Robert Wood, Louise Leahy, Bob Charlson, Peter BlosseyUniversity of Washington

Chris Hostetler, Ray Rogers, Mark Vaughan, Dave Winker NASA Langley Research Center

The Nature and Extent of Optically-Thin Low Clouds

Photo, Bjorn Stevens, RICO

Why study optically-thin low clouds?

Marine low clouds according to MODIS

10 year cdf of cloud optical thickness

Clouds of all sizes contribute significantly to cloud cover

Each decade contributes equally

See also Wood and Field (2011)

…..implies that very small clouds contribute importantly to global cloud cover

…..what about albedo?

Cloud

Clear Cloudy Cloudy Cloudy Clear

CALIOP Data• Use CALIOP Vertical Feature Mask (VFM) at highest

resolution (90 m FOV, 330 m spacing)

• Only low clouds (ztop<3 km) included

• An optically-thin cloud is defined as a cloud detected at full-resolution that does not fully attenuate the lidar signal such that the surface is also detected in the same profile

• For a cloud uniform across the FOV, this would corresponding to cloud optical depth (cld) less than 3

• Clouds broken at the FOV scale are also classified as optically thin

Integrated attenuated backscatter

𝛾′532𝑛𝑚❑ =∫

𝑧 1

𝑧 2

𝛽′ (𝑧 )𝑑𝑧

Integrated Attenuated Backscatter (IABS) (sr-1)

Optically-thin low cloud ubiquitous

Low Cloud Coverfcld 0.50 (0.25)

Optically-thin Low-Cloud Coverfthin 0.23 (0.09)

Optically-thin fraction of Low Cloudfthin,cld 0.45 (0.28)

Optically thin fraction decreases with cloud cover

• Clouds with >90% optically thin profiles are termed “majority optically thin”

• Clouds with >90% optically thick profiles are termed “majority optically thick”

• Most clouds smaller than a few km consist primarily of optically thin shots

• Clouds > 100 km in length are mainly optically thick

Optically thin low clouds are small

Cloud Length at Median Cloud Cover

(L50)

• Most optically thin low clouds in any given region are significantly smaller than optically thick clouds in that region

All low clouds

Majority optically thin low clouds

Majority optically thick low clouds

L50 [km]

Comparison with Higher Resolution Lidar

• NASA’s airborne High Spectral Resolution Lidar (HSRL).

• 4 spatially and temporally matched HSRL underflights of CALIOP over the tropical and subtropical western Atlantic.

• Temporal coincidence within ± 15 minutes

• HSRL footprint 8 x 60 m, contiguous FOVs

HSRLCALIOP

Optically-Thin Fraction as a Function of Scale

~𝑓 h𝑡 𝑖𝑛 ,𝑐𝑙𝑑=

1

∫𝐿𝑚𝑖𝑛

𝐿𝑚𝑎𝑥

𝐿𝑛(𝐿)𝑑𝐿∫𝐿𝑚𝑖𝑛

𝐿𝑚𝑎𝑥

𝑓 h𝑡 𝑖𝑛 ,𝑐𝑙𝑑𝐿 𝐿𝑛(𝐿)𝑑𝐿

Cloud length distribution alone explains three-quarters of the variance in fthin,cld:

r2 = 0.73 domain-wider2 = 0.77 over the Tropics

Implies that knowledge of how the marine cloud length distribution varies, is sufficient to predict geographical variation in fthin,cld across most of the ocean!

Cloud top heights

Large Eddy Model

CGILS: CFMIP/GCSS Intercomparison

S6: Trade cumulus regime 6 day runs

(∆x=100m, ∆z=40m)

Contribution to albedoCu

mul

ative

con

trib

ution

to c

loud

alb

edo

Liquid water path [g m-2]

=3

Land and Ocean contrasts

OceanLand

Ocean shows tight coupling between low cloud cover and optically thin fraction of low clouds, whereas land does not

Nighttime L50

The Scale of Optically-Thin CloudsDaytime L50

Conclusions• Over the non-polar oceans, optically-thin clouds comprise 45% of

marine low clouds with cloud top height below 3 km

• The optically-thin fraction of marine low cloud varies inversely with marine low-cloud cover, and reaches a maximum (> 0.80) in trade wind regions.

