study of the relationship between ice phase and cloud-top … · 2016-07-29 · study of the...

1
Study of the relationship between ice phase and cloud-top temperature with MODIS data Catherine Naud and Anthony Del Genio, GISS, NY Abstract: Cloud feedback in GCM climate change simulations is sensitive to predictions of cloud phase because of the different microphysical and radiative properties of liquid and ice. Usually the liquid-ice transition is parameterized as a simple function of temperature, but ice occurrence in clouds is known to depend on the underlying dynamics of the cloud system. We are using MODIS data to gain insights into this behaviour that might serve as a guide to improved parameterizations. Cloud top temperatures were obtained from 5-min level 2 MODIS TERRA and AQUA day-time granules, along with SWIR phase for all available winters (Dec-Jan-Feb) from the start of the missions until February 2004. Cloudy pixels in each granules were selected over the northern parts of the Atlantic and Pacific oceans in the area covered by the storm tracks and aggregated in 2x2.5° grid boxes. For each grid box, all pixels identified as ice only and ice or water are selected and sorted into temperature bins defined every 2K from 200 to 300K. Histograms can then be obtained that give the fraction of ice cloud pixels over the total number of ice and water cloud pixels in a given temperature range. The grid boxes are further aggregated along the best estimate of the storm track trajectory obtained from a contemporary NCEP reanalysis storm track atlas (Chandler and Jonas, 1999) and divided into a west, central and east region. The storm trajectories are chosen so that the storm frequency of occurrence does not vary much over the trajectory. However, we find that both mean intensity (central pressure) and severity (peak intensity) increase along the track. Atlantic ocean NCEP storm frequency (%) for all winters from December 2000 to February 2004. NCEP storm mean intensity (mb) for all winters from December 2000 to February 2004. NCEP storm severity (mb) for all winters from December 2000 to February 2004. MODIS ice cloud fraction versus cloud- top temperature for all months and both platforms accumulated in the western (solid), central (dashed) and eastern (dot- dash) portions of the storm track.There are at least 10,000 pixels in each 2K temperature range. NCEP storm frequency (%) for all winters from December 2000 to February 2004. NCEP storm mean intensity (mb) for all winters from December 2000 to February 2004. NCEP storm severity (mb) for all winters from December 2000 to February 2004. For 4 different temperature ranges, ice cloud fraction versus storm frequency (top), mean intensity (central) and severity (bottom). The ranges are: 230- 240K (column 1), 240-250K (column 2), 250-260K (column 3) and 260- 270K (column 4). Variations along the estimated storm mean trajectory of storm frequency (top), mean intensity (centre) and severity (bottom) as the storms move eastward. Pacific Ocean For 4 different temperature ranges, ice cloud fraction versus storm frequency (top), mean intensity (central) and severity (bottom). The ranges are: 230- 240K (column 1), 240-250K (column 2), 250-260K (column 3) and 260- 270K (column 4). MODIS ice cloud fraction versus cloud- top temperature for all months and both platforms accumulated in the western (solid), central (dashed) and eastern (dot- dash) portions of the storm track.There are at least 10,000 pixels in each 2K temperature range. Variations along the estimated storm mean trajectory of storm frequency (top), mean intensity (centre) and severity (bottom) as the storms move eastward. Preliminary Conclusion: Over the Atlantic storm track region, we found that as the clouds move eastward, the ice cloud fraction at a given temperature tends to increase, implying that heterogeneous freezing becomes more important as the storms travel eastward. Furthermore, there is a correlation between ice cloud fraction and storm intensity, at least for storms with a large enough intensity (pressure < 980mb). As storm intensity increases, ice cloud fraction increases, suggesting a dynamic control. However, over the Pacific ocean, the behaviour is inverted: the ice cloud fraction decreases as storms move eastward and is anticorrelated with storm intensity, suggesting another influence, perhaps varying aerosol concentration.

