CFMIP workshop @Hongo campus, Univ. Tokyo 25-28 September 2017,

Presentation at 9:30-9:45, on 26th. (Tuesday)

Cloud feedbacks from CloudSat/CALIPSO to EarthCARE

H. Okamoto, K. Sato and S. Katagiri (Research Institute for Applied Mechanics, Kyushu University)

KU-products with refinements

1. Cloud mask from CloudSat/CALIPSO (KU-mask) Improved from Hagihara et al., 2010JGR, 2014JGR, with ground-based multiple scattering polarization lidar (MFMSPL) (Okamoto et al. 2016 Opt. Express).

2. Cloud particle type from CloudSat/CALIPSO(KU-type) Improved from Yoshida et al., 2010JGR by MFMSPL. CloudSat- and CloudSat/CALIPSO synergy-type algorithms Kikuchi et al., 2017JGR (in press).

3. Ice and water cloud microphysics from CloudSat/CALIPSO (KU-micro) improved from Okamoto et al., 2010 JGR, Sato and Okamoto 2011 JGR. Sato et al., in prep.

2

(ΔH=1.1km, ΔH=240m)

2

　EarthCARE (Earth Clouds Aerosol Radiation Explorer)

JAXA-ESA joint mission

1. 94GHz Doppler cloud radar (CPR)

2. 355nm high spectral resolution lidar (ATLID) (ΔH=1km or 10km, ΔZ=100m)

3. Multi-spectral imager (MSI) 7channels (0.69, 0.865, 1.65, 2.21,8.8, 10.8, 12.0µm)

4. Broad band radiometer (BBR) 3 views

launch date :2019 altitude : 393.14km

JAXA product, ESA product

KU-products are extended and will be distributed as JAXA-EarthCARE-products (ΔH=1km or 10km, ΔZ=100m)

(Illingworth et al., 2015 BAMS)

3

composite diagram using the CALIPSO data (see Sect. 2.5

for the method).Before examining the cloud changes in 2 9 CO2 and

their differences among the models, we compare the mean

cloud fraction between CALIPSO and GCMs (shading inFig. 9). The satellite-based estimate of the cloud fraction

(Fig. 9a) reveals the following characteristics: a maximum

of more than 30% occurring at the highest value of LTS,and a gradual increase of the cloud layer altitude as LTS

decreases. These features of the mean low-cloud fractionmay also be seen when we make the longitude-height

section along the subtropical eastern oceans (Wang et al.

2004). All the GCMs not only fail to reproduce the clouddistribution derived from CALIPSO but also show different

types of bias. Namely, low clouds are overestimated for

low LTS in MIROC5, CCSM3, and GISS ER, whereasoverall they are underestimated in MIROC3, CCCma, and

GFDL CM2.0. The cloud layer is too thin in MRI GCM.

The causes of these biases would involve various factorsand are beyond the scope of this study, but we need to bear

them in mind when comparing the cloud change in the

2 9 CO2 runs.The divergence of the mean cloud distribution in GCMs

prevents us from detecting and understanding the

consistent change in the cloud fraction in the 2 9 CO2

experiments (contours in Fig. 9). Yet, we can identify someconsistency although it may not necessarily explain the

different magnitude and sign of the total low-cloud change.

For example, a relatively large change in the cloud fractionis found at small LTS in models that overestimate the mean

cloud there (e.g., MIROC5, CCSM3, and GISS ER). At

large LTS, many models show an increase and decrease ofclouds above and below the mean cloud layer, suggesting

an upward shift of the cloud layer. This accompanies anasymmetry in the cloud amount change, either a greater

increase (e.g., MIROC5, CCSM3, and GISS ER) or

decrease (e.g., MIROC3, INM, and MRI), probablyresulting in a non-zero change of the low-cloud amount.

