-100 has a
mounting bracket designed for easy installation on research aircraft.
*
Each sensor card has a slightly different resistance, so the actual operating temperature is determined through analysis
of clear air flight data.
Modular Sensor Circuit Card

17
Figure 6: Modular circuit card containing the slave and sensing coils for the DMT LWC-100. The card is designed to be
plugged into the main probe housing and to be easily replaceable (Adapted from DMT user manual).
Figure 7 shows the more advanced Wheatstone bridge circuit used in the DMT
LWC-100 probe. This particular circuit uses a pulse width modulated system of heating
the coil, thus reducing heat dissipation in the power field effect transistors (FET) that heat
the master and slave coils. This in turn leads to a lower electronics failure rate (see the
DMT LWC-100 manual for more details on this).
Figure 7: Pulse Width Modulated Wheatstone bridge circuit in the DMT LWC-100 that allows for much less heat to be
dissipated in the power FET’s that heat the master and slave coils (adapted from the DMT LWC-100 user’s manual).

18
Similar Nu versus Re parameterizations are used for evaluating the LWC with the
LWC-100 as described above for the original King probe. The LWC-100 is described by
DMT, and the University of Wyoming’s internal instrumentation documentation, to be able
to measure LWC, at 25 Hz, from 0 to 3 g mPi
. The LWC-100 has an accuracy of
0.05 g mPi
. Additionally, the manufacturer specifies operating limits for the LWC-100 at
between ±40℃, up to 40,000 ft (~12 km) MSL, and airspeeds up to 200 m s-1
.
Outside of cloud, the power consumed by the sensor wire is solely due to convective
heat loss by air flowing around the sensor wire. This term is then assumed to remain
constant for flight legs at a constant altitude and airspeed, even in cloud. The Nu versus
Re relationships are evaluated by fitting of the clear-air measurements of total power
delivered to the hot-wire, the ambient temperature, and the true airspeed of the UWKA in
a guess-and-check iterative convergence algorithm in MATLAB, developed at the
University of Wyoming by Dr. Al Rodi. The algorithm first calculates the baseline
temperature of the sensor wire at a given constant altitude out of cloud then calculates
appropriate temperature coefficients that refine the in cloud convective power term for a
given airspeed. However, there is a limitation. In order for the code to provide reasonable
temperature coefficients to determine the convective power term there needs to be a certain
amount of data points that are out of cloud. Typically, this range of out-of-cloud data points
is on the order of 30% to 40%. These ranges are somewhat arbitrary and subjective,
however, during COPE-MED each of the research flights contained enough out of cloud
data points – on the order of 65% or so – in order to obtain good calculations of the
convective term for the calculations of LWC.

19
The LWC-100’s first major limitation is in its inability to effectively measure LWC
at the extremes of droplet diameter spectra. For example, Biter at al. (1987) showed that
the King probe (and by extension the LWC-100) is limited by collection efficiency
considerations. Specifically, collection efficiencies fall below unity on the small and large
ends of the droplet diameter spectrum. On the small diameter end, for droplets with
diameter less than about 5 µm, a fraction of the droplets will tend to follow the airflow
around the the sensor & not fully evaporate so that the collection efficiency decreases more
and more from unity as the diameters decrease. This is partially compensated for by their
interaction with the “heat field” of the wire, however this will still tend to lead to an
underestimation of the LWC. Nevzorov and Shugaev (1992), showed that the collection
efficiency for small liquid droplets impacting the hotwire element is given by:
ϵ =
dEll
m
dEll
m
+ dn
m ( 6 )
where dEll and dH are parameterizations related to the droplet diameter. Korolev et. Al.
(199&) picked dH = 1.7 µm for the calculation of the hotwire element at nominal aircraft
speeds. As shown in Figure 8 this gives the collection efficiency to be greater than 95% for
sizes greater than about 5 µm.

