AN N ALS OF GEOPH YSICS, 56, Fast Track - 1, 2013; 10.4401/ ag - 6337 1 Study of thin clouds at CN R-IMAA Atmospheric Observatory (CIAO) MARCO ROSOLDI*, PILAR GUMÀ CLARAMUN T, FABIO MADON N A, ALDO AMODEO, IOAN N IS BIN IETOGLOU, GIUSEPPE D’ AMICO, ALDO GIUN TA, LUCIA MON A, N IKOLAOS PAPAGIAN N OPOULOS, GELSOMIN A PAPPALARDO Con siglio N azion ale d elle Ricerch e, Istitu to d i Metod ologie p er l’ An alisi Am bien tale, Tito Scalo, Poten za, Italy I. IN TRODUCTION p ace-based m easu rem en ts allow u s to stu d y atm osp h eric variables at global scale, bu t grou n d -based observation s are n ecessary for calibration an d valid ation [GCOS, 2006]. Up com in g satellite m ission s aim at sh ow in g th e ben efit com in g from th e availa- bility of tech n ological im p rovem en ts an d in - n ovation s. H ow ever, research is n ecessary to im p rove retrieval algorith m s of atm osp h eric variables an d to assess th e real ad van ces in th e kn ow led ge of w eath er an d clim ate. Qu ality - con trolled vertical p rofiles of atm osp h eric key variables p rovid ed by grou n d -based ad van ced atm osp h eric observatories rep resen t th e op ti- m al basis for th e satellite cal/ val p rogram s. So far, atm osp h eric observatories are w orkin g tow ard s th e d evelop m en t of n ew observation strategies an d th e fu ll exp loitation of th e sy n - ergy am on g active an d p assive p rofilin g se n - sors. Th is is h igh ly relevan t for th e n ew gen er- ation satellites equ ip p ed w ith m u ltip le sen sors on board th e sam e observation p latform . Grou n d -based observation s p rovid e lon g -term m on itorin g of p aram eters th at, at p resen t, ca n - n ot be m on itored from sp ace observation s. For in stan ce, th e stu d y of th in liqu id w ater clou d s as w ell as th e m ech an ism s lead in g to d rop let activation is ch allen gin g for satellite p assive sen sors, w h ereas th ey are im p ortan t for w eath er an d clim ate stu d ies. Th e im p act of th in clou d s on clim ate is h igh ly u n certain . Th in liqu id w ater clou d s, w h ich are low or m id level su p er -cooled clou d s ch aracter- ized by a liqu id w ater p ath (LWP) less th an 100 g m -2, are d ifficu lt to be observed accu rat e- ly an d large d iscrep an cies exist am on g d iffe r- en t observation tech n iqu es [Tu rn er et a l., 2007]. Th e im p ortan ce of th ese clou d s is rela t- ed to th eir exten sive global m ean coverage: low an d m id level clou d s, often con tain in g liq- u id w ater, are ch aracterized by a m ean LWP valu e of 51 an d 60 g m -2, an d cover 27.5% an d 19% of th e global su rface, r esp ectively, w h ile th e global m ean clou d coverage is 68.6% [Rossow an d Sh iffer, 1999]. Grou n d -based p rofilin g tech n iqu es, su ch as ad van ced m u lti-w avelen gth lid ar an d Dop p ler rad ar tech n iqu es, allow u s to stu d y aerosol an d clou d m icrop h ysics; th ey can be co n sid - ered as th e n eed ed brid ge betw een in -situ an d satellite m easu rem en ts to fill in th e gap s in th e u n d erstan d in g of th e p h ysical m ech an ism s lead in g to clou d form ation . In th is w ork, tw o exam p les of p ossible solu - tion s tow ard s th e im p rovem en t of th e accu racy in th e estim ation of th in clou d p rop erties are p rop osed . Th e stu d y is based on th e m u lti- w avelen gth lid ar, Dop p ler rad ar an d m icro- w ave rad iom eter m easu rem en ts p erform ed at * Corresp on d in g au th or: Marco Rosold i, Con siglio N azio- n ale d elle Ricerch