atmosphere
Article
Lidar and Ceilometer Observations and Comparisons
of Atmospheric Cloud Structure at Nagqu of
Tibetan Plateau in 2014 Summer
Xiaoquan Song 1,2 , Xiaochun Zhai 1 , Liping Liu 3 and Songhua Wu 1,2, *
1
2
3
*
Ocean Remote Sensing Institute, Ocean University of China, Qingdao 266100, China;
[email protected] (X.S.); [email protected] (X.Z.)
Laboratory for Regional Oceanography and Numerical Modeling, Qingdao National Laboratory for
Marine Science and Technology, Qingdao 266100, China
State Key Laboratory of Science Weather, Chinese Academy of Meteorological Science, Beijing 100081, China;
[email protected]
Correspondence: [email protected]; Tel.: +86-532-6678-2573
Academic Editor: Robert W. Talbot
Received: 12 September 2016; Accepted: 5 December 2016; Published: 17 January 2017
Abstract: In the project of the Third Tibetan Plateau Experiment of Atmospheric Science (TIPEX III),
the intensive observation of cloud and precipitation in Nagqu was conducted from 1 July to 31 August
2014. The CL31 ceilometer and a WAter vapor, Cloud and Aerosol Lidar (WACAL) were deployed and
focused on studying the cloud macroscopic characteristics and vertical distribution. The statistical
result of CL31 ceilometer in continuous operation mode shows that the cloud occurrence is about
81% with a majority of simple one-layer cloud. The cloud base and top height are retrieved by
improved differential zero-crossing method using lidar data. The results of cloud base height (CBH)
are compared with CL31 ceilometer, showing a good consistency with each other, however, in some
cases, the CL31 ceilometer overestimates the CBH and is also validated by synchronous radiosonde
data. The time snippet comparisons of cloud property between CL31 ceilometer and lidar imply
that the cloud properties have obvious diurnal variations with “U” shape distribution. The cloud
development including the time-spatial distribution features also has distinct diurnal variations based
on the lidar measurement. The detection range of lidar goes beyond the maximum height of CL31
ceilometer, offering substantial observations to the analysis of cirrus cloud radiation characteristics
and formation mechanism.
Keywords: Tibetan Plateau; cloud property; CL31 ceilometer; lidar
1. Introduction
Clouds play a vital role in the radiative transfer within the Earth’s atmosphere surface system
and thus for climate system [1,2]. Different types of cloud have varying effects on the atmosphere.
Generally, clouds exert both a cooling effect on the surface by reflecting shortwave radiation back into
space and a warming effect by trapping longwave radiation emitted from the Earth’s surface [3–6].
Due to its complex climate feedback and uncertainty in the process of climate change, the study of
cloud macroscopic properties, such as cloud type, cloud base and top height, and temporal and spatial
distributions in high temporal and spatial resolution are significant for describing the impact of clouds
in a changing climate.
The Tibetan Plateau is a vast elevated plateau in the middle of the Eurasian continent with
averaged elevation above 4500 m mean sea level (m.s.l.). It has an important role in the global and
regional climate system by its unique dynamic and thermodynamic forcing effect [7,8], and the thorough
research on cloud distribution is quite essential for understanding the climate of Asian monsoon region
Atmosphere 2017, 8, 9; doi:10.3390/atmos8010009
www.mdpi.com/journal/atmosphere
Atmosphere 2017, 8, 9
2 of 2
Atmosphere 2017, 8, 9
2 of 17
monsoon region and Tibetan Plateau. With the development of remote sensing, satellite data have
been applied to the Tibetan Plateau cloud characteristics research. Wang et al. [9] analyzed the
Tibetan
Plateau
cloud
vertical
distribution
its seasonal
using
and
Tibetan
Plateau.
Withoccurrence,
the development
of remote
sensing,and
satellite
data have variation
been applied
to
CloudSat/Cloud-Aerosol
and Infrared
Pathfinder
(CALIPSO)
data,
and
the
Tibetan Plateau cloudLidar
characteristics
research.
Wang Satellite
et al. [9] Observations
analyzed the Tibetan
Plateau
cloud
found that thin
cloud
with a range
of its
4–11
km arevariation
most frequent
this area and a higher grid
vertical
occurrence,
vertical
distribution
and
seasonal
using in
CloudSat/Cloud-Aerosol
Lidar
and
resolution
of GCM isSatellite
needed Observations
for accurate description
of data,
cloudand
occurrence
andthin
location.
[10]
Infrared
Pathfinder
(CALIPSO)
found that
cloudZhao
withetaal.
range
analyzed
the
distribution
of Tibetan
Plateau
totalgrid
water
path (TWP)
andof
itsGCM
seasonal
variation
of
4–11 km
are
most frequent
in this area
andcloud
a higher
vertical
resolution
is needed
for
using
CloudSat
data,
finding
that
the
value
of
TWP
in
warm
season
is
larger
than
cool
season’s
and
accurate description of cloud occurrence and location. Zhao et al. [10] analyzed the distribution of
the vertical
distribution
ofwater
precipitation
clouds
is using
quite CloudSat
different from
East Asia’s.
Tibetan
Plateau
cloud total
path (TWP)
and in
itsTibetan
seasonalPlateau
variation
data, finding
that
Fu
et
al.
[11]
analyzed
a
case
study
of
the
precipitation
cloud
in
the
valley
of
Tibetan
Plateau,
the value of TWP in warm season is larger than cool season’s and the vertical distribution of precipitation
indicating
that thePlateau
topographic
is the
this
cloud
clouds
in Tibetan
is quiteforcing
different
frommain
East factor
Asia’s.ofFu
et al.
[11] formation.
analyzed a However,
case studythe
of
deficiency
of ground-based
observation
is the main
constraint
for the
accurate
retrieval using
satellite
the
precipitation
cloud in the
valley of Tibetan
Plateau,
indicating
that
the topographic
forcing
is the
data. factor
Recently,
lidar
has formation.
been an effective
quantitative
tool of
of ground-based
ground-based observation
cloud detection
to
main
of this
cloud
However,
the deficiency
is thedue
main
its high temporal-spatial
resolution
andsatellite
measurement
accuracy
[12].
et an
al.effective
[13] carried
out the
constraint
for the accurate retrieval
using
data. Recently,
lidar
hasLiu
been
quantitative
research
on microphysical
macroscopic
cloud and precipitation
distribution
tool
of ground-based
cloudand
detection
due tofeatures
its high of
temporal-spatial
resolutionvertical
and measurement
using multiple
wavelength
andout
lidar
during
TIPEX III, and
found
that the statistics
accuracy
[12]. Liu
et al. [13]radar
carried
thedevices
research
on microphysical
and
macroscopic
features
parameters
cloud havevertical
obviousdistribution
diurnal variations.
He et wavelength
al. [14] analyzed
the lidar
cirrusdevices
cloud
of
cloud andofprecipitation
using multiple
radar and
geometry
and
optical
characteristics
in
summer
Tibetan
Plateau
using
micro-pulsed
lidar,
finding
during TIPEX III, and found that the statistics parameters of cloud have obvious diurnal variations.
thatetthere
exists
obvious
between
Tibetan
and same latitude
area. Liu
[15] analyzed
He
al. [14]
analyzed
thedifference
cirrus cloud
geometry
andPlateau
optical characteristics
in summer
Tibetan
Plateau
the aerosol
optical characteristics
Nagqu
lidar, and
indicated
thatPlateau
Nagqu area
using
micro-pulsed
lidar, findingin
that
thereusing
existsmicro-pulsed
obvious difference
between
Tibetan
and
has
a
better
air
quality
than
other
areas’
in
China.
same latitude area. Liu [15] analyzed the aerosol optical characteristics in Nagqu using micro-pulsed
a indicated
part of thethat
TIPEX
III, the
University
of China
operated
the CL31 ceilometer
lidar,As
and
Nagqu
areaOcean
has a better
air quality
than(OUC)
other areas’
in China.
(short
for
CL31)
and
the
WACAL
in
summer
of
2014
to
carry
out
the
research
on
microphysical
As a part of the TIPEX III, the Ocean University of China (OUC) operated thethe
CL31
ceilometer
characteristics
and
the
vertical
structure
of
cloud
in
Tibetan
Plateau.
In
this
paper,
the cloud
(short for CL31) and the WACAL in summer of 2014 to carry out the research on the microphysical
characteristicsand
based
on CL31
observation
is infirstly
analyzed.
Then,
the comparison
of the CBH
characteristics
the vertical
structure
of cloud
Tibetan
Plateau. In
this paper,
the cloud characteristics
between
CL31
and
the
WACAL
is
carried
out.
The
time
snippet
comparison
of
CBH
and
theand
cloud
based on CL31 observation is firstly analyzed. Then, the comparison of the CBH between CL31
the
occurrence
between
these
two
devices
is
analyzed
after
the
determination
of
the
data
acquirability
of
WACAL is carried out. The time snippet comparison of CBH and the cloud occurrence between these
the WACAL.
the
averaged
spatial distribution
of acquirability
different layer
measured
by the
the
two
devices is Finally,
analyzed
after
the determination
of the data
of clouds
the WACAL.
Finally,
WACAL
is
analyzed
at
different
observation
periods,
further
making
up
the
constraint
of
CL31
averaged spatial distribution of different layer clouds measured by the WACAL is analyzed at different
undetectableperiods,
condition
above
7.5 km.
observation
further
making
up the constraint of CL31 undetectable condition above 7.5 km.