Upload: others

Post on 20-Apr-2020

9 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Study of the relationship between ice phase and cloud-top … · 2016-07-29 · Study of the relationship between ice phase and cloud-top temperature with MODIS data Cat h erine Naud

Stud

y of

the

rela

tions

hip

betw

een

ice

phas

e an

d cl

oud-

top

tem

pera

ture

with

MO

DIS

dat

aC

athe

rine

Nau

dan

d A

ntho

ny D

el G

enio

, GIS

S, N

Y

Abs

trac

t: C

loud

feed

back

in G

CM

clim

ate

chan

ge si

mul

atio

ns is

sens

itive

topr

edic

tions

of c

loud

pha

se b

ecau

se o

f the

diff

eren

t mic

roph

ysic

al a

nd ra

diat

ive

prop

ertie

s of l

iqui

d an

d ic

e. U

sual

ly th

e liq

uid-

ice

trans

ition

is p

aram

eter

ized

as a

sim

ple

func

tion

of te

mpe

ratu

re, b

ut ic

e oc

curr

ence

in c

loud

s is k

now

n to

dep

end

on th

e un

derly

ing

dyna

mic

s of t

he c

loud

syst

em.

We

are

usin

g M

OD

IS d

ata

to g

ain

insi

ghts

into

this

beh

avio

ur th

at m

ight

serv

e as

a g

uide

to im

prov

ed p

aram

eter

izat

ions

. C

loud

top

tem

pera

ture

s wer

e ob

tain

ed fr

om5-

min

leve

l 2 M

OD

IS T

ERR

A a

nd A

QU

A d

ay-ti

me

gran

ules

, alo

ng w

ith S

WIR

pha

se fo

r all

avai

labl

e w

inte

rs (D

ec-J

an-F

eb)f

rom

the

star

t of t

he m

issi

ons u

ntil

Febr

uary

200

4. C

loud

y pi

xels

in e

ach

gran

ules

wer

e se

lect

ed o

ver t

he n

orth

ern

parts

of t

he A

tlant

ic a

nd P

acifi

c oc

eans

in th

e ar

ea c

over

ed b

y th

e st

orm

trac

ks a

nd a

ggre

gate

d in

2x2

.5°

grid

box

es. F

or e

ach

grid

box

, all

pixe

ls id

entif

ied

as ic

e on

ly a

nd ic

e or

wat

er a

re se

lect

ed a

nd so

rted

into

tem

pera

ture

bin

s def

ined

eve

ry 2

K fr

om 2

00 to

300

K. H

isto

gram

s can

then

be

obta

ined

that

giv

e th

e fr

actio

n of

ice

clou

d pi

xels

ove

r the

tota

l num

ber o

f ice

and

wat

er c

loud

pix

els i

n a

give

n te

mpe

ratu

re ra

nge.

The

grid

boxe

s are

furth

er

aggr

egat

ed a

long

the

best

est

imat

e of

the

stor

m tr

ack

traje

ctor

yob

tain

ed fr

om a

con

tem

pora

ry N

CEP

rean

alys

is st

orm

trac

k at

las (

Cha

ndle

r and

Jona

s, 19

99) a

nd d

ivid

ed in

to a

wes

t, ce

ntra

l and

eas

t reg

ion.

The

stor

m tr

ajec

torie

s are

cho

sen

so th

at th

e st

orm

freq

uenc

y of

occ

urre

nce

does

not

var

y m

uch

over

the

traje

ctor

y. H

owev

er, w

e fin

d th

at b

oth

mea

n in

tens

ity (c

entra

l pre

ssur

e) a

nd se

verit

y (p

eak

inte

nsity

) inc

reas

e al

ong

the

track

.

Atla

ntic

oce

an

NC

EP st

orm

freq

uenc

y (%

) for

all

win

ters

from

Dec

embe

r 200

0 to

Fe

brua

ry 2

004.

NC

EP st

orm

mea

n in

tens

ity (m

b)

for a

ll w

inte

rs fr

om D

ecem

ber 2

000

to F

ebru

ary

2004

.

NC

EP st

orm

seve

rity

(mb)

for a

ll w

inte

rs fr

om D

ecem

ber 2

000

to

Febr

uary

200

4.

MO

DIS

ice

clou

d fr

actio

n ve

rsus

clo

ud-

top

tem

pera

ture

for a

ll m

onth

s and

bot

h pl

atfo

rms a

ccum

ulat

ed in

the

wes

tern

(s

olid

), ce

ntra

l (da

shed

) and

eas

tern

(dot

-da

sh) p

ortio

ns o

f the

stor

m tr

ack.