e, Istitu to d i Metod ologie p er l’ An alisi Am bien tale, C.d a S. Loja, I – 85050, Tito Scalo, Poten za, Italy; em ail: m arco.rosold i@im aa.cn r.it. S AN N ALS OF GEOPH YSICS, 56, Fast Track - 1, 2013; 10.4401/ ag - 6337 2 th e CN R-IMAA Atm osp h eric Observatory (CIAO). Th e observatory is located in Poten za, Sou th ern Italy, on th e Ap en n in e Mou n tain s (40.60N , 15.72E, 760 m a.s.l.), less th an 150 km from th e West, Sou th an d East coasts [Mad on - n a et al., 2011]. Th e first exam p le ad d resses th e en h an cem en t of Dop p ler rad ar sen sitivity ach ieved ad op tin g a sp ectra avera gin g over th e tim e, w ith th e a im to d etect tin y d rop lets, as th ose form in g fog or th in clou d s. In th e se- con d exam p le, p articu lar atten tion h as been p aid to an op tically th in broken strat ocu m u lu s clou d th at gives u s th e op p ortu n ity to in vest i- gate th e d ifferen ces betw een n on -satu ration an d satu ration con d ition s in an aerosol layer an d th e d rop let activation m ech an ism s, u sin g Ram an lid ar m easu r em en ts. II. TH IN CLOUDS: RADAR SEN SITIVITY Sen sitivity of m illim eter -w avelen gth Dop p ler rad ars allow s to stu d y m icrop h ysical p rop e r- ties of n on -p recip itatin g clou d s. Th ese rad ars p rovid e accu rate reflectivity m easu rem en ts over a d yn am ic ran ge of ap p roxim ately seven ord ers of m agn itu d e, from -50 d BZ to +20 d BZ th rou gh ou t th e trop osp h ere [Kollias et al., 2001]. A large fraction of clou d s is com p osed of h igh con cen tration s of sm all h yd rom eteors w ith d iam eters from a few to ten s of m icrom e- ters, w ith reflectivity valu es from –60 d BZ to – 40 d BZ [Kollias et al., 2001]. Wh en th e reflec- tivity of th e sm all h yd rom eteors is very w eak (less th an -50 d BZ), th ese targets fall ou tsid e th e rad ar d etectable ran ge. Th e p rim ary m easu rem en t p rovid ed by Do p - p ler rad ars is th e Dop p ler sp ectru m . Typ ically, th e first th ree m om en ts of th e rad ar Dop p ler sp ectru m are p rovid ed an d u sed in rad ar d ata an alysis: th e sign al-to-n oise ratio (SN R – zero- th m om en t), th e m ean Dop p ler velocity (first m om en t) an d th e Dop p ler sp ectru m w id th (s e- con d m om en t). For system s op eratin g in zen - ith p oin tin g m od e, th e typ ical tim e resolu tion of a sin gle rad ar vertical p r ofile is of th e ord er of secon d s, w h ile th e typ ical vertical re solu tion is of a few ten s of m eters. In ord er to in crease th e rad ar sen sitivity, th e sp ectra p re-p rocessin g of rad ar m easu rem en ts p erform ed at CIAO station is p erform ed u sin g tw o d ifferen t op tion s: (1) “rou tin e” p re- p rocessin g, based on averagin g 200 sp ectra w ith a vertical resolu tion of 30 m an d a tim e resolu tion of 10 s; (2) “averagin g” p re - p rocessin g, based on averagin g 3600 sp ectra (averagin g th e sp ectra over 3 ran ge gates an d 1 m in u te) w ith a vertical resolu tion of 90 m an d a tim e resolu tion of 1 m in u te. Th e secon d o p - tion allow s u s to en h an ce th e rad ar sen sitivity of 5-10 d BZ, u p to abou t -55 d BZ. Figu re 1 sh ow s th e tem p oral evolu tion of rad ar reflectivity factor Ze in th e “rou tin e” p re - p rocessin g op tion (p an el a) an d in th e “avera g- in g” op tion (p an el b), an d th e lin ear d ep olar i- zation ratio (LDR) in th e “averagin g” op tion (p an el c), m easu red on 10 October 2011 by r a- d ar at CIAO. All th ese tim e series are clu tter filtered . Th