2. Data and Methodology
Lidar observations in summer of 2014 as a part of the TIPEX III were performed
performed at the Nagqu
Meteorological Bureau (31.48
(31.48°◦ N, 92.06
92.06°◦ E, 4508 m m.s.l.) on the Tibetan Plateau. The location of the
Nagqu site and the topography of the Tibetan Plateau
Plateau are
are shown
shown in
in Figure
Figure 1.
1. Nagqu is located in the
central part
part of
of the
theTibetan
TibetanPlateau.
Plateau.The
Theelevated
elevatedsurface
surfaceheat
heatand
and
the
rising
over
plateau
lead
central
the
rising
airair
over
thethe
plateau
lead
to
to anticyclonic
circulation
divergence
in the
upper
troposphere
lower
stratosphere
anticyclonic
circulation
andand
divergence
in the
upper
troposphere
andand
lower
stratosphere
[16].[16].
1. Location of the experimental observation site and the topography of the Tibetan Plateau
Plateau
Figure 1.
(DEM: Digital
Digital Elevation
Elevation Model).
Model).
(DEM:
Atmosphere 2017, 8, 9
3 of 17
Atmosphere
2017, 8, 9
2.1. Set Up
Introduction
3 of 3
2.1. Set Up Introduction
2.1.1. CL31
CL31
The2.1.1.
CL31
measures backscattered light at wavelength of 910 nm with measurement ranging
from 0 to 7.6The
km.
Themeasures
minimum
vertical and
resolutions
are
5 mmeasurement
in height and
2 s in time,
CL31
backscattered
lighttemporal
at wavelength
of 910 nm
with
ranging
respectively.
The
resolution
inand
thistemporal
study resolutions
amounts to
CL31
from 0 to
7.6 time
km. The
minimumused
vertical
are16
5 ms.inThe
height
and 2can
s in detect
time, three
respectively.
The time resolution
in this
studylaser
amounts
16 s. The CL31The
can single-lens
detect three cloud
cloud layers
simultaneously
with a used
pulsed
diode
lidartotechnology.
technology
layers
simultaneously
with
a
pulsed
diode
laser
lidar
technology.
The
single-lens
technology
provides excellent performance at low altitudes and reliable operation in all weather. Furthermore,
fast
provides excellent performance at low altitudes and reliable operation in all weather. Furthermore, fast
measurement also provides reliable detection of thin cloud layers below a solid cloud base. Figure 2a
measurement also provides reliable detection of thin cloud layers below a solid cloud base. Figure 2a
shows the
photo
of CL31.
TheThe
basic
parameters
arelisted
listed
of Table
shows
the photo
of CL31.
basic
parametersof
of CL31
CL31 are
in in
thethe
left left
of Table
1 [17].1 [17].
Figure 2. Pictures of: (a) CL31; and (b) WACAL.
Figure 2. Pictures of: (a) CL31; and (b) WACAL.
Table 1. Component parameters of the WACAL system and CL31.
Table 1. Component parameters of the WACAL system and CL31.
Specification
CL31
Laser type
InGaAs diode
Specification
CL31
Wavelength (nm)
910
Pulse
energy
(mJ)
0.0012
Laser type
InGaAs diode
Repetition rate (kHz)
10
Wavelength
(nm)
910Lens/96
Telescope/aperture
(mm) Single
0–7.6
Pulse energyDetection
(mJ) range (km)
0.0012
Range resolution (m)
5
Repetition rate (kHz)
10
2–120
Temporal resolution (s)
programmable
Telescope/aperture (mm)
Single
Lens/96
WACAL
Nd:YAG
354.7/532/1064WACAL
410/120/700 Nd:YAG
0.03
354.7/532/1064
Light Bridge/304.8
0.05–25 410/120/700
3.75
0.03
1–120
programmable
Light Bridge/304.8
Detection range (km)
0–7.6
0.05–25
Range resolution (m)
5
3.75
2.1.2. WACAL
The lidar measurements are conducted with the fully automatic portable multi-wavelength
resolution
1–120(OUC).
programmable
RamanTemporal
and polarization
lidar(s)WACAL 2–120
[18] byprogrammable
Ocean University of China
Figure 2b shows
the photo of WACAL, and the basic parameters are listed in the right of Table 1. WACAL measures
backscattered light at wavelength of 355 nm, 532 nm and 1064 nm and Raman scattered light at 387 nm
2.1.2. WACAL
and 407 nm. For cloud measurement, the 532 nm channel with pulse energy of 120 mJ is used. The beam
is used to decrease
the beam divergence
and ensure the
high optical
transmittance. Raman
Theexpander
lidar measurements
are conducted
with the angle
fully automatic
portable
multi-wavelength
After
a
laser
pulse
is
transmitted
into
the
atmosphere,
the
backscattered
signal
is
and polarization lidar WACAL [18] by Ocean University of China (OUC). Figure 2bcollected
shows by
the photo
telescope and then transmitted to the detection system. The 532 nm signal is separated by the dichroic
of WACAL, and the basic parameters are listed in the right of Table 1. WACAL measures backscattered
long-pass filter, and then the 532 nm parallel-polarized and perpendicular-polarized signals are
light at wavelength
of 355
nm, 532
nmsplitter
and 1064
nm
and
Raman
light at
387(PMTs)
nm and
separated by the
polarizing
beam
(PBS)
and
detected
byscattered
photomultiplier
tubes
to 407 nm.
For cloud
measurement,
the 532ratio.
nm The
channel
with
pulse energyand
of 120
mJ is used. The beam
expander
retrieve
the depolarization
sum of
parallel-polarized
perpendicular-polarized
signals
is used
to study
The vertical
and temporal
resolutions
are 3.75
m in height andAfter
30 s a laser
is used to
decrease
thecloud
beamstructure.
divergence
angle and
ensure the
high optical
transmittance.
pulse is transmitted into the atmosphere, the backscattered signal is collected by telescope and then
transmitted to the detection system. The 532 nm signal is separated by the dichroic long-pass filter, and
then the 532 nm parallel-polarized and perpendicular-polarized signals are separated by the polarizing
beam splitter (PBS) and detected by photomultiplier tubes (PMTs) to retrieve the depolarization ratio.
The sum of parallel-polarized and perpendicular-polarized signals is used to study cloud structure.
Atmosphere 2017, 8, 9
4 of 17
The vertical and temporal resolutions are 3.75 m in height and 30 s in time, respectively. In contrast to
CL31, the more powerful laser of WACAL enables measurements up to 15 km, offering substantial
observations to the analysis of high cloud.
2.1.3. Model GTS1 Digital Radiosonde
In operation with Model GFE (L) 1 Secondary Radar, Model GTS1 digital radiosonde can measure
the atmospheric temperature, pressure, relative humidity, wind direction and wind velocity from
ground up to altitude of 30 km. The Radar captures and automatically tracks signals transmitted from
a radiosonde attached to a balloon to obtain atmospheric conditions, calculating wind direction and
wind velocity while the balloon is being uplifted, and it inputs the received data along with angle
data and time signal into the data processing unit for real time data processing. GTS1 has come into
use in some upper air stations as the new generation upper air sounding in China because of its high
sampling rate, high anti-jamming capability, modularity and digitization characteristics. The basic
parameters of GTS1 digital radiosonde are listed in Table 2 [19].
Table 2. Component parameters of the GTS1 radiosonde.
Meteorological Sensor
Temperature
Specification
Technical Parameter
Range
−90–50 ◦ C
0.2 ◦ C (−80–50 ◦ C)
Accuracy (standard deviation)
0.3 ◦ C (−90–−80 ◦ C)
Humidity
Resolution
0.1 ◦ C
Range
0% RH~100% RH
5% RH (T ≥ 25 ◦ C)
Accuracy (standard deviation)
10% RH (T ≤ 25 ◦ C)
Resolution
1% RH
Range
Pressure
Accuracy (standard deviation)
1060 hPa~5 hPa
2 hPa (1050 hPa~500 hPa)
1 hPa (500 hPa~5 hPa)
Resolution
0.1 hPa
2.2. Methodology
2.2.1. Improved Differential Zero-Crossing Method Using WACAL
The methods of CBH retrieval using lidar data mainly include the threshold method [20] and
differential zero-crossing method [21]. In threshold method, the CBH is defined as the level where there
is firstly an increase in signal above the clear background level and of a magnitude equal to N times the
standard deviation of the background fluctuations Furthermore, it is required that the signal should
be increased continuously for at least M successive height intervals. Thus, it has a relative complex
adjustment process since the threshold value M and N need to be adjusted repeatedly based on the
properties of cloud and aerosol so that it is hard to realize the automatic processing. The differential
zero-crossing method judges the point where the value of the first derivative of the lidar backscatter
power changes from negative to positive as the CBH, but there may exist misjudgment when the
Signal Noise Ratio (SNR) is lower or the fluctuation of noise is too large. In this section, the improved
differential zero-crossing method is used [22], and it is based on differential zero-crossing method,
and additionally three threshold conditions are also used to effectively avoid misdiagnoses. Figure 3
shows the flow chart of cloud properties retrieval based on improved differential zero-crossing method.