Despite large differences in the mean cloud fraction and

its changes among GCMs, the change in the thermody-namic condition Ps has a common structure, which repre-

sents a shift of the PDF peak to larger values (bottom

panels in Fig. 9). While the degree of the PDF shiftdepends on the model (for example, it is large in CCSM3

but small in INM), this coincidence indicates that the

changing thermodynamic constraint as found in the twoMIROC models (Fig. 8b, d) is a robust part of the climate

change. If the cloud change at a given LTS (contours) does

Fig. 9 Regime composite ofthe cloud fraction in the lowertroposphere over the tropicaloceans as sorted by LTS(shading), together with its PDF(curves at the bottom of eachpanel): a CloudSAT/CALIPSOfrom June 2006 to May 2007,b MIROC3.2, c MIROC5,d–i CFMIP1 models, all fromcontrol runs. The vertical axis isthe normalized pressure. In b–i,contours indicate the differencebetween the control and either4 9 CO2 or 2 9 CO2 runs(intervals ±1, 2, 4, 6, 8, 10%),and blue (red) curves in thebottom panels are the PDF forthe control (4 9 CO2 or2 9 CO2) run. The greyshading in the PDF gives thedefinition of stable regime (seetext)

M. Watanabe et al.: Fast and slow timescales

123

20

15

10

5

0

Altit

ude

(km

)

-90 -60 -30 0 30 60 90Latitude (degrees)

10-22 4 6 8

10-12 4 6 8

100

Cloud Fraction

C4, Oct_06

1.KU-mask: CF sorted by LTS in tropics.

(Watanabe et al., 2011Clim Dyn)

There are large variations of vertical-LTS of low clouds among GCMs. They failed to reproduce patterns from CloudSat/CALIPSO.

KU-mask

(Hagihara et al., 2010JGR)

4

KU-type MIROC5 GCM

2. KU-type: ice-water partition

(Watanabe et al., 2010 J. Climate)

(Yoshida et al., 2010 JGR, Hirakata et al., 2014 JTECH)

δ-X diagram was used for ice/water partition in KU-type by 2 curves and 3 lines. X=log(βi/βi+1), proxy of extinction.

waterice ice water100

80

60

40

20

0

Dep

olar

izat

ion

ratio

[%]

2.01.51.00.50.0-0.5-1.0X

1 2 3 4 5

3D-ice

water

unknown2unknown1

2D-plate

Cloud schemes in MIROC5 GCM generally captures temperature-latitude pattern of ice-water ratio, except for high latitudes.

5

3. KU-micro: evaluation of vertical grid spacing of NICAM

KU-micro (Okamoto et al., 2010 JGR, Sato and Okamoto 2011 JGR)

ß, δ, Ze →　Reff, IWC, mixing ratio of 2D-plate

IWC in NICAM is underestimated in tropics (Seiki et al., GRL 2015).

Higher vertical resolution <400m is necessary to reproduce IWC in NICAM with ΔH=14 and 28km.

6

−80 −60 −40 −20 0 20 40 60 800

2

4

6

8

10

12

14

16

18a. GOCCP

Hei

gh

t (k

m)

−80 −60 −40 −20 0 20 40 60 800

2

4

6

8

10

12

14

16

18b. ST

−80 −60 −40 −20 0 20 40 60 800

2

4

6

8

10

12

14

16

18c. KU

−80 −60 −40 −20 0 20 40 60 800

2

4

6

8

10

12

14

16

18 d. GOCCP − ST

Latitude (°N)

−80 −60 −40 −20 0 20 40 60 800

2

4

6

8

10

12

14

16

18 e. GOCCP − KU

Cloud Fraction Bias (%)−20 −16 −12 −8 −4

Cloud Fraction (%)0 3 6 9 12 15 18 21 24 27 30

−80 −60 −40 −20 0 20 40 60 800

2

4

6

8

10

12

14

16

18a. GOCCP

Hei

gh

t (k

m)

−80 −60 −40 −20 0 20 40 60 800

2

4

6

8

10

12

14

16

18

b. ST

−80 −60 −40 −20 0 20 40 60 800

2

4

6

8

10

12

14

16

18

c. KU

−80 −60 −40 −20 0 20 40 60 800

2

4

6

8

10

12

14

16

18 d. GOCCP − ST

Latitude (°N)

−80 −60 −40 −20 0 20 40 60 800

2

4

6

8

10

12

14

16

18 e. GOCCP − KU

Cloud Fraction (%)0 3 6

Cloud Fraction Bias (%)−20−16−12 −8 −4 0 4 8 12 16 20

−80 −60 −40 −20 0 20 40 60 800

2

4

6

8

10

12

14

16

18a. GOCCP

Hei

gh

t (k

m)