e clu tter filterin g is carried ou t after th e estim ation of th e n oise level in th e rad ar p ow er sp ectru m [Bau er-Pfu n d stein , 2007]. In figu re 1a, th e black lin e sh ow s th e h eigh t of th e m eltin g layer retrieved u sin g su rface m ea s- u rem en ts of tem p eratu re an d th e m od eled tem p eratu re p rofiles of a stan d ard atm osp h ere. In both p rocessin g op tion s, Ze in d icates th e p resen ce of ice clou d at arou n d 7 km above sea level (a.s.l.) at 13:00 UT an d low -level liqu id w ater clou d s below 3 km a.s.l., altern ated w ith occasion al rain sh ow ers. Moreover, liqu id w a- ter layers are d etected u p abou t 2 km a.s.l. be- tw een 00:00 an d 04:00 UT an d betw een 17:00 an d 20:00 UT, on ly u sin g th e “averagin g” p re- p rocessin g op tion (figu re 1b). Th is is d u e to th e tin y size of th e d rop lets of th ese layers, likely fog, w h ose liq u id p h ase is con firm ed by th e absen ce of sign atu re in th e LDR (fig u re 1c). Th e d iscu ssed case sh ow s h ow th e averagin g of rad ar sp ectra over larger vertical an d tem - p oral d om ain s a llow s th e rad ar to exten d its ow n sen sitivity: th is is h igh ly relevan t in th e stu d y of th in clou d s an d fog. An en h an ced ra- d ar sen sitivity allow s u s to exten d th e size ran ge of h yd rom eteors d etected by th e rad ar as w ell as to con tribu te to a better exp loitation AN N ALS OF GEOPH YSICS, 56, Fast Track - 1, 2013; 10.4401/ ag - 6337 3 Figu re 1: (a) Tim e series of vertical p rofiles of clu tter filtered rad ar Reflectivity Factor (Ze) m easu red on 10 October 2011 at CIAO u sin g th e “rou tin e” p rocessin g (tim e resolu tion is 10 secon d s an d vertical resolu tion is 30 m eters), black lin e in d i- cates th e m eltin g layer h eigh t; (b) sam e as rep orted in p an el (a) bu t u sin g th e “avera gin g” p rocessin g (tim e resolu tion is 1 m in u te an d vertical resolu tion is 90 m eters); (c) tim e series of vertical p r ofiles of clu tter filtered rad ar Lin ear Dep olarization Ratio (LDR) for th e sam e d ate, obtain ed u sin g th e “averagin g” p rocessin g. of in stru m en t syn ergies. Th e stu d y of aero- sol, w ater vap or an d liqu id w ater in th in clou d s can also be con sid ered as a good o p - p ortu n ity for th e qu an tification of th e aerosol in d irect effect. Th is w ork is cu rren tly on g o- in g at CIAO alon g w ith th e evalu a tion of d ow n d rafts an d u p d rafts o bserved w ith th e Dop p ler rad ar. AN N ALS OF GEOPH YSICS, 56, Fast Track - 1, 2013; 10.4401/ ag - 6337 4 III. TH IN CLOUDS: RAMAN LIDAR AN D DROPLET ACTIVATION Aerosol p articles m ay be activated in to d r o- p lets w h en su p er -satu ration con d ition s are reach ed in a risin g air m ass. Clou d d rop let activation is a h igh ly com p lex an d n on lin ear p rocess: on ly a fraction of p articles can grow beyon d th e critical sizes to form d rop lets, d ep en d in g on several factors su ch as size d istribu tion , ch em ical com p osition an d u p - d raft velocity [Min g et al., 2006]. From th e grou n d based rem ote sen sin g p oin t of view , th e p ossibility of in vestigatin g th is p rocess is related to th e cap ability of th e in - stru m en ts to p rovid e accu rate estim ation of aerosol an d clou d s p rop erties. Th in clou d s are on e of th e p ossible targets th at allow u s to stu d y th ese p rocesses. Th ou gh th in liqu id w ater clou d s are d ifficu lt to be observed a c- cu rately an d large d iscrep an cies exist am on g d ifferen t