The raw lidar data need pre-processing including the noise removal and temporal-spatial average to
improve their SNR. In order to decrease the cloud layer misdiagnoses due to the aerosol layer and
background noise, three threshold conditions are carried out in this algorithm. Firstly, the background
Atmosphere 2017, 8, 9
5 of 17
Atmosphere
2017, 8,
9
5 of 5 e
noise
gradient
threshold
is calculated. Specifically, the mean value K and the standard deviation
of the background noise first derivative are calculated, and the candidate cloud region should meet
P(r) of
where
is the
after
pre-processing,
andsignal
n is a after
positive
with fine tuning
inis
the
condition
dP(lidar
r)/drsignal
≥ K+
ne, data
where
P(r) is the lidar
datavalue
pre-processing,
and n
atmospheric
conditions,
the analyzed
area is regarded
asthe
cloudless
area.
a different
positive value
with fine
tuning in otherwise,
different atmospheric
conditions,
otherwise,
analyzed
area
cloud
base andthe
peak
intensityof
should
calculated.
Specifically,
meanbe
isSecondly,
regardedthe
as threshold
cloudless of
area.
Secondly,
threshold
cloudbebase
and peak
intensity the
should
value M ofSpecifically,
the background
noisevalue
is calculated,
the candidate
region should
be candidate
satisfied
calculated.
the mean
M of theand
background
noisecloud
is calculated,
and the
Ppeak (r) and
Pbase (r)peak
that the
lidarshould
signal intensity
difference
the cloud
peak difference
cloud base
is
cloud
region
be satisfied
that thebetween
lidar signal
intensity
between
the cloud
Ppeak
(r)than
and cloud
baseis,
Pbase
(r(r)) is P
larger
than aM, that is, Ppeak (r) value
− Pbasewith
(r) ≥
aM,
and ainisdifferent
a positive
Ppeak
aM , that
larger
fine
tuning
base (r) ≥ aM , and a is a positive
value
with
fine
tuning
in
different
atmospheric
conditions,
the
tuning
standard
of
which
has been
atmospheric conditions, the tuning standard of which has been listed in Table 3. Otherwise,
the
listed
in
Table
3.
Otherwise,
the
analyzed
area
is
regarded
as
cloudless
area.
Thirdly,
the
threshold
analyzed area is regarded as cloudless area. Thirdly, the threshold of CBH is determined based on
ofthe
CBH
is determined
based on that
the ground-based
observations
that
averaged
CBHoverlap
is about
ground-based
observations
the averaged low
CBH is about
512the
± 148
m withlow
the lidar
512
±
148
m
with
the
lidar
overlap
into
consideration.
Thus,
the
layer
below
500
m
above
ground
into consideration. Thus, the layer below 500 m above ground level (AGL) is eliminated [23]. Figure
level
(AGL)
is
eliminated
[23].
Figure
4
shows
the
case
study
using
improved
differential
zero-crossing
4 shows the case study using improved differential zero-crossing method [22]. The left diagram is
method
[22]. the
Themiddle
left diagram
is is
lidar
signal,
the middle
is the
first derivative
the lidar
lidar signal,
diagram
the first
derivative
of thediagram
lidar signal
(dashed
line) with of
candidate
signal
(dashed
line)
with
candidate
cloud
layer
(solid
line
value
equals
to
1).
Since
the
amplitude
cloud layer (solid line value equals to 1). Since the amplitude difference of the first derivative of lidar
difference
of theby
first
derivative
lidar
signal induced
cloud, aerosol
andcases,
background
signal induced
cloud,
aerosolofand
background
noise by
is significant
in most
the finalnoise
cloudis
significant
in most
cases,
final cloud
layers shown
in thethreshold
right diagram
are selected
using
three
layers shown
in the
rightthe
diagram
are selected
using three
conditions
mentioned
above.
After the conditions
determination
of CBH, the
cloudAfter
top height
is determined where
the the
lidarcloud
signaltop
is less
thanis
threshold
mentioned
above.
the determination
of CBH,
height
determined
where
the dP(r)
lidar /signal
than the
and
When
the
r)/dr
|dP(and
| ≤ K + ne. the
dr ≤ Kis+less
ne . When
the cloud base’s
and
thecloud
cloud base’s
is optical
thick
impenetrable,
point
cloud
is
optical
thick
and
impenetrable,
the
point
with
a
minimum
value
above
the
cloud
peak
height
with a minimum value above the cloud peak height in the candidate area is regarded as the cloud
intop
theheight.
candidate
area isthat
regarded
as theand
cloud
top height.inItadjacent
is notedcloud
that comparison
and
It is noted
comparison
combination
layers should
be combination
done when
inthe
adjacent
cloud
layers structure.
should be done when the cloud has complex structure.
cloud has
complex
Figure3.3.Flow
Flowchart
chart of
of cloud
cloud properties
properties retrieval
retrieval based
Figure
based on
on improved
improvedzero-crossing
zero-crossingmethod.
method.
Atmosphere 2017, 8, 9
Atmosphere 2017, 8, 9
6 of 17
6 of 6
5
5
5
dp/dr
candidate
Cb and Cp
Altitude[km]
Jul.15 13:28
4.5
4.5
4
4
4
3.5
3.5
3.5
2.92
2.92
2.61
2.49
2.49
2.61
2.49
1.97
1.97
2
1.5
1.5
1.5
1
1
1
dp/dr
Cb and Cp
4.5
cloud peak
0.5 0
10
10
1
10
2
10
（a）Lidar signal [a.u.]
3
0.5
-2
0
2
（b）Scaled signal [a.u.]
4
0.5
-2
cloud base
0
2
（c）Result [a.u.]
4
Figure 4. Case
Case study
study using
using improved
improved differential zero-crossing method [22].
Figure
Table 3. Tuning
Tuning standard of threshold value in different
Table
different altitude
altitude and
and time
time range.
range.
Altitude
Altitude
0.5–5 km
0.5–5 km
5–10
km
5–10 km
>10>10
kmkm
Time
Range
Time Range
daytime
daytime
nighttime
nighttime
daytime
daytime
nighttime
nighttime
daytime
daytime
nighttime
nighttime
Threshold
Value a
Threshold Value a
2–4
2–4
1–1.5
1–1.5
20–30
20–30
1–1.5
1–1.5
10–20
10–20
2.2.2.
Method Using
Using Radiosonde
Radiosonde
2.2.2. Relative
Relative Humidity
Humidity (RH)
(RH) Threshold
Threshold Method
Radiosonde
withhigh
high
vertical
resolution
has widely
been widely
used atmospheric
to obtain atmospheric
Radiosonde with
vertical
resolution
has been
used to obtain
parameters
parameters
including
those
of
cloud
vertical
structure.
Methods
have
been
developed
determine
including those of cloud vertical structure. Methods have been developed to determinetolocations
of
locations
of cloud
layers
and theirfrom
boundaries
from
radiosonde [24,25].
measurement
[24,25].
According
to
cloud layers
and their
boundaries
radiosonde
measurement
According
to the
radiosonde
the
radiosonde and
characteristics
and Nagquconditions,
atmospheric
the WR95
is used
to
characteristics
Nagqu atmospheric
theconditions,
WR95 method
[23] ismethod
used to[23]
obtain
cloud
obtain
cloud
vertical
structure.
The
cloud
base
and
top
locations
are
identified
using
the
following
vertical structure. The cloud base and top locations are identified using the following criteria:
criteria:
maximum
RH)(>RH
> 87%; minimum
( RHmin )and
> 84%;
the
cloud
base
of thelayer
lowest
max )minimum
maximum
RH (RHmax
87%;
RH (RHRH
the and
cloud
base
of the
lowest
of
min ) > 84%;
layer
is set
500 m
AGL. Moreover,
the minimum
of low-middle
and high
cloudof
is cloud
set at 500
mat
AGL.
Moreover,
the minimum
thicknessthickness
of low-middle
cloud andcloud
high cloud
are
cloud
aremset
to 60
30 m,
m and
60 m, respectively,
andofthe
range of
adjacent
clouds
should
be larger
set to 30
and
respectively,
and the range
adjacent
clouds
should
be larger
than
300 m.than
It is
300
m.that
It isanoted
that a transformation
RH with
respect
to ice
at levels
where the temperature
noted
transformation
of RH with of
respect
to ice
at levels
where
the temperature
is below 0 ◦ is
C
below
°Cconsidered,
should be considered,
that is,
should0be
that is,
RH
RH × (Ew /Ei )
(1)
new =
RH
(1)
new = RH × (E w / E i )
where Ew and Ei are saturation vapor pressure of water and ice, respectively; and RH is the measured
where E w and E i are saturation vapor pressure of water and ice, respectively; and RH is the
relative humidity of radiosonde. The calculation of Ew and Ei are based on GG46 and GG46i
measured relative humidity of radiosonde. The calculation of E w and E i are based on GG46 and
equations [26,27], respectively,
GG46i equations [26,27], respectively,
log10 Ew = 10.79574(1 − TT1 ) −T5.02800 log10 ( TT T) + 1.50475 × 10−4 [1 − 108.2969(1−T/T1 ) ]
1
1
(2)
log10 −
5.02800log
) + 1.50475 ×10−4 [1 − 108.2969(1−T/T ) ]
) ] + 0.78614
10 (
− 104.76955−(1T−)T−1 /T
−0.42873 ×
10 E3w[1= 10.79574(1
T
(2)
1
1
−0.42873 ×10−3 [1 − 104.76955(1−T1 /T) ] + 0.78614
T
T
T
log10 Ei = 9.09685(1 − 1 ) − 3.56654 log10 ( 1 ) + 0.87682(1 −
) + 0.78614
T T1
T
T
T1
T
1
log10 E i = 9.09685(1 −
) − 3.56654 log10 (
) + 0.87682(1 −
) + 0.78614
(3)
(3)
T
T1
where T1 = 273.15 K, T1 = 273.15 + t,Tand t is centigrade
temperature.