20
Figure 8: Droplet effective diameter vs. the collection efficiency plotted using Nevzorov and Shugaev (1992) for the
LWC-100 probe, assuming op = 1.7 qr at nominal aircraft speeds as per Korolev et. al. (1997).
On the large end, as drop diameters reach hundreds of microns, drops will begin to
experience partial evaporation, particularly due to effects such as partial mass shedding
(Biter et. al. 1987). Shedding is typically an issue associated with high LWC loading, or
too large a wire temperature, but it is not necessarily associated with large hydrometeor
size unless there is a correlation between drizzle drops and LWC. Biter, et al (1987) showed
that for drizzle sized drops the response of the hotwire probes drops off from unity, and
can be nearer 50% for drops ~150 − 200 µm at 60 m sPu
. Thus, for very large drops, at
aircraft speeds, one expects the LWC-100 to tend to underestimate the LWC.
As the definition of LWC is in terms of the relevant particle size interval, e.g. all
droplets smaller than drizzle drop sizes, and the LWC-100 wire is relatively small, the
collection efficiency is essentially unity for all droplets larger than ~5 µm, particularly
since all drops in the volume swept by the wire will tend to impact. Therefore one is usually

21
only concerned with the collection efficiency issues on the small diameter end of the DSD
while assuming the collection efficiency to be unity above the drizzle drop range.
The second major limitation of the LWC-100 its inability to compensate for
changes in flow angle. Figure 9 illustrates how the airflow may be different relative to the
sensing element of the LWC-100. In the cases where the flow angle changes, the
convective power term increases but this is not accounted for in calculations so residual
power is assumed to be due to evaporation of LWC. The LWC erroneously calculated in
those cases, therefore shows a noticeable increase. This typically occurs in cases where the
sideslip angle, the angle between the forward velocity of the aircraft and the relative wind,
increases. Figure 10 shows an out-of-cloud segment of the July 9, 2013 calibration flight
that took place during COPE-MED. The jump in LWC is on the order of between 0.10 and
0.15 g mPi
for sideslip angles −5° ≤ x ≤ 5°. Figure 11 shows a scatterplot between the
aircraft sideslip and LWC showing that there is a correlation between the two. These
“jumps” in LWC have not been compensated for in the processing of the COPE-MED data
for the LWC-100.
Figure 9: Simplified schematic of air flow changes relative to the LWC-100 sensing element (shown as the orange circle).
The LWC is mounted in a rigid position and can-not move relative to the incoming air flow.

22
Figure 10: Plot of a segment of the July 9, 2013 calibration flight that occurred mostly outside of cloud. Where the sideslip
angle increases in magnitude there is a noticible jump in apperent LWC (~0.10 to 0.15 g m-3
) measured by the LWC-
100.
Figure 11:Sctter plot of the Sideslip Angle vs. LWC for the LWC-100 during the same time period as in Figure 10 above.

23
3.2.2 THE NEVZOROV LWC/TWC PROBE
The Nevzorov probe provides yet another improvement to the King type hot-wire
LWC probe. The Nevzorov Total Water Content (TWC)/Liquid Water Content Probe
(LWC) was originally developed in the Cloud Physics Laboratory of the Russian Central
Aerological Observatory in the mid-1970s (Korolev et. al. 1997). It is a hot-wire instrument
that has two main sensor wires that are held at a constant temperature. The Nevzorov
operates on the same principle as the LWC-100 but has the capability to measure both the
LWC and the total condensed water, TWC. This makes it particularly useful for
measurements in mixed phase clouds, allowing for determination of liquid relative to liquid
plus ice (phase discrimination) in those clouds. During COPE-MED, the calculation of Ice
Water Content (IWC) by subtracting the measured LWC from the measured TWC
wouldn’t work. In fact, in some cases the values of TWC were lower than the LWC due to
the uncertainties in collection efficiencies of each sensor and the offsets applied in
processing the data.
Like the LWC-100 the LWC (and the TWC) is calculable from first principles from
the energy provided to melting of ice and evaporation of liquid impacting the sensor wires.
However, unlike the LWC-100, the Nevzorov probe also has a reference wire for each
sensor type that allows for the direct measurement of the convective term. As discussed in
the previous section, this value must be calculated for the LWC-100 probe using an
iterative guess-and-check algorithm and requires more out-of-cloud data points than in
cloud.