observation tech n iqu es [Tu rn er et al., 2007], th ey are effectively observed th rou gh grou n d -based rem ote sen sin g rath er th an th rou gh p assive satellite rem ote sen s- in g. At CIAO, tw o m u lti-w avelen gth Ram an lid ars, PEARL (Poten za EArlin et Ram an Lid ar) an d MUSA (Mu lti-w avelen gth System for Aerosol), are op eration al [Mad on n a et al., 2011]. Both system s em it ligh t p u lses at 355, 532 an d 1064 n m an d d etect th e elastically backscattered rad iation by atm osp h eric co n - stitu en ts an d th e Ram an backscattered rad ia- tion by n itrogen m olecu les at 387 an d 607 n m , an d by w ater vap or m olecu les at 407 n m , th e last on ly for th e PEARL syst em . Ver- tical p rofiles of aerosol op tical p rop erties an d w ater vap or m ixin g ratio are retrieved from lid ar sign als acqu ired w ith a raw vertical r e- solu tion of 15 m (7.5 m for 1064 n m ) for PEARL an d 3.75 m for MUSA, an d a raw tim e resolu tion of 60 secon d s. Figu re 2a sh ow s th e tem p oral evolu tion of th e lid ar ran ge-corrected sign al at 1064 n m m easu red by MUSA on 4 Au gu st 2011 from 19:22 to 21:16 UT. Th e p lot sh ow s th e p r e- sen ce of broken clou d s betw een 3 an d 4.5 km of altitu d e a.s.l. Th e observ ation of broken clou d s gives u s th e op p ortu n ity to calcu late th e vertical p rofiles of aerosol op tical p rop e r- ties an d w ater vap or con ten t in tw ofold m an n er: “w ith clou d s”, by averagin g all li- d ar sign als, an d “w ith ou t clou d s”, by skip - p in g th e sign als sh ow in g th e p resen ce of clou d s. Th e n u m ber of skip p ed sign als in th e p rofiles “w ith ou t clou d s” is exp ressed th rou gh th e so-called skip p ed fraction , d efi- n ed as th e n u m ber of skip p ed sign als d iv i- d ed by th e total n u m ber of sign als. Th e clou d con tam in ated lid ar retu rn s h ave been id en tified by visu al in sp ection an d assu m in g th at a stron g n egative slop e in th e n itrogen Ram an lid ar sign als at 387 n m corr esp on d s to th e p resen ce of liqu id w ater d r op lets. Th is clou d d etection tech n iqu e h as been u sed p r e- lim in arily in ord er to verify th e reliability of all th e p rofiles con tain in g clou d s. H ow ever, th e p roced u re can be au tom ated , an d th is w ill be d on e in th e n ear fu tu re th rou gh th e u se of th e EARLIN ET Sin gle Ca lcu lu s Ch ain (SCC) for th e au tom atic retrieval of th e aer o- sol op tical p rop erties startin g from raw lid ar sign als [D’ Am ico et al., 2012]. Figu re 2b sh ow s th e vertical p rofiles of aer o- sol extin ction coefficien t at 355n m “w ith clou d s” (red lin e) an d “w ith ou t clou d s” (black lin e) obtain ed u sin g Ram an lid ar tech n iqu e, averagin g lid ar sig n als over th e sam e tim e in terval of 17 m in u tes, from 19:33 to 19:50 UT. Besid es, figu re 2c sh ow s th e ver- tical p rofiles of relative h u m id ity “w ith clou d s” (red lin e) an d “w ith ou t clou d s” (black lin e) estim ated w ith Ram an lid ar tech n iqu e, averagin g lid ar sign als from 19:33 to 19:50 UT, an d calibrated w ith w ater vap or m easu rem en ts p erform ed by a co -located an d sim u ltan eou s rad io -sou n d in g. In both figu res 2b an d 2c th e p rofiles h ave an effe c- tive vertical resolu tion of 780 m an d a skip p ed fraction of 0.29. Fin ally, in figu res 2d an d 2e th e sam e as in figu res 2b an d 2c is r e- p orted , bu t averagin g lid ar sign als over a tim e in terval of 103 m in u tes, from 19:33 to 21:16 UT, w ith an