Long-Term Statistics of Arctic Mixed-Phase Cloud Properties Retrieved From Doppler Spectra

Alexis N. Ortiz, Johannes Verlinde, Yaosheng Chen

Abstract

Mixed-phase clouds in the Arctic are highly influential within determining the surface ra-

diation budget, yet are poorly understood. These clouds contain both liquid water droplets and

ice crystals in the same volume, each with different impacts on the radiation budget. In order to

study the surface budget, it is necessary to separate and analyze the properties from each

phase. Using Ka-band zenith radar (KAZR) Doppler velocity spectra from the Atmospheric Radi-

ation Measurement (ARM) Program North Slope of Alaska (NSA) site, we separated contribu-

tions from the cloud liquid drop and the ice precipitation modes based on a continuous wavelet

transform and fuzzy logic techniques (Yu et al., 2014). Cloud liquid drop and precipitation reflec-

tivities, vertical air motions, and reflectivity-weighted mean fall velocities are retrieved for 836

hours of observations of single-layer mixed-phase clouds from September to December, 2011

to 2014. Our analysis reveal a relationship between the maximum precipitation reflectivities and

minimum cloud temperature similar to that of the temperature dependence of ice crystal vapor

depositional mass growth reported by Chen and Lamb (1994). These findings could be the re-

sult of a temperature dependence of the ice crystal mass growth in the observed clouds, al-

though other factors such as various scattering processes could also play a role.

Additional results suggest a temperature dependence between maximum precipitation reflectivi-

ties and mean vertical velocities in mixed-phase clouds, revealing that clouds with temperatures

between -20°C and -12°C were most likely to contain stronger downdrafts and larger ice particu-

lates. However, future research is needed to validate these results as well as to achieve greater

understanding of these influential clouds in the radiation budget.

Introduction

1

Mixed-phase clouds, those that are composed of supercooled liquid water droplets and

ice particles in the same volume of air, play an important role in the radiation budget over the

Arctic region. Since these clouds are made up of both liquid water and ice particles, differentiat-

ing the contributions from each precipitation mode is essential for expanding our understanding

of mixed-phase clouds within the radiation budget. Previous studies have discovered that

mixed-phase clouds are often found in multiple layers, containing complicated ice growth pro-

cesses (Herman and Goody 1976; Pinto et al. 2001; Verlinde et al. 2013). Most global models

do not have the ability to interpret such complex growth processes, and instead incorrectly treat

ice particulates as spheres. In short, there is a severe lack of understanding of the development

of mixed-phase clouds and the processes that occur within them.

However, recent funding from the U.S. Department of Energy (DOE) Atmospheric Radia-

tion Measurement (ARM) Program led to the installation of the ARM Climate Research Facility

at Barrow, Alaska, on the North Slope of Alaska (NSA) in 2011 (Bharadwaj et al. 2011). This

Ka-band zenith pointing radar collected data describing the characteristics of mixed phase

clouds over the region. From 2011 to 2014, analyzing the KAZR data using a continuous

Gaussian wavelet transform and fuzzy logic techniques which were suitable for conditions only

containing a single, stratiform layer of mixed-phase clouds over the NSA site, we concluded that

there were 836 available hours worth of data. This data was analyzed for comparisons between

radar reflectivities and minimum cloud temperatures, suggesting that there is a relationship be-

tween the maximum precipitation reflectivity and minimum cloud temperature, to that of ice crys-

tal mass growth by vapor deposition and the temperature reported by Chen and Lamb (1994).

Our results reveal a greater understanding of the processes that occur within mixed-phase

clouds over the Arctic region and will allow us to better parameterize these processes within our

global models, resulting in greater accuracy of the global radiation budget.

Data and Methods

2

Most of our data was collected via the KAZR instrument at the NSA site in Barrow,

Alaska. This zenith pointing radar operates at a frequency of approximately 35 GHz, remotely

probing the composition of mixed-phase clouds at millimeter wavelengths. From September to

December, 2011 to 2014, the instrument collected doppler radar reflectivities of both the cloud

liquid water droplets and ice crystals, in addition to the vertical velocities of these particles. In

addition, microwave radiometers and radiosondes located at the ARM Climate Research site in

Barrow, AK were used to collect additional information about the cloud structure, such as cloud-

base and cloud-top heights, liquid water paths, and cloud thicknesses. This background infor-

mation is needed to ensure that the mixed-phase clouds were single-layer, and not multi-layer,

as our applied algorithm is not designed to analyze multi-layer mixed-phase clouds. Once our

data was collected, we ran a few algorithms in MATLAB to concatenate the available hours that

were single-layer mixed-phase clouds. We then de-noised the spectra values, and applied a

Gaussian continuous wavelet transform and fuzzy logic techniques to separate the cloud liquid

drop mode and the precipitation mode from one another.