Then, the cloud structure
below 0T◦ C= can
be determined
from
(1)–(3). temperature. Then, the cloud structure below
273.15K
+ t ,Equations
, T = 273.15
and t is centigrade
where
1
1
0 °C can be determined from Equations (1)–(3).
Atmosphere 2017, 8, 9
Atmosphere 2017, 8, 9
7 of 17
7 of 7
3.
3. Results
Results
3.1. Cloud Characteristics Statistics of CL31
The data acquirability
acquirability of CL31 is firstly analyzed to guarantee sufficient samples and the
accuracy
accuracy of follow-up
follow-up work
work since
since there
there may exist data missing
missing due
due to mistaken
mistaken operative
operative mode,
mode,
although
continuous
observation
mode
during
the the
whole
experiment
period.
The
although CL31
CL31 isisinina a24-h
24-h
continuous
observation
mode
during
whole
experiment
period.
time
resolution
of CL31
is 16 s,
12,600
data size
in size
the 56-day
observations.
Figure
The time
resolution
of CL31
is thus
16 s,one
thushour
one has
hour
has 12,600
data
in the 56-day
observations.
5a
is the5a
data
acquirability
per hour,per
andhour,
the x-axis
each 1-h interval
overinterval
24 h, forover
instance,
Figure
is the
data acquirability
and represents
the x-axis represents
each 1-h
24 h,
the
point
at
1
represents
the
data
acquirability
between
00:00
and
01:00
Local
Standard
Time
(LST,
for instance, the point at 1 represents the data acquirability between 00:00 and 01:00 Local Standard
LST
UTCLST
+ 8 h).
It can
that
data
at 00:00–09:00
are all 100%
with
no with
data
Time=(LST,
= UTC
+ 8beh).seen
It can
bethe
seen
thatacquirability
the data acquirability
at 00:00–09:00
are all
100%
deficiency,
and the others
areothers
all above
95%.
Thus,
the data
missing
level
is endurable
reasonable,
no data deficiency,
and the
are all
above
95%.
Thus,
the data
missing
level isand
endurable
and
which
has
a
negligible
impact
on
the
following
statistical
results.
reasonable, which has a negligible impact on the following statistical results.
100
Cloudless
One-layer cloud
Two-layer cloud
Three-layer cloud
Vertical Visibility
Probability (% )
99
98
3% 1%
19%
12%
97
96
95
94
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
Time (LST)
65%
Figure
Figure 5.
5. CL31
CL31 (a)
(a) Data
Data acquirability;
acquirability; and
and (b)
(b) the
the probability
probability distribution
distribution of
of cloudless
cloudless (red),
(red), one-layer
one-layer
cloud
(green),
two-layer
cloud
(blue),
three-layer
cloud
(yellow)
and
vertical
visibility
(black)
cloud (green), two-layer cloud (blue), three-layer cloud (yellow) and vertical visibility (black) during
during
the
the experiment
experiment period.
period.
It can be seen in Figure 5b that the cloud occurrence is about 81% during the experiment period,
It can be seen in Figure 5b that the cloud occurrence is about 81% during the experiment period,
and the one-layer cloud has the highest occurrence frequency of 65% with multi-layer cloud
and the one-layer cloud has the highest occurrence frequency of 65% with multi-layer cloud frequency
frequency of 13%, so the cloud structure at Nagqu in summer is relatively simple. It is noted that
of 13%, so the cloud structure at Nagqu in summer is relatively simple. It is noted that sometimes CL31
sometimes CL31 “cloud status” is “4”, representing the atmospheric condition is full obscuration
“cloud status” is “4”, representing the atmospheric condition is full obscuration determined but no
determined but no cloud base detected. In this case, the results represent “vertical visibility as
cloud base detected. In this case, the results represent “vertical visibility as calculated” and “highest
calculated” and “highest signal detected”. During the experiment period, the occurrence of “4”
signal detected”. During the experiment period, the occurrence of “4” almost happens in the rainy
almost happens in the rainy weather, its diurnal variation is obvious, the daytime occurrence is rare
weather, its diurnal variation is obvious, the daytime occurrence is rare but has significant increase at
but has significant increase at night and before dawn, which corresponds to the special weather
night and before dawn, which corresponds to the special weather phenomenon of “sunny daytime,
phenomenon of “sunny daytime, rainy nighttime” at Nagqu.
rainy nighttime” at Nagqu.
The cloud occurrence of CL31 is defined as the ratio of the cloudy profiles and the total profiles.
The cloud occurrence of CL31 is defined as the ratio of the cloudy profiles and the total profiles.
The occurrence frequency of a specific layer cloud (e.g., one-layer cloud) at specific time interval is
The occurrence frequency of a specific layer cloud (e.g., one-layer cloud) at specific time interval is
defined as ratio of the profiles having the specific layer cloud and the total profiles. It can be seen in
defined as ratio of the profiles having the specific layer cloud and the total profiles. It can be seen in
Figure 6 that the averaged cloud occurrence is between 70% and 90% each day, which is consistent
Figure 6 that the averaged cloud occurrence is between 70% and 90% each day, which is consistent
with the results from millimeter-wave radar and satellite data [13,28]. It can be also seen that the
with the results from millimeter-wave radar and satellite data [13,28]. It can be also seen that the
diurnal variation of cloud occurrence is obvious, that is, the cloud occurrence has a minimum value
diurnal variation of cloud occurrence is obvious, that is, the cloud occurrence has a minimum value
between 13:00 and 14:00 LST, and with the enhancement of the surface radiation, the atmospheric
between 13:00 and 14:00 LST, and with the enhancement of the surface radiation, the atmospheric
activities become more convective with cloud occurrence rapidly increasing. Moreover, the
activities become more convective with cloud occurrence rapidly increasing. Moreover, the occurrence
occurrence frequency of multilayer cloud is relative low and stable, on the other hand, the one-layer
frequency of multilayer cloud is relative low and stable, on the other hand, the one-layer cloud
cloud occurrence frequency has an obvious variation with a maximum proportion.
occurrence frequency has an obvious variation with a maximum proportion.
Atmosphere2017,
2017,8,8,99
Atmosphere
8 of
8 of
178
90
80
C loud C over (% )
70
60
50
One-layer cloud
Two-layer cloud
Three-layer cloud
All
40
30
20
10
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
Time (LST)
Figure
Figure6.6.Averaged
Averageddiurnal
diurnalvariation
variationof
ofthe
thecloud
cloudoccurrence
occurrence(black),
(black),one-layer
one-layercloud
cloud(red),
(red),two-layer
two-layer
cloud
cloud(green),
(green),and
andthree-layer
three-layercloud
cloud(blue)
(blue)ofofCL31
CL31during
duringthe
themeasurement
measurementperiod.
period.
3.2.CBH
CBHComparison
Comparisonbetween
betweenCL31
CL31and
andWACAL
WACAL
3.2.
Comparisonsbetween
betweendifferent
differenttypes
typesof
ofinstruments
instrumentshave
haveshown
showntypical
typicalfeatures
featuresof
ofinstruments
instruments
Comparisons
and provided
provided many
many different phenomena of
of
and
of clouds
clouds [29–31].
[29–31]. Feister
Feisteretetal.al.[29]
[29]made
madea acomparison
comparison
macroscopic
cloud
data
from
ground-based
measurements
using
VIS/NIR
andand
IR instruments
and
of
macroscopic
cloud
data
from
ground-based
measurements
using
VIS/NIR
IR instruments
showed
that that
CBHCBH
differences
between
the instruments
in theinlower
heightheight
range range
are mainly
due to
and
showed
differences
between
the instruments
the lower
are mainly
the different
principlesprinciples
of measurement
including the
geometries
measurements,
the assumption
due
to the different
of measurement
including
theof geometries
of measurements,
on
lapse
rate
used
in
the
Nubiscope
cloud
algorithm,
and
the
definition
of
CBH
for
the respective
the assumption on lapse rate used in the Nubiscope cloud algorithm, and the definition
of CBH
instrument.
Liu et al.
[30] used two
(CL31,two
CL51)
and a whole-sky
cloud-measuring
for
the respective
instrument.
Liuceilometers
et al. [30] used
ceilometers
(CL31, infrared
CL51) and
a whole-sky
system cloud-measuring
and found that even
if and
the same
different
types with
may different
introducetypes
different
infrared
system
foundceilometers
that even if with
the same
ceilometers
may
observation
results
of
CBH.
Although
the
commercial
ceilometer,
like
Vaisala
CL31,
has
been
introduce different observation results of CBH. Although the commercial ceilometer, like Vaisala CL31,
extensively
used in cloud
thestudy,
CBH the
retrieval
still need still
to beneed
investigated,
especially
has
been extensively
used study,
in cloud
CBH algorithm
retrieval algorithm
to be investigated,
to
look
into
the
data
from
different
apparatus.
especially to look into the data from different apparatus.
Thecomparison
comparisonof
offirst
firstCBH
CBH(CBH1)
(CBH1) between
between CL31
CL31 and
and WACAL
WACALhas
has been
been analyzed
analyzed in
in Figure
Figure 7.
7.
The
Notedthat
thatCL31
CL31has
hasthe
theability
abilityof
of24-h
24-hcontinuous
continuousobservation
observation while
while the
the WACAL
WACALisisonly
onlyoperated
operated
Noted
at
about
09:00–12:00,
14:00–19:00,
and
20:00–23:00
each
day
except
for
the
passes
of
CALIPSO,
the
at about 09:00–12:00, 14:00–19:00, and 20:00–23:00 each day except for the passes of CALIPSO, the
matching time should
WACAL
sampling
time.