Ther

e ar

e at

leas

t 10,

000

pixe

ls in

eac

h 2K

te

mpe

ratu

re ra

nge.

NC

EP st

orm

freq

uenc

y (%

) for

all

win

ters

from

Dec

embe

r 200

0 to

Fe

brua

ry 2

004.

NC

EP st

orm

mea

n in

tens

ity (m

b)

for a

ll w

inte

rs fr

om D

ecem

ber 2

000

to F

ebru

ary

2004

.

NC

EP st

orm

seve

rity

(mb)

for a

ll w

inte

rs fr

om D

ecem

ber 2

000

to

Febr

uary

200

4.

For 4

diff

eren

t tem

pera

ture

rang

es, i

ce

clou

d fr

actio

n ve

rsus

stor

m fr

eque

ncy

(top)

, mea

n in

tens

ity (c

entra

l) an

d se

verit

y (b

otto

m).

The

rang

es a

re: 2

30-

240K

(col

umn

1), 2

40-2

50K

(col

umn

2), 2

50-2

60K

(col

umn

3) a

nd 2

60-

270K

(col

umn

4).

Var

iatio

ns a

long

the

estim

ated

stor

m

mea

n tra

ject

ory

of st

orm

freq

uenc

y (to

p), m

ean

inte

nsity

(cen

tre) a

nd

seve

rity

(bot

tom

) as t

he st

orm

s mov

e ea

stw

ard.

Paci

fic O

cean

For 4

diff

eren

t tem

pera

ture

rang

es, i

ce

clou

d fr

actio

n ve

rsus

stor

m fr

eque

ncy

(top)

, mea

n in

tens

ity (c

entra

l) an

d se

verit

y (b

otto

m).

The

rang

es a

re: 2

30-

240K

(col

umn

1), 2

40-2

50K

(col

umn

2), 2

50-2

60K

(col

umn

3) a

nd 2

60-

270K

(col

umn

4).

MO

DIS

ice

clou

d fr

actio

n ve

rsus

clo

ud-

top

tem

pera

ture

for a

ll m

onth

s and

bot

h pl

atfo

rms a

ccum

ulat

ed in

the

wes

tern

(s

olid

), ce

ntra

l (da

shed

) and

eas

tern

(dot

-da

sh) p

ortio

ns o

f the

stor

m tr

ack.

Ther

e ar

e at

leas

t 10,

000

pixe

ls in

eac

h 2K

te

mpe

ratu

re ra

nge.

Var

iatio

ns a

long

the

estim

ated

stor

m

mea

n tra

ject

ory

of st

orm

freq

uenc

y (to

p), m

ean

inte

nsity

(cen

tre) a

nd

seve

rity

(bot

tom

) as t

he st

orm

s mov

e ea

stw

ard.

Prel

imin

ary

Con

clus

ion:

Ove

r the

Atla

ntic

stor

m tr

ack

regi

on, w

e fo

und

that

as t

he c

loud

s mov

e ea

stw

ard,

the

ice

clou

d fr

actio

n at

a g

iven

tem

pera

ture

tend

s to

incr

ease

, im

plyi

ng th

at h

eter

ogen

eous

free

zing

bec

omes

mor

e im

porta

nt a

s the

stor

ms t

rave

l eas

twar

d. F

urth

erm

ore,

ther

e is

aco

rrel

atio

n be

twee

n ic

e cl

oud

frac

tion

and

stor

m in

tens

ity, a

t lea

st fo

r sto

rmsw

ith a

larg

e en

ough

inte

nsity

(pre

ssur

e <

980m

b). A

s st

orm

inte

nsity

incr

ease

s, ic

ecl

oud

frac

tion

incr

ease

s, su

gges

ting

a dy

nam

ic c

ontro

l.

How

ever

, ove

r the

Pac

ific

ocea

n, th

e be

havi

our i

sinv

erte

d: th

e ic

e cl

oud

frac

tion

decr

ease

s as s

torm

s mov

e ea

stw

ard

and

isan

ticor

rela

ted

with

stor

m in

tens

ity, s

ugge

stin

g an

othe

r inf

luen

ce, p

erha

ps v

aryi

ng a

eros

ol c

once

ntra

tion.