−80 −60 −40 −20 0 20 40 60 800

2

4

6

8

10

12

14

16

18

b. ST

−80 −60 −40 −20 0 20 40 60 800

2

4

6

8

10

12

14

16

18

c. KU

−80 −60 −40 −20 0 20 40 60 800

2

4

6

8

10

12

14

16

18 d. GOCCP − ST

Latitude (°N)

−80 −60 −40 −20 0 20 40 60 800

2

4

6

8

10

12

14

16

18 e. GOCCP − KU

Cloud Fraction (%)0 3 6

Cloud Fraction Bias (%)−20−16−12 −8 −4 0 4 8 12 16 20

−80 −60 −40 −20 0 20 40 60 800

2

4

6

8

10

12

a. GOCCP

Hei

gh

t (k

m)

−80 −60 −40 −20 0 20 40 60 800

2

4

6

8

10

12

b. ST

−80 −60 −40 −20 0 20 40 60 800

2

4

6

8

10

12

c. KU

−80 −60 −40 −20 0 20 40 60 800

2

4

6

8

10

12

d. GOCCP − ST

Latitude (°N)

−80 −60 −40 −20 0 20 40 60 800

2

4

6

8

10

12

e. GOCCP − KU

Cloud Fraction Bias (%)−20 −16 −12 −8 −4

Cloud Fraction (%)0 3 6 9 12 15 18 21 24 27 30

−80 −60 −40 −20 0 20 40 60 800

2

4

6

8

10

12

a. GOCCP

Hei

gh

t (k

m)

−80 −60 −40 −20 0 20 40 60 800

2

4

6

8

10

12

b. ST

−80 −60 −40 −20 0 20 40 60 800

2

4

6

8

10

12

c. KU

−80 −60 −40 −20 0 20 40 60 800

2

4

6

8

10

12

d. GOCCP − ST

Latitude (°N)

−80 −60 −40 −20 0 20 40 60 800

2

4

6

8

10

12

e. GOCCP − KU

Cloud Fraction (%)0 3 6

Cloud Fraction Bias (%)−20−16−12 −8 −4 0 4 8 12 16 20

4. Comparison of ice and water cloud fraction from CALIPSO

ice cloud fraction

water cloud fraction

cloud

Cesana et al., 2016 JGR due to different treatment of noise, clouds/aerosol partitions, resolutions, fully attenuated pixels, multiple scattering

KU

NASA KUFrenchFrench NASA

The ST and KU liquid fractions remain consistent before and after the tilt change. Another possibility concernsthe different treatments of aerosols in the algorithms, which might generate some discrepancies over thedust-polluted regions such as the Saharan desert or the China basin.

5.3. Cloud Phase Transition Against Temperature

Previous studies showed that the cloud phase transition between liquid and ice clouds (i.e., the ratio of iceclouds to all clouds, referred as to Phase Ratio, PR) could be represented in a simple way using the tempera-ture [e.g., Bower et al., 1996; Korolev et al., 2003; Hu et al., 2010; Cesana and Chepfer, 2013]. This method lets useasily compare different data sets on the same plot using the same thermodynamical variable—the tempera-ture—as reference. Here we took advantage of the in situ measurements to validate the temperature used inthe CALIPSO products (GEOS5-DAS in GOCCP and ST, and European Centre for Medium-Range WeatherForecasts-auxiliary product (ECMWF-AUX) provided by the CloudSat team in KU). We averaged the aircrafttemperature onto the CALIPSO horizontal sampling grid and then computed the mean absolute error(MAE) and the maximum error (ME) for all/cloud/clear sky pixels (according to the in situ measurements).The results are presented in Table S1. In the Arctic as well at midlatitudes, the MAE is lower than 2°C andthe ME no more than 4°C.