Our case selection consisted of cloud layer temperatures between -40°C and 0°C, with a stan-

dard deviation of

cloud base heights

less than 50 meters.

Cloud liquid water

paths were greater

than 25 g m^-2,

cloud top heights

were lower than

2500 m, and cloud

thicknesses were

between 200 and

3

2000 m. We then ran statistical comparisons between maximum reflectivities (dBZ) of the cloud

liquid drop mode and the precipitation mode against the minimum cloud temperature (°C). In ad-

dition, we generated comparisons between mean vertical velocities in the cloud layer (m/s)

against cloud thicknesses (m). Lastly, we examined a possible temperature dependence on ice

crystal mass growth by comparing maximum precipitation reflectivities against maximum cloud

liquid drop reflectivities at various temperatures, as well as against fall velocities. Our hypothe-

sis was that we would see enhanced fall velocities and higher reflectivities in temperatures be-

tween -20°C and -10°C, as Chen and Lamb (1994) had suggested that ice crystal mass growth

is most prominent in this temperature interval.

Results

4

From the KAZR instrument at the NSA site, we were able to analyze the frequency of

joint distribution between the maximum precipitation reflectivity (dBZ) and minimum cloud tem-

peratures (°C) gathered from radiosonde data. Using bin widths of 1 dBZ and 1°C respectively,

we discovered that the highest frequency distribution was located between the temperatures of -

20°C and -10°C, with a smaller secondary peak near -5°C. When comparing maximum cloud

liquid drop reflectivity (dBZ) against minimum cloud temperature, there was very little variance in

cloud reflectivity, with peak values occurring in the -10°C to 0°C range, but no distinct parabolic

shape in the data was observed.

Next, we wanted to determine that if given the same maximum cloud liquid drop reflectiv-

ity, would there be a noticeable difference in maximum precipitation reflectivity, at various tem-

peratures. At minimum cloud temperatures less than -20°C, we observed the highest frequency

of maximum precipitation reflectivity around -20 dBZ, with a range of large frequencies between

-30 and -10 dBZ. At minimum cloud temperatures between -20°C and -12°C, the highest fre-

quency of maximum precipitation reflectivity was most concentrated between 0 and -10 dBZ,

with some observations greater than 0 dBZ. Between -12°C and -7°C, the greatest frequency of

joint distribution was between -30 dBZ and -20 dBZ, while the range of -7°C and -3°C showed

similar results in the -30 dBZ to -20 dBZ range. It should be noted that the greatest values of

joint frequency distribution between maximum precipitation reflectivity and cloud liquid drop re-

5

flectivity were between temperatures of -20°C and -12°C, and that much of the data was located

in higher maximum precipitation values, as opposed to the rest of the other temperature profiles,

where the greatest frequency distribution was located at lower maximum precipitation reflectivity

values.

We also compared the frequency of joint distribution between precipitation reflectivity

and fall velocity at various temperatures, searching for which temperature range would produce

the largest fall velocities, signaling larger particulates. For minimum cloud temperatures less

than -20°C, the highest frequency between precipitation reflectivities and fall velocities were ob-

served around velocities of 0.5 m/s and reflectivities between -20 dBZ and -10 dBZ. In the tem-

perature range of -20°C and -12°C, there was a large swath of high frequency values with fall

velocities between 0.5 m/s and 1 m/s, with precipitation reflectivities most common between 0

dBZ and -10 dBZ, but noticeable reflectivities present even up to 10 dBZ. For temperatures be-

tween -12°C and -7°C, highest frequencies were observed when fall velocities were near 0.5

m/s and precipitation reflectivities between -30 dBZ and -20 dBZ. For temperatures between -

7°C and -3°C, most common frequencies were observed with fall velocities near 0.5 m/s and

precipitation reflectivities near -30 dBZ. It should be noted that the temperature range of -20°C

6

to -12°C recorded the highest frequency values of all the other temperature profiles, while the

warmest temperature profile (-7°C to -3°C) generated the lowest frequency values. Also, every

profile except the warmest profile appeared to show a slight linear trend between fall velocity

and precipitation reflectivity, whereas the warmest profile showed constant fall velocities of 0.5

m/s for all precipitation reflectivities.