24
The basic schematic of the Nevzorov probe is shown in Figure 12. It consists of a
movable vane that allows the sensors to pivot parallel to the incoming air flow. This helps
achieve the maximum collection efficiency and it also helps prevent droplets and ice
crystals from impacting the reference sensors providing greater accuracy of the convective
power term in cloud. The vane also typically has a leading edge, heated wire to prevent
icing of the apparatus in super-cooled clouds (Korolev et. al. 1997).
Figure 12: Schematic of the standard design of the Nevzorov LWC/TWC probe.
The Nevzorov probe hot-wire sensor wires are made of single-layer enamel-
covered nickel wire wound tightly around a solid copper rod and then cemented to the
opposite edges of a flat plastic plate (that is part of the vane assembly). The diameter of the
sensor wire and rod assembly is 1.8 mm and the length is 16 mm. The LWC sensor wire
operates in a similar manner as the LWC-100. At 90° C the sensor wire resistance ranges
between 2.5 and 3.5 Ω (Korolev et. al. 1997). The LWC sensor wire is also inherently
designed such that ice particles will impact, shatter, and split away from the sensor’s

25
convex surface, with minimal transfer of heat. Liquid water will form a thin film on the
wire and evaporate with a greater heat loss (Korolev et. al. 1997).
Figure 13: The LWC sensor's phase discrimination capability. The solid ice particles shatter and deflect away from the
wire wrapped cylinder's convex surface with minimal heat loss (figure reproduced from Korolev, et. al. 1997).
The LWC sensor wire can capture most moderately sized droplets, however it
suffers from the same limitation the LWC-100 where, as larger droplets impinge on the
sensor wire, some fraction of the liquid sheds prior to complete evaporation. The collection
efficiency of the LWC sensor wire approaches 100% for diameters between ~5 µm and
~100 µm. The estimate of LWC then decreases significantly, mostly due to shedding
effects causing partial evaporation, for drops on the large end of the DSD, just as with the
LWC-100 (Biter et. al. 1987).
Figure 14 shows the schematic of a typical Nevzorov probe electronic control
circuit. This circuit is somewhat more advanced than the LWC-100’s. Both the main
collector sensor and reference wires form two temperature-dependent legs of a wheatstone
bridge circuit with amplifiers. The adjustable resistors R’ and R” in the figure are mounted
on the control box and are used to set the temperature of each sensor wire. Laboratory

26
calibrations determine the relationship required between R’, R” and the required
temperature (Korolev et. al. 1997).
Figure 14: Nevzorov Probe control circuit (adapted from Korolev, et. al., 1997).
The calibration of the Nevzorov LWC probe requires flying flight legs at a constant
altitude in order to get a baseline reading for the reference sensors and the LWC sensing
wire. These data are then used to determine clear air corrections to the sensor wire to match
the reference wire. This was done for the COPE-MED campaign using 3 flight levels. At
each level five, straight and level runs were performed for ~1 minute at three different
flight speeds. The flight levels chosen to calibrate the Nevzorov probe for COPE-MED
were 23,000 ft, 13,000 ft, and 3,000 ft.
Under ideal conditions the LWC probe wire will measure a value of 0 g mPi
in dry,
cloud free air (Korolev et. al. 1997). However, due to the sensitivity of a hot-wire probe to
convective heat losses, this value will drift. There are several sources of drift. Drift that can
occur due to changes in airspeed, external temperature changes, changes in pressure
(altitude), and random fluctuations in clear, non-cloudy air. The random fluctuations of
straight and level flight are estimated to be ±0.002 g mPi
, which is an order of magnitude

27
less than the King probe. The internal circuitry is designed to compensate for larger random
fluctuations and Korolev, et. al. tested the other sources of drift. Table 2 quantifies the three
major sources of drift. Since each has an order of magnitude of ~10Pi
the error induced
from these is considered negligible (Korolev et. al. 1997).
Table 2: Magnitudes of baseline drift of the LWC in the Nevzorov LWC/TWC probe due to various sources.
Temperature Drift: ~0.5×10Pi
g mPi

10℃
Pressure/Altitude Drift: ~5×10Pi
g mPi
kmP{
Airspeed Drift: ~3×10Pi
g mPi
10 m sP{

The Nevzorov LWC sensor coils suffer from the same droplet diameter limitations
as the LWC-100. That is lower collection efficiency at smaller diameters and potentially
incomplete evaporation due to mass shedding at larger diameters. The collection efficiency
for the Nevzorov probe is, again, given by Equation 6, adapted from Nevzorov and
Shugaev (1992) using the values for the parameters given by Korolev et. al. (1997).
3.3 OPTICAL PARTICLE COUNTER PROBES
Optical particle counters (OPCs) infer the diameter of a particle by measuring the
intensity of light scattered by that particle from a monochromatic laser beam. Within
certain solid angular limits, the response measured by a photo detector is a direct
measurement of the particle’s diameter, assuming a mostly spherical particle and a