Figure 6 shows the relationship between the PR and the temperature, for the CALIPSO products (GOCCP inred, ST in magenta, and KU in green), the aircraft in situ measurements (black + circle line), and the averagebetween five passive sensor satellites of the Global Energy and Water Cycle Experiment-Cloud Assessment(GEWEX-CA) [Stubenrauch et al., 2013; blue line and blue shade]. To make this figure, we created Phase-Temperature composites by breaking the PR into 3° temperature bins. For the CALIPSO products, weused the monthly mean files over 1 year (thick dashed lines). As GOCCP and KU relations are sensitive tothe change of CALIOP nadir-pointing angle (0.3° off nadir before 2008, to 3° off nadir after), we used the year2008 to build these relations (ST ice and liquid cloud amount remain consistent before and after the tilt, notshown). The change of nadir angle decreased specular returns of horizontally oriented ice crystals. Thisresulted in less false cloud detection and less false liquid cloud determination since ice crystal plates producethe same signature as liquid droplets [e.g., Sassen et al., 2012]. In GOCCP, we assumed undefined-phase

Figure 5. same as Figure 4 for liquid clouds.

Journal of Geophysical Research: Atmospheres 10.1002/2015JD024334

CESANA ET AL. CALIPSO CLOUD PHASE VALIDATION 5802

Large differences in three CALIPSO global products.7

4-2. Depolarization ratio due to multiple-scattering The depolarization ratio for the four tilt angles was estimated using eight channels (Fig.

4). Depolarization ratio ( )δ θ for four different angles θ (mrad) was estimated by

(0) ( .2) / ( .1),(10) ( .4) / ( .3),(20) ( .6) / ( .5), (30) ( .8) / ( .7),

att att

att att

att att

att att

Ch ChCh ChCh Ch andCh Ch

δ β βδ β βδ β βδ β β

⊥

⊥

⊥

⊥

====

�

�

�

�

(1).

3500

300025002000

1500

Hei

ght [

m]

80x103706050403020

Time in 5 March 2015

CH2/CH1 (0mrad)

1.00.80.60.40.20.0Depolarization ratio

35003000250020001500

Hei

ght [

m]

80x103706050403020

Time in March 05 2015

CH4/CH3 (10mrad)

3500

3000

2500

2000

1500

Hei

ght [

m]

80x103706050403020

Time in March 5, 2016

CH6/CH5 (20mrad)

3500

3000

2500

2000

1500H

eigh

t [m

]

80x103706050403020

Time in March 5, 2016

CH8/CH7 (30mrad)

(a)

(b)

(c)

(d)

3500

3000

2500

2000

1500

Hei

ght [m

]

8x105765432

Time in March 05 2015

NIES lidar

(e)

Fig. 4. Same as Fig. 2 but for depolarization ratio. (a) For on-beam channel, (b) for 10 mrad, (c) for 20 mrad, (d) for 30 mrad, and (e) for NIES lidar.

Vol. 24, No. 26 | 26 Dec 2016 | OPTICS EXPRESS 30060

Ľ�

2.1!OûĞýƵOĝ����ƮƆƗƶűģÌ!MFMSPL&(Mul*&+&Field&of&view&+&Mul*ple&–&Sca7ering&–&Polariza*on&–&Lidar)&!

Nd:YAG&laser� Beam&Expander�

Ch.2�Ch.4�Ch.6�Ch.8�

Ch.1�Ch.3�Ch.5�Ch.7�

Amplifer�

PC�

AD!Comveter�

Parallel!channel!

Perpendicular!channel!

0 10 20 30 40 [!mrad!]�ƸTiltƹ�parallel&channel&

Perpendicular&channel&

ŒƏƑƚƪ¡Ëœ�

ŒMFMSPLœ�

OnYbeam�offYbeam�

OnYbeam!!!:!Ch.1,!2!OffYbeam1!:!Ch.3,!4!OffYbeam2!:!Ch.5,!6!OffYbeam3!:!Ch.7,!8�

ĠÏ%þïĸĸƽ!!6m!�Ĥ%þïĸĸƽ!!10Ø!ü·CƽŪŜųƸÇJÕƹ!Ėèü·ƽ2014/6/7!~�

4/14�

5. Multi-Field of View MultipleScattering Polarization Lidar

8 telescopes are used and total FOV ~

70mrad.