Next, we compared the frequency distributions between cloud thicknesses and mean

vertical velocities in the cloud layer, in an attempt to observe any trend suggesting that either

larger or smaller particulates were observable in updrafts or downdrafts. With bin widths of 30 m

for cloud thickness and 0.025 m/s for mean vertical velocity, we observed a circular pattern for

the frequency values, with the highest values centered around cloud thicknesses of 400 m and

vertical velocities near 0 m/s. When normalizing the data for velocity bin widths, the highest fre-

quencies were observed for cloud thicknesses around 500 m for all mean vertical velocities,

with no observable trend. Similarly, when breaking these results down by minimum cloud tem-

peratures, for temperatures between -20°C and -12°C, there was no observable largest fre-

quencies, however, moderate frequencies were observed between cloud thicknesses of 800 m

and 1200 m. When comparing cloud thickness with mean vertical velocity for temperatures less

7

than -20°C or greater than -

12°C, the largest frequencies

were evident in the 500m

cloud thickness band, but con-

sistent with all mean vertical

velocities.

We also compared mean

vertical velocities in the cloud

layer against the minimum

cloud temperature, in an at-

tempt to reveal any additional relationships between features in the updrafts or downdrafts of

the cloud layer. There was no distinguishable pattern in the frequency of joint distribution plot,

although the highest frequency values occurred when the minimum cloud temperature was near

-10°C with a secondary peak around -15°C. However, when normalizing by the velocity bins,

there was a very apparent peak in frequency when the minimum cloud temperature was -15°C

with mean vertical velocities around -0.6 m/s, the strongest downdrafts values within our plot.

There was also a consistent frequency value for the entire velocity profile between -5°C and -

10°C. Further investigating into downdrafts and mean vertical velocities, we decided to compare

the mean vertical velocities against maximum precipitation reflectivity. Our results displayed a

fairly spherical pattern, with frequency values suggesting a slight relationship between the two

variables. As maximum precipitation increases from -30 dBZ to 0 dBZ, mean vertical velocities

decreased from ~0 m/s to -0.6 m/s. When normalizing by velocity bins, there was a very evident

pattern between the two variables. Large frequency values were evident for maximum precipita-

tion values between -25 dBZ and -15dBZ with mean vertical velocities between 0 m/s and 0.6

m/s, whereas for maximum precipitation reflectivity values between -20 dBZ and 10 dBZ, mean

vertical velocities decreased from 0 m/s to -0.6 m/s as precipitation reflectivity values increased.

8

This pattern was observable in all four minimum cloud temperature ranges: < 20°C, -20°C ~ -

12°C, -12°C ~ -7°C, and -7°C ~ -3°C. However, for minimum cloud temperatures between -20°C

and -12°C, maximum frequency values were observed for reflectivities near 0 dBZ, with mean

vertical velocities near -0.6 m/s.

Discussion

9

Analyzing the relationship between maximum precipitation reflectivity and minimum

cloud temperature, it appears that there is a striking resemblance to that of the ice crystal mass

growth by vapor deposition and the tempera-

ture reported as suggested in Chen and Lamb

(1994). Shown below, Chen and Lamb sug-

gested the greatest mass of ice crystals for

cloud temperatures near -15°C, with a sec-

ondary peak between -10°C and -5°C. Our re-

sults, below, support this claim as there is a

similar parabolic pattern peaking at -15°C and

decreasing as temperature increases, with a

secondary smaller peak near -5°C. These

findings suggest that in this temperature

range, within single-layer mixed-phase

clouds in the Arctic, ice crystals are created

via vapor deposition at the rate suggested

by Chen and Lamb. However, when com-

paring the maximum cloud liquid drop re-

flectivity to minimum cloud temperature,

there was no such resemblance to Chen and Lamb’s findings. This is valid, as the precipitation

reflectivity mode encapsulates larger ice crystals, which would grow according to Chen and

Lamb’s findings, and therefore shows a similar pattern, whereas the liquid cloud drop mode con-

sists of smaller liquid water droplets, which do not grow in size according the rate suggested by

vapor deposition.