Thus,
the data
4 min 4are
chosen
matching
should be
bebased
basedononthe
the
WACAL
sampling
time.
Thus,
the every
data every
min
are
to compose
the dataset
with the
number
of 2610.ofIn2610.
Figure
the red
solid
andline
light
blue
dashed
chosen
to compose
the dataset
with
the number
In 7,
Figure
7, the
redline
solid
and
light
blue
fdata - ydata
line represent
the linear
fit for
all all
dataset
and
dashed
line represent
the linear
fit for
dataset
andthe
thedata
data excluding
excluding |fdata-ydata
1.0 *∗SD
SD,,
| >> 1.0
respectively,
where
fdata
is
the
fitted
value,
ydata
is
the
corresponding
CL31
CBH,
and
SD
represents
respectively, where fdata is the fitted value, ydata is the corresponding CL31 CBH, and SD represents
the
thestandard
standarddeviation
deviationof
ofydata.
ydata. The
The black
black dashed
dashed line
line is
is the
the weighted
weighted least
least square
square fit
fit for
for all
all dataset.
dataset.
The
black
and
pink
dots
represent
fdata-ydata
within
and
without
the
standard
deviation
of
|
|
The black and pink dots represent fdata - ydata within and without the standard deviation ofydata.
ydata.
The occurrence of the pink dots is mainly the result of different cloud layer comparison for these
The occurrence of the pink dots is mainly the result of different cloud layer comparison for these two
two devices. Non-strict time comparison may result in deviation in this process; moreover, when it
devices. Non-strict time comparison may result in deviation in this process; moreover, when it occurs
occurs due to low thin cloud, CL31 sometimes has fault decision by mistake, resulting in different
due to low thin cloud, CL31 sometimes has fault decision by mistake, resulting in different cloud
cloud layer comparison for these two devices. Basically, the CBH of CL31 and WACAL shows a good
layer comparison for these two devices. Basically, the CBH of CL31 and WACAL shows a good
consistency with the correlation of 0.82 and standard deviation of 0.99 km. After excluding the singular
consistency with the correlation of 0.82 and standard deviation of 0.99 km. After excluding the
value in pink dots, the correlation and standard deviation has been increased up to 0.92 and decreased
singular value in pink dots, the correlation and standard deviation has been increased up to 0.92 and
to 0.64 km, respectively.
decreased to 0.64 km, respectively.
Atmosphere 2017, 8, 9
Atmosphere 2017, 8, 9
9 of 17
9 of 9
Figure
Figure7.7.Scatter
Scatterdiagram
diagramofofthe
theCBH1
CBH1measured
measuredby
byWACAL
WACALand
andCL31.
CL31.
From
Fromthe
thefrequency
frequencydistribution
distributionof
ofthe
theCBH
CBHdifferences
differencesshown
shownin
inFigure
Figure8a,
8a,ititcan
canbe
befound
foundthat
that
about
46%
of
the
differences
are
within
±200
m,
and
about
80%
of
the
differences
are
within
±1000
about 46% of the differences are within ±200 m, and about 80% of the differences are within ±1000m.
m.
The
Thedeviation
deviationhas
hasan
anobvious
obviousleft-skewed
left-skeweddistribution;
distribution;that
thatis,
is,the
thecondition
conditionthat
thatWACAL
WACALCBH
CBHisis
smaller
smallerthan
thanCL31’s
CL31’soccurs
occursmuch
muchmore
moreoften
oftenthan
thanthe
theone
onethat
thatCL31
CL31CBH
CBHisissmaller
smallerthan
thanWACAL’s.
WACAL’s.
Figure
8b
shows
the
frequency
distributions
of
CBH
retrieved
with
WACAL
and
CL31
for
the
Figure 8b shows the frequency distributions of CBH retrieved with WACAL and CL31 for thedataset,
dataset,
respectively.
forall
allCBH
CBHtogether
togetherhas
hasthe
the
maximum
cantered
at 1000 m–
respectively. The
The histogram
histogram for
maximum
forfor
thethe
binbin
cantered
at 1000–2000
m,
2000
andashows
a progressive
decreasing
frequency
to the maximum
level of detection
of the
and m,
shows
progressive
decreasing
frequency
up to theupmaximum
level of detection
of the WACAL
WACAL
and
CL31.
is clear
that the
cloud
layers detected
by WACAL
much
more
than
CL31
in
and CL31.
It is
clearItthat
the cloud
layers
detected
by WACAL
are muchare
more
than
CL31
in the
range
the
range
of
1000
m–2000
m.
of 1000–2000 m.
In
Inview
viewof
ofthe
theabove
abovecondition,
condition,aacase
casestudy
studyatat20:00–23:00,
20:00–23:00,12
12August
Augusthas
hasbeen
beenanalyzed.
analyzed.Figure
Figure99
shows
showsthe
theTime–Height–Intensity
Time–Height–Intensitydiagram
diagramofofWACAL
WACALobservation,
observation,where
wherethe
theblack
blackdots
dotsand
andblue
blue
dots
represent
the
retrieved
CBH
by
WACAL
and
CL31,
respectively.
It
can
be
found
that
there
dots represent the retrieved CBH by WACAL and CL31, respectively. It can be found that thereexists
exists
obvious
obviousdifference
differencebetween
betweenthese
thesetwo
tworesults.
results.This
Thismay
maybe
bethe
theresult
resultofofthe
thedifference
differenceon
onthe
theCBH
CBH
definition
Specifically,
thethe
improved
differential
zero-crossing
method
of
definitionand
andretrieval
retrievalmethods
methods[26].
[26].
Specifically,
improved
differential
zero-crossing
method
WACAL regards the place where the first derivative of the lidar backscatter power changes from
of WACAL regards the place where the first derivative of the lidar backscatter power changes from
negative to positive as the CBH, while the CL31 regards the strongest backscattered signal point as
negative to positive as the CBH, while the CL31 regards the strongest backscattered signal point as
the CBH.
the CBH.
The cloud structure in this case is also retrieved using simultaneous radiosonde data and shown
The cloud structure in this case is also retrieved using simultaneous radiosonde data and shown in
in Figure 10. The red and blue solid line represent the RH measured from the radiosonde and
Figure 10. The red and blue solid line represent the RH measured from the radiosonde and calculated
calculated from the saturation vapor pressure of ice using Equations (1)–(3), respectively. The CBH
from the saturation vapor pressure of ice using Equations (1)–(3), respectively. The CBH is 2.71 km
is 2.71 km using WR95 method mentioned in Section 2.2.2, and shows a good consistency with the
using WR95 method mentioned in Section 2.2.2, and shows a good consistency with the WACAL result
WACAL result shown in Figure 9. It can be indicated that CL31 sometimes overestimates the CBH
shown in Figure 9. It can be indicated that CL31 sometimes overestimates the CBH compared with
compared with WACAL and radiosonde, which to some extent explains the situation that WACAL
WACAL and radiosonde, which to some extent explains the situation that WACAL CBH is smaller
CBH is smaller than CL31’s much more often than CL31 CBH is smaller than WACAL’s.
than CL31’s much more often than CL31 CBH is smaller than WACAL’s.
Atmosphere 2017, 8, 9
Atmosphere 2017, 8, 9
Atmosphere 2017, 8, 9
10 of 17
10 of 10
10 of 10
0.3
0.3
Frequency
Frequency
0.25
0.25
0.2
0.2
0.15
0.15
0.1
0.1
0.05
0.05
-2
-2
-1
0
1
-1
0
1
Difference (Lidar-CL31)
(km)
Difference (Lidar-CL31) (km)
800
800
800
800
600
600
600
600
Number
Number
Number
Number
0
0-3
-3
400
400
200
200
0
00
0
2
2
3
3
400
400
200
200
1
1
2 3 4 5 6
2Lidar
3 CBH1
4 5(km)
6
Lidar CBH1 (km)
7
7
8
8
0
00
0
1
1
2 3 4 5 6
2CL31
3 CBH
4 5(km)6
CL31 CBH (km)
7
7
8
8
Figure 8.
8. (a)
(a)CBH1
CBH1 differencebetween
between WACAL;
and
(b)
distribution
of the
CBH1
of WACAL
(left)
Figure
and
(b)(b)
distribution
of the
CBH1
of WACAL
(left)(left)
and
CBH1 difference
difference betweenWACAL;
WACAL;
and
distribution
of the
CBH1
of WACAL
and
CL31
(right)
during
the
measurement
period.
CL31CL31
(right)
during
the measurement
period.
and
(right)
during
the measurement
period.
Figure 9.
9. Atmospheric
Atmospheric backscattered
signal
of
WACAL
at 20:21–22:54,
20:21–22:54, 12
12 August
August 2014
2014 and
and CBH1
CBH1
Figure
backscattered signal
signal of
of WACAL
WACAL at
at
Figure
9. Atmospheric
backscattered
20:21–22:54, 12
August 2014
and CBH1
measured by
by WACAL(black)
(black) andCL31
CL31 (blue).
measured
measured
by WACAL
WACAL (black) and
and CL31 (blue).
(blue).