This is the first time that depolarization ratio for optically thick clouds was observed by ground-based lidar.

(Okamoto et al., 2016, Opt. Express.)

Fig. 2. MFMSPL installed at the National Institute of Environmental Studies (NIES), Tsukuba, Japan.

Table 1. MFMSPL system specifications.

Transmitter Laser Nd:YAG, Q-switched, linearly polarized (Quantel, Brilliant EaZy) Wavelength 532 nm Pulse energy 165 mJ/pulse Repetition rate 10 Hz Receiver Lens Diameter = 5 cm, focal length = 40 cm (CVI Laser Optics) FOV 10 mrad Detectors PMTs (Licel, PM-HV-20)

Bandpass filter Interference filter with 1 nm FWHM (Andover Corporation)

ND filters Transmittance: 1% for Ch.1 and 7% for Ch.2 (Sigma Koki Group) (ND filters were used only for channels 1 and 2)

Data acquisition • A/D converter for all of the channels, 25 MHz, 16-bit (Turtle Industry Co., TUSB-0216ADM)

3. Calibration procedures There are two steps used for calibration of the MFMSPL to obtain attenuated backscattering coefficient (ßatt) for each channel. The first is absolute calibration, which is performed by comparing signals observed by Mie-type NIES vertical-pointing lidar and those observed from channel 1 (vertical-pointing). Calibration of the NIES lidar was performed in advance, according to the procedure described in [16]. The NIES lidar was necessary for calibration because the MFMSPL cannot detect signals from altitudes higher than 8 km during the day, due to its limited sensitivity, and self-calibration was not possible.

Vol. 24, No. 26 | 26 Dec 2016 | OPTICS EXPRESS 30057 Simulation of space-borne lidar signals becomes possible. Evaluation of CALIPSO/ATLID algorithm can be done.

vertical

10mrad

20mrad

30mrad

40mrad

From Cesana et al., 2016 JGR

8

5-1. Ice and water cloud fraction by refined KU-type

20

15

10

5

-80 -60 -40 -20 0 20 40 60 80

0.300.250.200.150.100.05

14

12

10

8

6

4

2

0

-80 -60 -40 -20 0 20 40 60 80

0.200.150.100.05

14

12

10

8

6

4

2

0

Height [km]

-80 -60 -40 -20 0 20 40 60 80Latitude [degN]

Liquid Cloud Fration - Former Version -

0.20

0.15

0.10

0.05

0.00

20

15

10

5

0

Height [km]

-80 -60 -40 -20 0 20 40 60 80Latitude [degN]

0.30

0.25

0.20

0.15

0.10

0.05

0.00

Ice Cloud Fration - Former Version - iceice

WaterWater

Former

Former

Refined

Refined

Low level water cloud fraction increases in refined scheme.8

KU (new)

KU (new)

Ice cloud fraction

Water cloud fraction

6. Differences among three global products decrease.

KU(old)<GOCCP<ST -> GOCCP<KU(new)<ST

Might be interesting to revisit LTS-CF relation.

9

15

10

5

Height [km]

-80 -40 0 40 80Latitude

5 6 70.1

2 3

Particle type from CloudSat-only and CloudSat/CALIPSO

Sato et al., 2017 ( in prep.)

LWC from CALIPSO

Confidential manuscript submitted to Journal of Geophysical Research: Atmospheres

12

4.1 CPR Stand-alone Hydrometeor Particle Type 370

The CPR stand-alone algorithm consists of three main steps: (1) initial classification of 371 hydrometeor particle type based on radar reflectivity and temperature, (2) correction for 372 classification of cloud–precipitation, and (3) spatial continuity test. 373