28
specified index of refraction. These types of in situ probes cover a broad range of particle
diameters up to several tens of microns. However, the response of the photo detectors to
monochromatic light is generally non-monotonic. Thus, identical measured responses can,
in some cases, be attributed to multiple diameters of particles. This non-monotonic
behavior is due to the Mie resonances, which are dependent on the diameters and refractive
indices of the incident spherical particles. It can also heavily depend on the optical
geometry of the probe itself. The suppression of these resonance effects may be partially
accomplished by using a multimode laser.
OPC’s are droplet-sizing instruments, not LWC-measuring probes, so rather than
measuring LWC directly, LWC is derived by summing over the diameter bin distributions.
Given a N centered at droplet diameter, DB, the LWC is derived from the third moment as
follows:
LWC =
} ~
Ä ÅÇXÉÑÖ
nB DB
iIáàáX
Bâ{ ( 7 )
where nB represents the number of particles in the ith bin, nCHC@ä represents the total number
of instrument channels (or bins), and Vu@ãJäE the sample volume, which depends on the
geometry of the laser beam, airspeed, and the sample time.
3.3.1 THE FORWARD SCATTERING SPECTROMETER PROBE
The Forward Scattering Spectrometer Probe (FSSP) (Knollenberg 1976) was
originally manufactured by Particle Measuring Systems Inc. (PMS) out of Boulder, CO
and has been one of the most widely used in situ probes for measuring cloud droplets over

29
the last few decades. The FSSP detects, counts, and diameters single particles by measuring
the intensity of the light scattered by particles passing through a hybrid HeNe laser beam
focused at the center of an inlet tube that faces into the oncoming airstream as seen in
Figure 15.
Figure 15: Schematic of the optical configuration of the FSSP single particle scattering probe (figure adapted from
https://www.eol.ucar.edu/instruments/forward-scattering-spectrometer-probe-model-100).
The FSSP laser beam has a diameter of about 200 µm at its focal point. Laser light
that is scattered in the forward direction is directed by a right angle prism positioned about
38 mm away from the particle plane though a compound condensing lens. This lens focuses
the beam into a beam splitter onto a pair of photo-detectors. The "dump spot," in Figure
15, is a beam block on the opposite side of the inlet, which prevents direct light from the
laser from entering the collection optics and corrupting the scatter signal (Pinnick et al.
1981).

30
Mie scattering theory, which assumes a plane wave scattering from spherical
particles, is used to relate the scattering intensity to the particle diameter. The response
function is obtained by integration over a scattering angle range of α to β :
R x =
λm
2 π
S∥ θ, m, x
m
+ Së θ, m, x m
sin θ dθ
a
ì
( 8 )
where λ is the wavelength of the laser, S is the parallel and perpendicular components of
the the complex scattering function, m is the complex index of refraction, x =
} î
ï
is a
dimensionless size parameter for a particle of diameter D, and θ is the scattering angle
(Barnard & Harrison 1988). Liquid particles of refractive index ~ 1.33 are also typically
assumed, since one can calibrate the probe using any reasonable value of refractive index.
The FSSP has a total collection angle ranging from about 3° − 13°. However, this range
may vary from probe to probe, depending on its alignment (Pinnick et. al. 1981). The
electronics measure both the height and duration of the pulse created by a particle passing
through the beam.
The probe’s optical depth of field (DOF) is used to select droplets crossing the
center of the beam, those close to its focal point in the middle of the sampling tube. This is
accomplished by optically masking one of the detectors and the use of real-time electronics.
The masked detector only receives scattered light when the particles pass through the laser
beam displaced greater than ~ 1.5 mm on either side of the center of beam’s focus. When
the pulse from that masked detector exceeds that from the unmasked detector the particle
is rejected as being outside of the DOF. The FSSP’s multimode laser does not have a
uniform intensity across its beam diameter and thus, particles near the edges may be
undersized.