Atmosphere
Atmosphere2017,
2017,8,8,99
Atmosphere 2017, 8, 9
11
11of
of17
11
11 of 11
7
7
RH-Water
RH-Water
RH-Ice
RH-Ice
Min-RH
Min-RH
Max-RH
Max-RH
CBH(2.71km)
CBH(2.71km)
6
6
HeightAGL(km)
AGL(km)
Height
5
5
44
33
22
11
00
00
10
10
20
30
40
40
50
50
RH
RH(%)
(%)
60
60
7070
8080
9090
100
100
Figure10.
10.Radiosonde
Radiosonderelative
humidity
(RH)
Figure
10.
Radiosonde
relative humidity
humidity (RH)
(RH)profile
profileand
andresults
resultsof
ofobserved
observedcloud
cloudstructure.
structure.
Figure
profile
and
results
of
observed
cloud
structure.
Discussion
4.4.Discussion
Discussion
4.
4.1.Cloud
CloudCharacteristics
CharacteristicsDiurnal
DiurnalVariation
Variation Comparison
Comparison
between
4.1.
Cloud
Characteristics
Diurnal
Variation
Comparisonbetween
betweenCL31
CL31and
andWACAL
WACAL
4.1.
CL31
and
WACAL
SinceWACAL
WACAL
is not
not
24-h
continuous
observation
CL31,
the
should
bebe
Since
WACAL
24-h
continuous
observation
like
CL31,
thedata
dataacquirability
acquirability
should
Since
is is
not
24-h
continuous
observation
likelike
CL31,
the data
acquirability
should
be firstly
firstly
analyzed
to
make
the
further
work
more
reliable.
Figure
11
shows
the
data
acquirability
per
firstly analyzed
tothe
make
the further
workreliable.
more reliable.
11 shows
theacquirability
data acquirability
per
analyzed
to make
further
work more
FigureFigure
11 shows
the data
per hour,
hour,and
andthe
thex-axis
x-axis represents
represents each
each 1-h
interval
over
24
h,
which
isissame
asasthe
Figure
5a.
The
time
hour,
1-h
interval
over
24
h,
which
same
the
Figure
5a.
The
time
and the x-axis represents each 1-h interval over 24 h, which is same as the Figure 5a. The time resolution
resolutionof
of WACAL is
is 16 s,
s, thus 11 h
8325
data size
from 8 8July
toto
resolution
h has
hassize
8325
sizein
inthe
the37-day
37-dayobservations
observations
July
of
WACAL is WACAL
16 s, thus 1 16
h hasthus
8325 data
in data
the 37-day
observations
from 8 July from
to 15 August.
15August.
August.In
Inthis
this statistic,
statistic, the
the hour
hour with
observed
number
larger
than
isischosen
toto
make
15
with total
total
observed
number
larger
than4000
4000
chosen
make
In
this statistic,
the hour with
total observed
number
larger
than 4000
is chosen
to make
comparison
comparison
with
CL31.
As
a
result,
10
h,
at
10:00,
11:00,
12:00,
16:00,
17:00,
18:00,
19:00,
20:00,
21:00
comparison
with
CL31.
As
a
result,
10
h,
at
10:00,
11:00,
12:00,
16:00,
17:00,
18:00,
19:00,
20:00,
21:00
with CL31. As a result, 10 h, at 10:00, 11:00, 12:00, 16:00, 17:00, 18:00, 19:00, 20:00, 21:00 and 22:00,
and 22:00, are selected as WACAL dataset for diurnal comparison. Moreover, the time period of
andselected
22:00, are
selected as
WACAL
dataset comparison.
for diurnal comparison.
Moreover,
the of
time
period of
are
as WACAL
dataset
for diurnal
Moreover, the
time period
10:00–12:00,
10:00–12:00, 16:00–19:00, and 20:00–22:00 can represent the cloud characteristics at forenoon,
10:00–12:00,and
16:00–19:00,
represent
the cloud
characteristics
at and
forenoon,
16:00–19:00,
20:00–22:00and
can 20:00–22:00
represent thecan
cloud
characteristics
at forenoon,
afternoon,
night
afternoon, and night period, respectively.
afternoon,
and
night
period,
respectively.
period, respectively.
8000
8000
7000
7000
6000
6000
Number
Number
5000
5000
4000
4000
3000
3000
2000
2000
1000
1000
0
0
0
1
2
3
4
5
6
7
8
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
0
1
2
3
4
5
6
7
8
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
Time (LST)
Time (LST)
Figure 11. WACAL data acquirability during measurement period.
Figure
Figure 11.
11. WACAL
WACALdata
dataacquirability
acquirabilityduring
duringmeasurement
measurementperiod.
period.
Figure 12a shows the averaged diurnal variation of CBH spatial distribution measured by CL31.
The Figure
CBH distribution
proportion
at diurnal
a specific
time period
andspatial
height distribution
range, for instance,
from
12a shows the
averaged
variation
of CBH
measured
by00:00
CL31.
The CBH distribution proportion at a specific time period and height range, for instance, from 00:00
Atmosphere 2017, 8, 9
12 of 17
Figure
12a
Atmosphere
2017,
8, 9shows
the averaged diurnal variation of CBH spatial distribution measured by12CL31.
of 12
The CBH distribution proportion at a specific time period and height range, for instance, from 00:00 LST
LST
to 01:00
atkm,
1–2 is
km,
the ratio
the profiles
CBH occurs
1–2
thecloud
total cloud
to 01:00
LST LST
at 1–2
theisratio
of theofprofiles
wherewhere
CBH occurs
at 1–2at
km
tokm
the to
total
layers
layers
at 00:00–01:00
LST.that
Notethe
that
the statistic
is limited
the range
m to
kmtodue
the
at 00:00–01:00
LST. Note
statistic
is limited
in the in
range
from 0from
m to07.5
km7.5
due
theto
CL31
CL31
detection
limitation.
be found
that
themainly
CBH mainly
occurred
3 km
with
detection
range range
limitation.
It can It
becan
found
that the
CBH
occurred
below 3below
km with
distinct
distinct
variation.
Specifically,
CBH is generally
lower,
below
2 km, at forenoon.
With the
diurnal diurnal
variation.
Specifically,
the CBH the
is generally
lower, below
2 km,
at forenoon.
With the increasing
increasing
surface radiation,
thethermodynamic
Plateau thermodynamic
process becomes
moreatactive
at afternoon
of surface of
radiation,
the Plateau
process becomes
more active
afternoon
so that
so
that
the
CBH
is
gradually
lifted
and
high
cloud
begins
to
appear
with
a
wider
spatial
scale.
By
the CBH is gradually lifted and high cloud begins to appear with a wider spatial scale. By evening,
evening,
thedistribution
vertical distribution
cloud maximum
reaches maximum
value. Generally,
the variation
diurnal variation
the vertical
of cloud of
reaches
value. Generally,
the diurnal
of CBH
of
CBH distribution
spatial distribution
hasshape;
a “U” that
shape;
high mainly
cloud mainly
appears
at
spatial
has a “U”
is, that
highis,cloud
appears
beforebefore
dawn dawn
and atand
night,
night,
while,
conversely,
the
low-middle
cloud
is
the
main
cloud
type
during
daytime.
Figure
12b
is
while, conversely, the low-middle cloud is the main cloud type during daytime. Figure 12b is the
the
averaged
diurnal
variation
CBH
spatialdistribution
distributionmeasured
measuredby
byWACAL.
WACAL.The
The cloud analyzed
averaged
diurnal
variation
of of
CBH
spatial
analyzed
range
range is
is selected
selected as
as 0.5
0.5 km–7.5
km–7.5 km
km to make the data consistency
consistency with CL31’s.
CL31’s. It
It can
can be
be seen
seen that
that these
these
two
results
have
quite
uniformity
in
distribution
range,
variation
trend.
A
similar
U
shape
distribution
two results have quite uniformity
A similar U shape distribution
was
was also found
found using
using concurrent
concurrent cloud
cloud radar
radar measurement
measurement during
during the
the field
field experiment
experiment [13],
[13], and
and the
the
distance
between
cloud
radar,
ceilometer
and
Lidar
is
about
tens
of
meters,
which
can
be
regarded
distance between cloud radar, ceilometer and Lidar is about tens of meters, which can be regarded
as
as the same area. Furthermore,
Furthermore, the
the historical
historical radiosonde
radiosonde observations
observations from
from 136
136 ground-reference
ground-reference
meteorological
meteorological stations
stations between
between 1987
1987 and
and 2001
2001 in
in China
China region
region have
have shown
shown that
that the
the diurnal
diurnal change
change
of
cloud
occurrence
has
five
types,
single-peak
in
the
morning,
single-peak
in
the
afternoon,
double
of cloud occurrence has five types, single-peak in
morning, single-peak in the afternoon, double
peaks
peaks with
with higher
higher peak
peak in
in the
the morning,
morning, double
double peaks
peaks with
with higher
higher peak
peak in
in the
the afternoon
afternoon and
and double
double
peaks with
same
value
[32].
Unfortunately,
there
is
no
ground-reference
meteorological
station
in
with same value [32]. Unfortunately, there is no ground-reference meteorological
in the
the
whole Tibetan Plateau. The
reasons
for
the
special
U
shape
distribution
cannot
be
explained
well
The reasons
the special U shape distribution cannot be explained well yet
yet
and
and should
should be
be further
further analyzed.
8
8
7
7
0.5
3
0.2
5
4
3
0.4
0.3
Frequency
0.3
Frequency
4
Height AGL (km )
Height AGL (km )
0.4
5
0.2
2
2
0.1
1
0
0:00
0.5
6
6
2:00
4:00
6:00
8:00
10:00 12:00 14:00 16:00 18:00 20:00 22:00
Time (LST)
0
1
0
0:00 2:00 4:00 6:00 8:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00
Time (LST)
0.1
0
Figure
diurnal
variation
of CBH
spatial
distribution
measured
by: (a) by:
CL31;(a)and
(b)
Figure 12.