effective vertical resolu tion AN N ALS OF GEOPH YSICS, 56, Fast Track - 1, 2013; 10.4401/ ag - 6337 5 Figu re 2: (a) Tim e series of ran ge-corrected sign al at 1064 n m obtain ed from MUSA on 4 Au gu st 2011 from 19:22 to 21:16 UT u sin g a tim e resolu tion of 1 m in u te an d a vertical r esolu tion of 3.75 m ; (b) vertical p rofiles of 355 n m aerosol extin ction coefficien t retrieved u sin g lid ar sign als in clu d in g clou d s (“w ith clou d s” - red lin e) an d skip p in g clou d s (“w ith ou t clou d s” - black lin e) an d averagin g th e lid ar sign als over 17 m in u tes, from 19:33 to 19:50 UT; (c) vertical p rofiles of relative h u m i d ity “w ith clou d s” an d “w ith ou t clou d s”, corresp on d in g to th e p rofiles rep orted in p an el (b), calibrated u sin g a co-located an d sim u ltan eou s rad iosou n d in g. In both (b) an d (c) p an els, th e p rofiles h ave an effective ve rtical resolu tion of 780 m (skip p ed fraction = 0.29); (d ) vertical p rofiles of 355 n m aerosol extin ction coefficien t retrieved u sin g lid ar sign als in clu d in g clou d s (“w ith clou d s” - red lin e) an d skip p in g clou d s (“w ith ou t clou d s” - black lin e) an d averagin g th e lid ar sign als over 103 m i- n u tes, from 19:33 to 21:16 UT; (e) vertical p rofiles of relative h u m id ity “w ith clou d s” an d “w ith ou t clou d s”, corresp on d in g to th e p rofiles rep orted in p an el (d ), calibrated u sin g a co-located an d sim u ltan eou s rad iosou n d in g. In both (d ) an d (e) all th e p rofiles h ave an effective vertical resolu tion of 660 m (skip p ed fraction = 0.40). AN N ALS OF GEOPH YSICS, 56, Fast Track - 1, 2013; 10.4401/ ag - 6337 6 of 660 m an d a skip p ed fraction of 0.40. Th e error bars in th e p rofiles of figu res 2b an d 2c are larger th an th ose of figu res 2d an d 2e, becau se th e sm aller th e tim e in terval over w h ich th e lid ar sign als are averaged , th e lo- w er th e sign al to n oise ra tio. In figu res 2b an d 2d , extin ction coefficien t p rofiles for both “w ith clou d s” an d “w ith ou t clou d s” cases are fou n d to be in good agr e- em en t reach in g an altitu d e of abou t 3.2 km a.s.l. Th is su ggests th at aerosol p articles d o n ot n u cleate. In th e region a bove, “w ith clou d s” p rofiles (red lin es) rise sign ifican tly com p ared to “w ith ou t clou d s” p rofiles (black lin es) becau se of th e aerosol activation an d th e d rop let form ation . From abou t 4.5 km a.s.l., th e extin ction p rofiles “w ith clou d s” an d “w ith ou t clou d s” are in good agreem en t again . Th erefore, th e layer exte n - d in g from 3.2 km to 4.5 km a.s.l. can be co n - sid ered as th e region w h ere aerosol p a rticles are activated an d d rop lets form . Th ese r e- su lts seem to be con firm ed by th e relative h u m id ity p rofiles sh ow n in figu res 2c an d 2e. Su ch p rofiles sh ow a sign ifican t in crease of relative h u m id ity w ith in th e activation la- yer an d a satu ration con d ition reach ed in th e m id d le p art of th e activation region w h ere d rop lets form . Th e in terp retation of su ch p rofiles n eed s fu rth er in vestigation . Th e p resen ted case stu d y reveals th e a d - van tage of stu d yin g broken th in liqu id w ater clou d s, as th ey offer th e p ossibility to ch ara c- terize th e aerosol activation region an d to sep arate it from th e region n ot affected by th e activation p rocess. IV. SUMMARY Qu ality-con trolled vertical p rofiles of atm o s- p h eric key variables p rovid ed by ad