• Optically-thin marine low clouds are predominantly small clouds, with many smaller than CALIOP field of view.

• The cloud length distribution of all low clouds explains 3/4 of the geographical variance in the optically-thin fraction of marine low clouds.

• The largest optically-thin clouds are found over land regions, despite lower cloud cover over land.

Leahy, L. V., R. Wood, R. J. Charlson, C. A. Hostetler, R. R. Rogers, M. A. Vaughan, and D. M. Winker. On the nature and extent of optically-thin low clouds. L. V. Submitted to J. Geophys. Res.

Canonical modes of mesoscale

variabilityin subtropical and tropical marine low

clouds

No mesoscale cellular convection

Closed mesoscale cellular convection

Open mesoscale cellular convection

Cellular but disorganized

Wood and Hartmann (2006), J. Climate

Predominant mesoscale low cloud modes

Mean precipitation

rates

• CloudSat light rain (Lebsock and L’Ecuyer 2011)

• Cloud base precipitation rate

• Much of the warm rain from shallow cloud systems comes from open cell MCC over the global oceans

Two questions:1) How often is the lidar FOV partially filled with optically-thick cloud?

Model Optically-thin Cloud threshold:

2) How does sensor resolution affect fcld?

Bounded Cascade Model

𝜏>3𝜏∗=3

Surface

60 m Clear

𝑛∗=𝑛0𝑒− 2𝜂𝜏∗ 𝑛=𝑛0𝑒

−2𝜂𝜏

𝑛𝑛∗

>1

CALIOP Detector

FOV 1 FOV 2

(no surface signal)

z

Bounded Cascade Model

𝑤𝑖=(1− 2𝑝)2𝐻 (𝑖− 1)

Model Input Parametersp: intermittency (0 < p < 0.5)H: scaling (0 < H < : mean clear and cloud optical depth

(Wood and Field, 2011)

Bounded Cascade Model

Lognormal Optical Depth Distribution

Model Input Variables to simulate Trade Cumulus Region cloud observations

H was chosen to reproduce observed CALIOP cloud size distribution exponent ( there is an inverse relationship between H and Wood and Field, 2011).

p was set to simulate observed cloud cover. was adjusted to match CERES albedo for this region (0.16). Model derived albedo is

estimated from model optical depth values.

Instrument fcld fthin,cldObserved:    

CALIOP 0.26(0.11) 0.84(0.13)Model      

CALIOP 0.27 0.93HSRL 0.26 0.93HSRL Full Res 0.22 0.79

ClearCloud

Optically-thin Cloud

Potential Overestimate in fthin,cld and fcld

Potential overestimate in

fthin,cld is ~6%

Potential overestimate in fcld

is ~20%

Nighttime Land Low-Cloud Cover

Low Cloud Coverfcld 0.21 (0.21)

0.50 (0.25)

Low Cloud Coverfthin 0.10 (0.06)

0.23 (0.09)

Optically thin Fraction of Low Cloudfthin,cld 0.55 (0.28)

0.45 (0.28)

Land and Ocean Low-Cloud Cover

Ocean Nighttime

Land Nighttime

Nighttime L50

Optically-Thin Clouds as a Function of ScaleDaytime L50

Conclusions: Land• Optically-thin clouds comprise 0.10-0.20 of low clouds over land, with a mean value

0.15.

• Optically-thin fraction of low cloud has maxima over all southern hemisphere continents, and arid desert regions. In the northern hemisphere, values less than 0.4 are not observed.

• Horizontally extensive optically-thin clouds are observed over land at night, in mid-latitude regions.

• Variation of optically-thin cloud as a function of cloud length has the same overall relationship for land and ocean data, however, clouds over land at night are optically thinner. Daytime optically-thin clouds closely match the marine daytime and nighttime data.