12. Averaged
Averaged
diurnal
variation
of CBH
spatial
distribution
measured
CL31;
WACAL
during
measurement
period.
and (b) WACAL during measurement period.
Figure 13a shows the averaged diurnal variation of CBH temporal distribution measured by
Figure 13a shows the averaged diurnal variation of CBH temporal distribution measured by
CL31. The CBH frequency at a specific time period and height range, for instance, from 00:00 LST to
CL31. The CBH frequency at a specific time period and height range, for instance, from 00:00 LST to
01:00 LST at 1–2 km, is the ratio of the profiles where CBH occurs at 1–2 km to the total profiles at
01:00 LST at 1–2 km, is the ratio of the profiles where CBH occurs at 1–2 km to the total profiles at
00:00–01:00 LST. It can be found that the diurnal variation of CBH frequency is similar to the CBH
00:00–01:00 LST. It can be found that the diurnal variation of CBH frequency is similar to the CBH
distribution proportion diurnal variation. In Figure 13a, the CBH frequency of low-middle cloud is
distribution proportion diurnal variation. In Figure 13a, the CBH frequency of low-middle cloud is
obviously higher than the high cloud’s, and the high value area of low cloud occurs at 06:00–15:00.
obviously higher than the high cloud’s, and the high value area of low cloud occurs at 06:00–15:00.
From 16:00 to the next morning, the CBH frequency of low cloud is relatively stable, ranging from
From 16:00 to the next morning, the CBH frequency of low cloud is relatively stable, ranging from
20% to 30%, while the CBH frequency of high cloud begins to increase gradually and reaches its
20% to 30%, while the CBH frequency of high cloud begins to increase gradually and reaches its
maximum at night. Figure 13b is the averaged diurnal variation of CBH frequency measured by
maximum at night. Figure 13b is the averaged diurnal variation of CBH frequency measured by
WACAL. Compared with CL31, both of them have a good consistency in distribution range and
WACAL. Compared with CL31, both of them have a good consistency in distribution range and
variation trend.
variation trend.
Atmosphere 2017, 8, 9
Atmosphere 2017, 8, 9
13 of 17
13 of 13
8
8
7
7
0.5
0.3
3
0.2
5
4
3
0.4
Frequency
4
Frequency
0.4
5
Height AGL (km)
Height AGL (km)
0.5
6
6
0.3
0.2
2
2
0.1
1
0
0:00
2:00
4:00
6:00
8:00
10:00 12:00 14:00 16:00 18:00 20:00 22:00
Time (LST)
0
1
0
0:00 2:00 4:00 6:00 8:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00
Time (LST)
0.1
0
Figure
distribution
measured
by:by:
(a) (a)
CL31;
andand
(b)
Figure 13.
13. Averaged
Averageddiurnal
diurnalvariation
variationofofCBH
CBHtemporal
temporal
distribution
measured
CL31;
WACAL
during
measurement
period.
(b) WACAL during measurement period.
4.2.
4.2. Cloud
Cloud Characteristics
Characteristics Statistics
Statistics of
of WACAL
WACAL
Since
CL31,
thethe
cloud
below
7.5 7.5
km km
can can
onlyonly
be measured,
and
Since the
the detection
detectionrange
rangelimitation
limitationofof
CL31,
cloud
below
be measured,
the
other
cloud
features
such
as
the
cloud
top
height
and
thickness
are
unattainable.
However,
the
and the other cloud features such as the cloud top height and thickness are unattainable. However,
WACAL
can
fill
these
gaps
because
of
its
higher
laser
transmitting
power
and
receiving
ability,
the WACAL can fill these gaps because of its higher laser transmitting power and receiving ability,
detecting
wider range
range and
andretrieving
retrievingaccurately
accuratelythe
thecloud
cloudfeatures
featuresininaddition
addition
the
CBH.
According
detecting wider
toto
the
CBH.
According
to
to
the
results
in
Figure
11,
the
selected
three
time
periods
can
represent
different
cloud
development
the results in Figure 11, the selected three time periods can represent different cloud development
processes,
cloud spatial-temporal
spatial-temporal distribution
be analyzed
analyzed
processes, so
so the
the cloud
distribution based
based on
on these
these three
three periods
periods can
can be
and
and compared
compared using
using WACAL
WACAL data.
data. Figure
Figure 14
14 shows
shows the
the averaged
averaged spatial
spatial distribution
distribution of
of one-,
one-, two-,
two-,
threeand
four-layer
clouds
measured
by
WACAL
at
forenoon,
afternoon
and
night,.
The
CBT,
three- and four-layer clouds measured by WACAL at forenoon, afternoon and night,. The CBT, CTH,
CTH,
ΔT(shown
(shownwith
withgreen
greencolor
colorininFigure
Figure14),
14),and
andN
Nrepresent
representthe
the averaged
averaged cloud
cloud base
base height,
∆T
height, averaged
averaged
cloud
top
height,
averaged
cloud
thickness
and
the
occurrence
number
of
the
specific
cloud
cloud top height, averaged cloud thickness and the occurrence number of the specific cloud type,
type,
respectively.
respectively. The
The value
value in
in bracket
bracket presents
presents the
the corresponding
corresponding occurrence
occurrence frequency
frequency of
of the
the specific
specific
cloud
thinner cloud
cloud type.
type. ItIt can
can be
be found
found that
that the
the cloud
cloud mainly
mainly occurs
occurs below
below 44 km
km with
with relative
relative thinner
cloud
thickness
in
the
forenoon,
and
the
cloud
structure
of
four-layer
cloud
is
similar
to
the
three-layer
thickness in the forenoon, and the cloud structure of four-layer cloud is similar to the three-layer
cloud’s,
cloud’s, while
while the
the cloud
cloud has
has been
been lifted
lifted distinctly
distinctly with
with all
all clouds
clouds getting
getting thicker
thicker in
in the
the afternoon,
afternoon, the
the
spatial
scale
has
been
extended
to
6
km
and
the
cloud
structure
of
four-layer
cloud
is
quite
different
spatial scale has been extended to 6 km and the cloud structure of four-layer cloud is quite different
from
the three-layer
three-layer cloud’s,
cloud’s,whose
whosetop
topcloud
cloudthickness
thicknesshas
hasdeveloped
developedfrom
from
0.19
km
0.47
km.
from the
0.19
km
to to
0.47
km.
TillTill
in
in
evening,
cloud
continuestotobecome
becomehigher
higherand
andthicker
thickerwith
withthe
the spatial
spatial scale
scale extending
extending to
thethe
evening,
thethe
cloud
continues
to
10 km,
km, and
and the
the top
top cloud
cloud thickness
thickness of
of the
the four-layer
four-layer cloud
cloud has
has developed
developed to
10
to 1.07
1.07 km,
km, which
which may
may be
be
the
result
that
the
cooling
effect
of
the
lower
cloud
is
relative
drastic
and
results
in
the
temperature
the result that the cooling effect of the lower cloud is relative drastic and results in the temperature
stratification
stratification instability,
instability, causing
causing more
more convective
convective condition
condition to
to contribute
contribute to
to the
the cloud
cloud development.
development.
Generally,
the
cloud
development
has
obvious
diurnal
variation
with
a
thicker,
Generally, the cloud development has obvious diurnal variation with a thicker, wider,
wider, and
and more
more
abundant
cloud
structure
process.
abundant cloud structure process.
Atmosphere 2017, 8, 9
Atmosphere 2017, 8, 9
14 of 17
14 of 14
Figure 14.
14. Averaged
Averaged spatial
spatial distribution
distribution of
of one-,
one-, two-,
two-, three-,
three-, and
and four-layer
four-layer clouds
clouds measured
measured by
by
Figure
WACALat:
at:(a)
(a)forenoon
forenoon(10:00–12:00);
(10:00–12:00);(b)
(b)afternoon
afternoon(16:00–19:00);
(16:00–19:00);and
and(c)
(c)night
night(20:00–22:00).
(20:00–22:00).
WACAL
Atmosphere 2017, 8, 9
15 of 17
5. Conclusions
As a part of the TIPEX III, the Ocean University of China operated the CL31 and the WACAL
from 7 July 2014 to 8 August 2014 to carry out the research on the microphysical characteristics
and the vertical structure of cloud in Tibetan Plateau. This paper summarizes the CL31 and
WACAL observations and comparisons of atmospheric cloud structure. The major findings are
summarized below.
(1)
(2)
(3)
(4)
The cloud occurrence at Nagqu is about 81% during the CL31 experimental period with obviously
diurnal variation. The cloud structure is relatively simple with a majority of single-layer cloud.
It can be found from the CBH comparison between CL31 and WACAL that the cases with obvious
deviation may result from the different cloud layer detection from these two devices. The CL31
sometimes overestimates the CBH compared with WACAL’s on the premise that the same cloud
layer is analyzed, which may be the result of the difference on the CBH definition and retrieval
methods, and this phenomenon has also been validated by synchronous radiosonde result.
The diurnal variations of CBH distribution proportion and CBH frequency from CL31 observation
have a similar variation trend with a distinct “U” shape, which analyze the cloud structure
variation at spatial and temporal aspects, respectively. In addition, in the time snippet comparison
between CL31 and WACAL, the results from these two devices have a good consistency in
distribution range, variation trend, the corresponding high value area and so forth.