31
Particles are also rejected by comparing their electronic pulse duration to the mean
pulse duration – the linear running average of the DOF selected pulse duration. If the pulse
duration is shorter than that mean, then the particle is rejected. If it is the same or longer,
the particle is accepted. This criteria is known as the velocity average control (VAC) and
allows for rejection of particles that pass through the beam edge as illustrated in Figure 16.
However, the VAC assumes the pulse duration only depends on the location of the particle
trajectory with respect to the beam and not on the particle diameter, making the VAC
criteria most efficient for the situation that all droplets have the same diameter
(monodisperse DSD).
Figure 16: Schematic of the VAC acceptance criteria of the FSSP. The green box, which covers 62% of the beam’s
diameter, represents the acceptance range. Particle A crosses the beam close to the beam edge and its pulse (red) is shorter
than the mean, so it is rejected. Particle B is within the acceptance region and has a pulse longer than the mean
(green).Particle C is on the mean pulse region (purple) and is also accepted. (based upon figure 1 in Coelho, et al. 2005)
Larger droplets in non-monodisperse spectra can produce statistically longer pulses
so the FSSP has a delay mode that can reduce the dependence of the pulse duration on the
droplet diameter by measuring the pulse duration at the mid-pulse amplitude (Coelho et al.
2005). Round-off errors on the shorter transit time measurement of droplets crossing the

32
beam edge also affect the VAC selection criteria since it is generally limited to 10 Hz (Dye
and Baumgardner 1984).
Due to the geometry of the probe and associated optics – roughly 80% of the
particles are rejected as being outside of the DOF. Of the remaining 20% of accepted
particles roughly half of those are rejected due to the VAC criteria. The remaining 10% of
the particles are then sized by the electronics. After sizing, the particles are categorized
into one of 15 bins, or channels. In general, the FSSP has 4 user-selectable diameter ranges.
Aside from the 2 − 47 µm range, there is also 2 − 30 µm, 1 − 15 µm, and 0.5 − 0.75 µm
ranges that can be selected. The FSSP was set to measure particles in the diameter range
2 µm to 47 µm during COPE-MED.
The diameter of the particle is determined from the pulse height by comparing it to
a predetermined set of pulse heights forming the upper limits of the 15 diameter channels
and registering a count into the appropriate channel. Counts taken over a period of time are
aggregated to generate a drop diameter distribution. However, since the laser beam
intensity is not uniform the intensity of scattered light by particles passing through the
beam is highly dependent on the path they take through the beam. This beam
inhomogeneity tends to produce broadening of the measured droplet spectrum
(Baumgardner and Spowart 1990). The accuracy of the particle by particle diameter
measurements of the FSSP is on the order of ±20%, due mostly to particle positioning
within the beam. The overall accuracy of the derived concentration by the FSSP to be on
the order of ±16% (Baumgardner 1996).
Since the liquid water content is not a measured quantity by the FSSP, but rather a
derived quantity obtained by integrating the diameter distribution measured by the FSSP,

33
uncertainties in the diameter measurements can lead to root sum squared (RSS) errors in
the liquid water content. These RSS errors can cause the derived LWC to be up to a factor
of three or higher than its true value. Baumgardner (1996) showed the accuracy of FSSP
LWC measurements fell between ±30%– 50%.
There are two fundamental measurement uncertainties that impact the FSSP
measurements of a population of droplets. The first is shattering of liquid and ice particles
impacting the shroud that can lead to miscounts. This is especially the case when
precipitation-sized ice particles are present (Korolev et. al. 2011, 2013). The second is
optical and electronic coincidence errors. Optical coincidence results from multiple
particles spaced close together such that they enter the laser beam and scatter as if they
were a single, irregularly shaped particle. Electronic coincidence (or dead time) is defined
as particles entering the laser during electronic reset time and thus not being detected
properly. The probability of optical and electronic coincidence error increases with
concentration. The probability is ~ 5% losses at 300 cmPi
to greater than 30% at
1000 cmPi
. Generally, algorithmic corrections are applied to account for these losses
(Baumgardner et al. 1985; Brenguier et al. 1989; Brenguier et al. 1994). They were applied
to the COPE-MED data used herein and in the rest of the COPE-MED investigations.
FSSPs are typically calibrated by the use of monodispersed glass beads. More
recently FSSPs have been calibrated using a stream of monodisperse water droplets
(Korolev 1985). Figure 17 shows the different Mie (theoretical response) curves for the
glass beads vs. liquid drops. Figure 17 shows the differences in the Mie curves for crown
glass beads and liquid water drops. Calibrations require adjusting instrument thresholds in
the probe electronics based on beads or drop diameters.