4.1.1 Initial Classification of Hydrometeor Particle Type 374

The hydrometeor type is initially determined based on the relationship of the dominant 375 hydrometeor type with radar reflectivity and temperature. For this purpose, we constructed a type 376 classification diagram that describes the initial guess of hydrometeor type using the relationship 377 (Fig. 4). This was derived from Fig. 2 by selecting the hydrometeor type that gave the highest 378 fractional occurrence exceeding the minimum threshold of 3 u 10-6 at a given radar reflectivity–379 temperature bin. The category of water was further separated according to temperature T into the 380 warm water ( CT qt 0 ) and the supercooled water ( CT q� 0 ). Furthermore, when the fractional 381

occurrence was below the threshold, we assigned rain to CT qt 0 , and snow to CT q� 0 based on 382 the fact that dominance of the precipitation coincides with the low occurrence of CPR/CALIOP 383 detection as argued in Section 3.2. In the next step described in Section 4.1.2, the temperature 384 threshold of 0 qC for rain-snow separation is modified to 2 qC [Liu 2008], in the convective 385 cloud types where falling snows typically reach below 0qC level. The convective cloud types 386

were defined by those with the temperature at the lowest hydrometeor top layer ( LHTTemp ) 387

below 0 qC. 388

389

390

Figure 4. Radar reflectivity–temperature diagram for the initial hydrometeor type classification 391 in the radar algorithm. In the precipitation correction, we modified the 0 qC threshold for rain 392 and snow separation to 2 qC for convective clouds where the temperature at the lowest 393 hydrometeor top was below 0 qC. 394

395

Kikuchi et al., 2017 JGR

7. New KU-products (will be distributed soon)Water microphysics from CALIPSO (based on fast semi-analytical methods

Ice microphysics from CS/CA refined.

Physical optics for CALIPSO, better ice scattering models (with Borovoi’s and Ishimoto’s group)

10

10-4

10-3

10-2

10-1

100

ß [/m

/sr]

for I

WC

=1g/

m3

4 6 810

2 4 6 8100

2 4 6 81000

Effective radius [µm]

2D-Plates (1.0) for 0.3 deg. 2D-Plates (1.0) for 3 deg. 2D-Plates (2.0) for 0.3 deg.2D-Plates (2.0) for 3 deg. 2D-Plates (0.5) for 3 deg.

70

60

50

40

30

20

10

0

Lida

r rat

io

6050403020100

Depolarization ratio [%]

Lser tilt angle=3 deg. 2D-plate with beta=0.5 2D-plate with beta=1 2D-plate with beta=2 2D-column with beta=0.5 2D-column with beta=1.0 2D-column with beta=2.0 Droxtals

Ice (and aerosol) particle type from HSRL function of ATLID

Simultaneous retrievals of vertical air motion, particle fall speed and particle mass flux from Doppler-CPR-ATLID.

-10

-8

-6

-4

-2

Vd

[m/s

]

30025020015010050Effective radius [µm]

Droxtal 94GHz radar

Droxtals

Columns

Oriented plates

8. New from EarthCARE

to evaluate cumulus convective parameterization and cloud parameterization

1.2

1.0

0.8

0.6

0.4

0.2

0.0

Signal [A.U.]

-2.5 0.0 2.5Wavelength shift [pm]

Particle

Molecule

12x103

8

4Hei

ght [

m]

414039383736Latitude [degree]

-6 -5 -4 -3 -2 -1 0Doppler velocity [m/s]

Lida ratio=σext/ß

11

Summary

1. Refined KU-mask shows increase of low-level cloud fraction. 2. Refined KU-type shows increase of water cloud fraction. 3. LUTs for ice particles are updated in order to bridge the gap in retrieved microphysics among nadir- and off-nadir CALIPSO periods and also EarthCARE.

4. Fast semi-analytical approach for lidar multiple scattering is developed. Water microphysics is retrieved by CALIPSO(Sato’s presentation).

5. Ground-base multiple scattering polarization lidar is developed and extension is planned to connect CALIPSO and ATLID on EarthCARE information.

6. EarthCARE is expected to provide vertical air motion, better particle typing, and cloud mass flux by Doppler function of CPR and high spectral resolution function of ATLID.

12

3200

3000

2800

2600

2400

2200

2000H

eigh

t [m

]

68x103666462605856

Time

10-8 10

-6 10-4

ßtotal [1/m/sr] for CH5 and CH6

3200

3000

2800

2600

2400

2200

2000

Hei

ght [

m]

68x103666462605856

Time

10-8 10

-6 10-4

ßtotal [1/m/sr] for CH5 and CH6

5-1. Refinement of KU-mask by MFMSPL

Former KU-maskRefined KU-mask

Current KU-mask underestimates cloud fraction in large attenuation in liar signals. Refined scheme overcomes the issue mainly by introducing different discrimination schemes at/below cloud top (above cloud bottom) for space-borne (ground-based) lidar.