The WACAL compares the averaged spatial distribution of one-, two-, three-, and four-layer
clouds from three different time periods, corresponding to different cloud development processes.
The averaged cloud thickness and vertical distribution range at night is about 2.5–2.9 times and
2.4–3.3 times as much as the clouds in the forenoon, and the occurrence frequency of multi-layer
cloud increases from 24% in the forenoon to 39% at night. Generally, the cloud development has
a distinct diurnal variation with a thicker, wider, and more abundant cloud structure process.
Acknowledgments: This work was partly supported by the National Natural Science Foundation of China (NSFC)
under grant 41375016, 41471309 and 91337103, by the China Special Fund for Meteorological Research in the
Public Interest under grant GYHY201406001. We appreciate the contribution made by Meteorological Bureau of
Tibet Autonomous Region, Nagqu Bureau of Meteorology. We also thank Guangyao Dai, Dongxiang Wang and
Qian Zhang for their help with the observations and data processing.
Author Contributions: All authors made great contributions to the presented work. Xiaoquan Song and
Xiaochun Zhai performed the algorithm and prepared the paper; Liping Liu and Songhua Wu designed the
experiments and provided the technical discussions; and Xiaochun Zhai performed the experiments and analyzed
the data.
Conflicts of Interest: The authors declare no conflict of interest.
References
1.
2.
3.
4.
5.
6.
Myhre, G.; Samset, B.H.; Schulz, M.; Balkanski, Y.; Bauer, S.; Berntsen, T.K.; Bian, H.; Bellouin, N.;
Chin, M.; Diehl, T.; et al. Radiative forcing of the direct aerosol effect from AeroCom Phase II simulations.
Atmos. Chem. Phys. 2013, 13, 1853–1877. [CrossRef]
Su, W.Y.; Loeb, N.G.; Schuster, G.L.; Chin, M.; Rose, F.G. Global all-sky shortwave direct radiative forcing
of anthropogenic aerosols from combined satellite observations and GOCART simulations. J. Geophys.
Res. Atmos. 2013, 118, 655–669. [CrossRef]
Zelinka, M.D.; Dennis, L.H. Why is longwave cloud feedback positive? J. Geophys. Res. Atmos. 2010,
115, D16117. [CrossRef]
Espinoza, H.R. Simple parameterizations of the radiative properties of cloud layers: A review. Atmos. Res.
1995, 35, 113–125. [CrossRef]
Cess, R.D.; Potter, G.L.; Blanchet, J.P.; Boer, G.J.; Ghan, S.J.; Kiehl, J.T.; Treut, H.L.; Li, Z.X.; Liang, X.Z.;
Mitchell, J.F.B.; et al. Interpretation of cloud-climate feedback as produced by 14 atmospheric general
circulation models. Science 1989, 245, 513–516. [CrossRef] [PubMed]
Stephens, G.L. Cloud feedbacks in the climate system: A critical review. J. Clim. 2005, 18, 237–273. [CrossRef]
Atmosphere 2017, 8, 9
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
16 of 17
Jian, M.Q.; Luo, H.B. Impact of the diurnal variation of the surface heating in the Tibetan Plateau on the
general circulation over the Asian monsoon region. J. Trop. Meteorol. 2002, 18, 269–275.
Ye, D.Z.; Gao, Y.X. Qinghai-Xizang Plateau Meteorology, 1st ed.; Science Press: Beijing, China, 2002; pp. 1–278.
Wang, H.; Luo, Y.L.; Zhang, R.R. Analyzing seasonal variation of clouds over the Asian monsoon regions
and the Tibetan Plateau region using CloudSat/CALIPSO. Chin. J. Atmos. Sci. 2011, 35, 1117–1131.
Zhao, Y.F.; Wang, D.H.; Yin, J.F. A study on cloud microphysical characteristics over the Tibetan Plateau
using CloudSat data. J. Trop. Meteorol. 2014, 30, 239–248.
Fu, Y.F.; Li, H.T.; Zi, Y. Case study of precipitation cloud structure viewed by TRMM satellite in a valley of
Tibetan Plateau. Plateau Meteorol. 2007, 26, 98–106.
Spinhirne, J.D. Micro pulse lidar. IEEE Trans. Geosci. Remote Sens. 1993, 31, 48–55. [CrossRef]
Liu, L.P.; Zheng, J.F.; Ruan, Z.; Cui, Z.H.; Hu, Z.Q.; Wu, S.H.; Dai, G.Y.; Wu, Y.H. The preliminary analyses of
the cloud properties over the Tibetan Plateau from the field experiments in clouds precipitation with the
various radar. Acta Meteorol. Sin. 2015, 73, 635–647.
He, Q.S.; Li, C.C.; Ma, J.Z.; Wang, H.Q.; Shi, G.M.; Liang, Z.R.; Luan, Q.; Geng, F.H.; Zhou, X.W. The properties
and formation of cirrus clouds over the Tibetan Plateau based on summertime lidar measurements.
J. Atmos. Sci. 2013, 70, 901–915. [CrossRef]
Liu, C. Measurements of the Aerosol over Nagqu of Tibet and Suburb of Beijing by Micro Pulse Lidar (MPL).
J. Photon. Study 2006, 35, 1435–1439.
Yanai, M.; Li, C.; Song, Z. Seasonal heating of the Tibetan Plateau and its effects on the evolution of the Asian
summer monsoon. J. Meteorol. Soc. Jpn. 1992, 70, 319–351.
Vaisala Ceilometer CL31. Available online: https://sensorexperts.com/node/221729 (accessed on
21 April 2016).
Wu, S.H.; Song, X.Q.; Liu, B.Y.; Dai, G.Y.; Liu, J.T.; Zhang, K.L.; Qin, S.G.; Hua, D.X.; Gao, F.; Liu, L.P.
Mobile multi-wavelength polarization Raman lidar for water vapor, cloud and aerosol measurement.
Opt. Express 2015, 23, 33870–33892. [CrossRef] [PubMed]
GTS1 Digital Radiosonde. Available online: http://www.docin.com/p-1209358545.html (accessed on
21 April 2016).
Winker, D.M.; Vaughan, M.A. Vertical distribution of clouds over Hampton, Virginia observed by lidar under
the ECLIPS and FIRE ETO programs. Atmos. Res. 1994, 34, 117–133. [CrossRef]
Pal, S.R.; Steinbrecht, W.; Carswell, A.I. Automated method for lidar determination of cloud-base height and
vertical extent. Appl. Opt. 1992, 31, 1488–1494. [CrossRef] [PubMed]
Wang, X.P.; Song, X.Q.; Yan, B.D.; Chen, C.; Liu, B.Y.; Wu, S.H. Research on the observation of cloud-base
height for the city of Zhuhai of China with lidar. J. Optoelectron. Laser 2013, 11, 2192–2197.
Wang, J.; Rossow, W.B. Determination of cloud vertical structure from upper-air observations.
J. Appl. Meteorol. 1995, 34, 2243–2258. [CrossRef]
Zhang, J.Q.; Chen, H.B.; Li, Z.Q.; Fan, X.H.; Peng, L.; Yu, Y.; Cribb, M. Analysis of cloud layer structure
in Shouxian, China using RS92 radiosonde aided by 95 GHz cloud radar. J. Geophys. Res. Atmos. 2010,
115, D00K30. [CrossRef]
Wang, J.; Rossow, W.B.; Uttal, T.; Rozendaal, M. Variability of cloud vertical structure during ASTEX observed
from a combination of rawinsonde, radar, ceilometer, and satellite. Mon. Weather Rev. 1999, 127, 2482–2502.
[CrossRef]
Goff, J.A.; Gratch, S. Low-pressure properties of water from −160 to 212 F. Trans. Am. Soc. Heat. Vent. Eng.
1946, 51, 125–164.
Alduchov, O.A.; Eskridge, R.E. Improved Magnus form approximation of saturation vapor pressure.
J. Appl. Meteorol. 1996, 35, 601–609. [CrossRef]
Eberhard, W.L. Cloud signals from lidar and rotating beam ceilometer compared with pilot ceiling. J. Atmos.
Ocean. Technol. 1986, 3, 499–512. [CrossRef]
Feister, U.; Möller, H.; Sattler, T.; Shields, J.; Görsdorf, U.; Güldner, J. Comparison of macroscopic cloud data
from ground-based measurements using VIS/NIR and IR instruments at Lindenberg, Germany. Atmos. Res.
2010, 96, 395–407. [CrossRef]
Liu, L.; Sun, X.J.; Liu, X.C.; Gao, T.C.; Zhao, S.J. Comparison of Cloud Base Height Derived from a
Ground-Based Infrared Cloud Measurement and Two Ceilometers. Adv. Meteorol. 2015, 2015, 853861.
[CrossRef]
Atmosphere 2017, 8, 9
31.
32.
17 of 17
Van Tricht, K.; Gorodetskaya, I.V.; Lhermitte, S.; Turner, D.D.; Schween, J.H.; Van Lipzig, N.P.M. An
improved algorithm for polar cloud-base detection by ceilometer over the ice sheets. Atmos. Meas. Tech.
2014, 7, 1153–1167. [CrossRef]
Wang, B.M.; Liu, X.N. The cloud diurnal variation in China. In Cloud Distribution of China, 1st ed.;
Chen, H., Ed.; China Meteorological Press: Beijing, China, 2009; Volume 8, pp. 82–106.
© 2017 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC-BY) license (http://creativecommons.org/licenses/by/4.0/).