34
Figure 17: Differences in the Mie scattering curves for glass beads (those that are typically used in the FSSP calibration
procedure) and liquid water droplets for the manufacturer’s specified scattering angle range (figure adapated from NCAR:
https://www.eol.ucar.edu/instruments/forward-scattering-spectrometer-probe-model-100).
In COPE-MED the FSSP was not calibrated prior to deployment. The UWKA
research team rather conducted a check on particle sizing using glass beads. Two diameters
of glass beads ~10 µm and ~20 µm were used to test the sizing of the FSSP. These
diameters corresponds to 8.9 µm and 17.3 µm water-equivalent diameter, taking into
account difference between index of refraction between crown glass and water. Results of
these tests suggest the FSSP was oversizing between 3 µm to 5 µm (Figure 18). Thus, the
FSSP is not used in the analysis of LWC in Chapters 4 and 5.

35
Figure 18: Histograms from the May 08, 2013 glass bead sizing test of the FSSP. The FSSP looks large for the 2 bead
tests. The solid grey line represents the expected diameter reported by the FSSP. The histograms show that the mean
value of the bead tests are both larger.
3.3.2 CLOUD DROPLET PROBE
The cloud droplet probe (CDP) is also a single particle, forward-scattering optical
spectrometer manufactured by Droplet Measurement Technologies. It was developed, in
part to improve upon the mechanical design of the FSSP and more importantly to address
inherent design issues such as particle shattering. It was also developed to improve upon
the other optical and electronic issues that impact the FSSP. Lance et. al. (2010 & 2012)
made some design and calibration improvements that also serve to improve the results of
measurements by the CDP.
The CDP measures diameter distributions of cloud droplets in a similar manner to
the FSSP. One marked difference is that the CDP uses a 658 nm, single spatial and
longitudinal mode, diode laser. The advantage of using a single mode laser is that it is more
spatially coherent and avoids the inherent phase inhomogeneities that are found in a

36
multimode laser (Siegman 1986). Figure 19 illustrates an ideal Gaussian beam profile, the
CDP’s beam profile will be close to the idealization but not as symmetric. The CDP’s beam
is an elliptical Gaussian beam, anamorphic because the beam diverges in the two
perpendicular directions, roughly 2 mm × 0.2 mm (Lance et. al. 2010). The single-mode
laser also avoids the inability of a multimode laser to be focused to a diffraction-limited
spot (Siegman 1986). The disadvantage of the single-mode laser is that the Mie resonances
are generally more pronounced, whereas in a multi-mode laser they are typically damped
out (Knollenberg et al. 1976).
Figure 19: Representation of the CDP’s single-mode laser intensity profile. The beam is an elliptical Gaussian, which
will have a profile similar to the plot shown. However, this plot is an idealization and the real beam profile isn’t as
symmetric as shown here because the beam diverges in the two perpendicular directions.
The optical set up of the CDP consists of optics and detectors that are separated into
a “qualifier” component and a “sizer” component. A mask with a rectangular slit
configuration that allows single slit diffraction patterns to be detected in such a way that
the detector can determine if the particle is out of focus or in focus, is used as a mask in

37
the qualifying optics. If the particle is in focus – and within the depth of field – then it is
qualified as being in the sample area. If it is out of focus it is qualified as being out of the
sample area. The sample area of the CDP is heavily dependent on the optical configuration
of the device (Lance et. al. 2010). Figure 20 shows the mechanical and optical
configuration of the CDP to scale. Figure 21 shows the difference in the signals the qualifier
receives when the particle is within and outside the depth of field.
Figure 20: Scale representation of the mechanical and optical set-up within the housing of the CDP (DMT CDP User’s
manual).
(a) (b)
Figure 21: (a) Signals when the drop is in the sample area. (b) Signals when the drop is outside the sample area. (Lance,
et. al., 2010)

38
The sample area lies within the Rayleigh range of the elliptical Gaussian beam,
therefore the beam very closely approximates a plane wave. Therefore, the light scattered
by a droplet within that sample area closely follows Mie scattering theory (Bohren and
Huffman 1983). An un-masked photo detector and a comparator circuit are used for particle
sizing. The digital signal obtained by the sizing optics is directly related to the droplet’s
diameter. The CDP can measure droplet diameters in the range from 2 − 50 µm and
binned according to diameter ranges in 30 data bins (Lance et. al. 2010). The default
manufacturer’s configuration is for the first bins for diameters between 3 − 14 µm are
1 µm wide and the bins from 15 µm − 50 µm are 2 µm wide. There is a non-monotonic
relationship between the forward scattered light signal and droplet diameter, particularly
in the 9 µm − 15 µm diameter range, so that the sizing resolution of the CDP is rather
limited in its ability to get the fundamental sizing accurate (Lance et. al. 2010).
In order to account for the Mie ambiguities seen in the CDP, the UWKA research
group combines bins where there are known Mie resonances that lead to such ambiguities.
Specifically, the 9 µm & 10 µm , 11 µm & 12 µm , and 13 µm & 14 µm bins are
combined. Figure 22 illustrates a schematic of the CDP’s data bins and which bins are
combined by the UWKA group.