Okamoto et al., 2017a

Doppler accuracy of EarthCARE CPR

16km mode (PRF=7200Hz)a:

10km horizontal integration: 1.1m/s for -19dB (worst case). 0.6m/s for -19dB (nominal case).

1km horizontal integration: 1.3m/s for -14dBZ (worst case).

0.8m/s for -14dBZ (nominal case). Theoretical limit =0.2m/s

trying from 0.8 to 0.2m/s…

ー20kmー

ー40kmー

ー20kmー

New Ver.

Refined scheme shows improvements in detection of clouds.

former scheme

5-2 Development of CALIPSO-cloud mask; example

Blue : fully attenuated pixel.

Low level cloud detections have been improved in largely attenuated regions by the refined scheme. Katagiri et al., 2017

16

Refined scheme

5-3. Refinement KU-type by MFMSPL

Current algorithm misclassified water as ice in regions where lidar signals are heavily attenuated. It leaded to underestimation of water cloud fraction. Refined algorithm overcomes the issue.

3200

3000

2800

2600

2400

2200

2000

Hei

ght [

m]

68x103666462605856

Time

210Cloud particle type for old type

3200

3000

2800

2600

2400

2200

2000H

eigh

t [m

]

68x103666462605856

Time

210Cloud particle type for old type

1:water, 2: ice

Former Refined

waterwaterice

Okamoto et al., 2017a

70

60

50

40

30

20

10

0

Lida

r rat

io

6050403020100

Depolarization ratio [%]

Lser tilt angle=3 deg. 2D-plate with beta=0.5 2D-plate with beta=1 2D-plate with beta=2 2D-column with beta=0.5 2D-column with beta=1.0 2D-column with beta=2.0 Droxtals

Droxtal

Column

Oriented plates

高スペクトル分解ライダと雲粒子タイプ ８

Confidential manuscript submitted to Journal of Geophysical Research: Atmospheres

12

4.1 CPR Stand-alone Hydrometeor Particle Type 370

The CPR stand-alone algorithm consists of three main steps: (1) initial classification of 371 hydrometeor particle type based on radar reflectivity and temperature, (2) correction for 372 classification of cloud–precipitation, and (3) spatial continuity test. 373

4.1.1 Initial Classification of Hydrometeor Particle Type 374

The hydrometeor type is initially determined based on the relationship of the dominant 375 hydrometeor type with radar reflectivity and temperature. For this purpose, we constructed a type 376 classification diagram that describes the initial guess of hydrometeor type using the relationship 377 (Fig. 4). This was derived from Fig. 2 by selecting the hydrometeor type that gave the highest 378 fractional occurrence exceeding the minimum threshold of 3 u 10-6 at a given radar reflectivity–379 temperature bin. The category of water was further separated according to temperature T into the 380 warm water ( CT qt 0 ) and the supercooled water ( CT q� 0 ). Furthermore, when the fractional 381

occurrence was below the threshold, we assigned rain to CT qt 0 , and snow to CT q� 0 based on 382 the fact that dominance of the precipitation coincides with the low occurrence of CPR/CALIOP 383 detection as argued in Section 3.2. In the next step described in Section 4.1.2, the temperature 384 threshold of 0 qC for rain-snow separation is modified to 2 qC [Liu 2008], in the convective 385 cloud types where falling snows typically reach below 0qC level. The convective cloud types 386

were defined by those with the temperature at the lowest hydrometeor top layer ( LHTTemp ) 387

below 0 qC. 388

389

390

Figure 4. Radar reflectivity–temperature diagram for the initial hydrometeor type classification 391 in the radar algorithm. In the precipitation correction, we modified the 0 qC threshold for rain 392 and snow separation to 2 qC for convective clouds where the temperature at the lowest 393 hydrometeor top was below 0 qC. 394

395

Kikuchi et al., 2017 In press, JGR

Ze-T diagramLidar ratio-dep. diagram

Based on Physical Optics