When analyzing the relationship between maximum precipitation reflectivity and maxi-

mum cloud liquid drop reflectivity, we divided our findings into four separate profiles determined

10

by various cloud temperatures. Our findings suggested that given the same maximum cloud liq-

uid drop reflectivites, clouds with minimum temperatures between -20°C and -12°C tended to

produce larger maximum precipitation reflectivities, as opposed to any other cloud temperature

range. Again, this is supported by the prediction that ice mass growth occurs most significantly

within this temperature range. Similarly, this temperature range also produced the highest fall

velocities, again suggesting that larger ice particles or aggregates are present in Arctic single-

layer mixed-phase clouds between the temperatures of -20°C and -12°C.

Furthermore, research conducted in 2014 by Chen, Yu, and Verlinde suggests that

within the updrafts of mixed-phase clouds in the Arctic, the mean updraft speed increases with

cloud thickness. However, when including the most recent data from the NSA site in Barrow,

Alaska, our findings suggest that no such relationship exists, as there is simply a constant mean

vertical velocity at cloud thicknesses of 500 m, with no observable trend as cloud thickness in-

creases. However, when comparing mean vertical velocity with maximum precipitation reflectiv-

ity, there does appear to be a relationship between downdrafts and precipitation reflectivities

when both increase. When normalizing the data, this relationship becomes even more apparent.

This suggests that as precipitation reflectivites increase, starting around -30 dBZ, so does the

mean downdraft speed of these ice particulates. When splitting these figures up by temperature

profiles, the highest frequency values are observed for very high precipitation reflectivites and

strong downdrafts in the temperature range of -20°C and -12°C. This agrees with both Chen

and Lamb’s and our own results, as larger ice particulates would be present in this temperature

range, resulting in greater downdrafts as the ice particle increases its downward speed as mass

increases.

Conclusion

This study explores the relationships between the macro and micro-physics, thermody-

namics, dynamics and additional features of single-layer mixed-phase clouds in the Arctic. Us-

11

ing collected data from the KAZR at the NSA site in Barrow, Alaska during the months of Sep-

tember to December, 2011 to 2014, we applied an algorithm based on the Gaussian continuous

wavelet transform and fuzzy logic techniques, in order to accurately separate the contributions

from the cloud liquid drop mode and the precipitation mode. Upon comparisons between the

maximum precipitation reflectivities and minimum cloud temperature, we witnessed a remark-

able resemblance to that of Chen and Lamb’s ice mass growth findings. We concluded that

within mixed-phase clouds in the Arctic, it is likely that ice particulates at various temperatures

increase in mass by vapor deposition at the rate suggested by Chen and Lamb, with the great-

est ice crystal mass located within single-layer mixed-phase clouds with minimum temperatures

near -15°C. Additionally, our results revealed that mixed-phase clouds within the temperatures

of -20°C and -12°C were most likely to produce the largest precipitation reflectivites and fall ve-

locities, validating the conclusion that larger ice particulates, or aggregates were present in

clouds at these temperatures. However, most recent available data disproved past results,

which suggested that within the updrafts of these mixed-phase clouds, the mean updraft speed

increased with cloud thickness. No relationship was apparent; however, comparisons between

mean vertical velocity and minimum cloud temperature suggests that the strongest downdrafts

are found in mixed-phase clouds with temperatures near -15°C, further validating our results.

Additionally, we discovered a strong relationship between mean vertical velocities in downdrafts

and larger maximum precipitation reflectivites, which suggested that downdraft speeds increase

as precipitation reflectivites increase. This pattern was most evident in cloud temperatures be-

tween -20°C and -12°C, which displayed the largest frequencies for vertical velocities near -0.6

m/s and precipitation reflectivites near 0 dBZ, suggesting that there are large ice particulates

with large downdraft speeds in these mixed-phase clouds, increasing their mass growth accord-

ing to Chen and Lamb’s findings.

Acknowledgements

12

This research was supported by the National Science Foundation through Grant AGS-

1263225, in partnership with the Undergraduate Research Experience hosted by the Pennsylva-

nia State University. Significant contributions were made possible with the help of Johannes

Verlinde and Yaosheng Chen.

13