39
Figure 22: Schematic of the CDP bins. In order to account for the Mie ambiguities the UWKA research group combines
certain bins, namely the 9 qr and 10 qr, 11 qr and 12 qr, and 13 qr and 14 qr diameter bins. The bin diameters
shown here are the bin edges.
The sample area of the CDP used in COPE-MED was determined through direct
measurement by the manufacturer to be 0.30 mmm
. The sample volume is determined by
multiplying the area by the true airspeed (TAS), which is typically ~85 m sP{
for the
UWKA. The CDP operates most efficiently in TAS ranges of 10 − 200 m sP{
(Lance et.
al. 2010) and the UWKA operates well within this effective range.
The sources of uncertainty in the CDP result from of both optical and electronic
considerations, and though a marked improvement over the FSSP, it still suffers from some
of the same fundamental sources of error as the FSSP. Although the CDP’s single-mode
laser avoids the greater spatial intensity and phase inhomogeneity, the CDP still
experiences some broadening of the droplet spectrum from those effects. Optical
coincidence can still cause undercounts or oversizing errors in the CDP.
Lance et al. (2012) ran a Monte Carlo simulation after installing the pin-hole
qualifier optics mask and found that the CDP undercounted on the order of ~25% for N
greater than 500 cmPi
and improvement on the unmodified CDP, which would
undercount ~50% for N greater than 400 cmPi
(Lance et al. 2010). The CDP that was in
use during COPE-MED had that pin-hole qualifier mask installed, and when N results are

40
compared to those derived from the FSSP counts, and after being algorithmically corrected,
for the low precipitation cases the CDP was found to only undercount on average between
10% for N greater than 500 cmPi
.
Figure 23: Absolute value of the percent difference of the derived N of the un-corrected CDP with the pin-hole qualifier
mask to the N derived from the algorithmically corrected FSSP DSD data. For low precipitation cases, like RF03, the
CDP only undercounts, on average, ~10% for N greater than 500 òrPi
(as shown by the red line).
Particle shattering is minimized as part of the CDP’s inherent design. The
aerodynamic arms upstream of the beam location and its open path help resolve this.
Khanal et al. (2015) show that in ice-only clouds, the effect of ice shattering is at the CDP
Mean ~10%

41
particle detection limit. Finally, random statistical uncertainties in counts, due to Poisson
statistics, will add to the overall uncertainty of the CDP. However, this uncertainty is less
than 5% for concentrations greater than 13 cmPi
when given a 1Hz sampling rate and an
airspeed of ~100 m sP{
(Lance et. al. 2010). During COPE-MED the CDP sampled at
both 1 Hz and 10 Hz.
The CDP is calibrated in a similar manner as the FSSP. Prior to COPE-MED, the
CDP was calibrated by the manufacturer using the glass bead method and checked in the
field periodically in order to ensure proper sizing. Figure 24 shows the histograms from
one of these sizing checks using 10 bead diameters ranging from 2µm to 60 µm. On
average, the CDP sized correctly for the calibrated diameter range.

42
Figure 24: These glass bead sizing tests of the CDP, for 10 different diameters of glass beads show that, on average, the
CDP is sizing correctly. As before, the solid grey lines indicate the expected value for that diameter of glass bead.
3.4 GERBER PVM-100A
The laser-diffraction particle-sizing class of optical probes measure laser light
scattered by an ensemble of droplets passing through the sample volume of the probe. The
technique, originally described by Wertheimer and Wilcock (1976), relates the diffracted

Atmospheric Science Master of Science Thesis

Atmospheric Science Master of Science Thesis

Recommended

Recommended

More Related Content

Similar to Atmospheric Science Master of Science Thesis

Similar to Atmospheric Science Master of Science Thesis (20)

Recently uploaded

Recently uploaded (20)

Atmospheric Science Master of Science Thesis