Summary• This analysis finds a prevalence of optically-thin clouds at the scale of an individual

CALIOP FOV (90 m), over land and ocean.

• Knowing how the marine cloud length distribution varies, is sufficient to accurately predict geographical variation in optically-thin fraction of cloud across most of the ocean. This is not the case over land.

• A simple fractal model simulation of a marine trade cumulus region indicates that the majority of FOV containing optically-thin clouds are partially filled. However, clouds filling these FOV are mostly optically-thin.

• Optically-thin fraction of low cloud may be overestimated by ~6% due to partial filling of the lidar footprint with optically-thick cloud.

• Low-cloud cover may be overestimated by as much as 20% by CALIOP when sampling in broken cloud fields.

• The largest optically-thin clouds are found over land regions, despite lower cloud cover over land.

Acknowledgements

Committee: Rob Wood (Chair), Tom Ackerman, Qiang Fu, Dennis Hartmann, LuAnne Thompson (GSR), Steve Riser (Acting GSR)

UW: Bob Charlson, Sarah DohertyHarry Edmon, David Warren, and Marc Michelson

CALIPSO: Mark Vaughan

EXTRA SLIDES

Next Steps

• High resolution LES model simulations of trade wind regions, and Sc regions.

• Radiative transfer modeling of optically-thin cloud shortwave and longwave effects of these clouds.

• Examine diurnal cycle

Local Solar Time vs Latitude for the CALIPSO Orbit

-90

-75

-60

-45

-30

-15

0

15

30

45

60

75

90

0 2 4 6 8 10 12 14 16 18 20 22 24

98o sun-synch orbit

Local Solar Time (hours)

La

titu

de

(d

eg

s)

Daytime

NIghttime

35

Daytime High-cloud-screened number of data points, 2 yrs

36

Bounded Cascade Model Cloud Length

Extra Slide on CALIOP

z(r, kp)=zsat(kp) - r cos [θ(kp)],

Multiple Scattering: 3-D RT calculationsCalculateda partitioning of 180º backscatter of laser radiation for CALIPSO lidar

Feature type optical depthb Single scatter, within FOVc

Multiple scatter, within FOVc

Multiple scatter, outside FOVc

water cloud 30 10% 29% 61% water cloud 9 17% 28% 55% water cloud 3 38% 34% 28%

cirrus 1 50% 18% 32% sulfate aerosol 1 33% 4% 63%

aRadiative transfer calculations courtesy of David M. Winker. bPhysical depths are 0.3 km for water cloud, 1 km for cirrus, 2 km for sulfate aerosol. cFOV: lidar field-of-view for CALIPSO detector telescope (130 rad full angle; 705 km altitude)

• 28-63% of 180-backscatter is not sensed by spaceborne lidar• depends on layer type and instrument geometry• calculations assume 3-D homogeneity

Data Description• Two years of CALIOP vertically-resolved Vertical Feature Mask (VFM) cloud data, at full

resolution.

• Cloud detection at 1064 nm, threshold in terms of optical depth:Daytime Nighttime Layer Depth (m)0.05 0.03 > 180 0.17 0.10 < 180

• Data are screened for high clouds (cloud top height > 3 km), and multi-layer low clouds, and binned into 5° x 5° grid boxes.

We don’t know a priori whether detecting the surface in a cloud containing profile is indicative of an optically-thin cloud, or just broken optically-thick clouds. ‘

Classifying a partially filled FOV (pFOV) as optically thin when the cloud intercepted is optically-thick cloud will lead to an overestimate of optically-thin cloud occurrence.

Marine Cloud RegimesOptically-Thin Fraction of Low Cloud

Stratocumulus (Sc)

Sc-Cu Transition

Cumulus (Cu)

Cloud Type Latitude Longitude

Sc 20°N - 30°N 120°W - 130°W

Sc-Cu 10°N - 20°N 130°W - 140°W

Cu 10°S - 20°S 150°W - 160°W

Marine Cloud Length Distribution

Cloud Regimes: Low Cloud Length