van ced atm osp h eric observatories, like CIAO, are h igh ly relevan t for d efin ition an d im p lem en - tation of su itable strategies for calibration an d valid ation for both geostation ary (e.g. Metop ) an d p olar orbitin g (e.g. OMI, Te r- ra/ Aqu a, ADM-Aeolu s an d Earth CARE) satellite p latform s. Th ey allow u s to stu d y m acrop h ysical an d m icrop h ysical p rop erties of aerosol an d clou d s over lon g term , in clu d - in g p aram eters ch allen gin g to be ad d ressed w ith satellite observation s. Th in clou d s rep resen t on e of th e m ajor sou rce of u n certain ty in th e estim ation of global rad iative balan ce an d in th e ch aracte r- ization of aerosol-clou d in teraction s. Tw o exam p les relative to th e u se of Dop p ler rad ar an d Ram an lid ar m easu rem en ts h ave been d iscu ssed . Th e first stresses th e n eed for u s- in g a su itable tem p oral an d vertical avera g- in g of rad ar sp ectra in ord er to in crease th e sen sitivity of rad ar in th e d etection of sm all d rop lets, as th ose form in g fog or th in clou d s. Th e secon d d escribes th e ad van tage of stu d - yin g th in broken clou d s, u sin g Ram an lid ar m easu rem en ts, to d istin gu ish th e region s w h ere aerosol p articles are activated or n ot. Com p lem en tary w ater vap or m ixin g ratio m easu rem en ts w ith Ram an lid ar con tribu te to id en tify th e d rop let form ation region . Th e p resen ted m eth od ology is cu rren tly u n d er in vestigation over a large d ataset of rad ar an d lid ar m easu rem en ts collected at CIAO in th e p eriod 2009-2012. Stu d y of th in liqu id w ater clou d s, as w ell a s th e related m ech a- n ism lead in g to d rop let activation , is ch a l- len gin g u sin g sp ace observation s on ly. Grou n d based lid ar an d rad ar observation s can stron gly con tribu te to im p rove satellite retrievals an d to ad vise on th e tech n ological requ irem en ts for fu tu r e satellite m ission s. ACKN OWLEDGMEN TS Th e au th ors gratefu lly ackn ow led ge Dr. Ma t- th ias Bau er for h is su p p ort in th e rad ar d ata p rocessin g. Th e fin an cial su p p ort for EARLIN ET by th e Eu rop ean Un ion u n d er gran t RICA 025991 in th e Sixth Fram ew ork Program m e is gratefu lly ackn ow led ged . Sin ce 2011 EARLIN ET is in tegrated in th e ACTRIS Research In frastru ctu re Project su p - p orted by th e Eu rop ean Un ion Seven th Fram ew ork Program m e (FP7/ 2007 - 2013) AN N ALS OF GEOPH YSICS, 56, Fast Track - 1, 2013; 10.4401/ ag - 6337 7 u n d er gran t agreem en t n . 262254. REFEREN CES [Bau er-Pfu n d stein , 2007] Bau er-Pfu n d stein , M., an d U. Gö rsd orf (2007). Target sep ar a- tion an d classification u sin g clou d rad ar d op p ler-sp ectra, in Proceed in gs 33rd In tern . Con f. on Rad ar Meteorology, Cairn s. [D’ Am ico et al., 2012] D’ Am ico, G., et al. (2012). EARLIN ET Sin gle Calcu lu s Ch ain for au tom atic lid ar d ata p rocessin g: first tests on op tical p rod u cts, Review ed an d Revised P a- p ers of th e 26th In tern ation al Laser Rad ar Con feren ce, Volu m e 1, p p . 331-334. [GCOS, 2006] GCOS rep ort N .107, System a t- ic Observation Requ irem en ts for Satellite- based Prod u cts for Clim ate, Su p p lem en tal d etails to th e satellite-based com p on en t of th e Im p lem en tation Plan for th e Global O b- servin g System for Clim ate in Su p p ort of th e UN FCCC, Sep tem ber, 2006. [Kollias et al., 2001] Kollias, P., B. A. A l- brech t, R. Lh erm itte, an d A. Savtch en ko (2001). Rad ar observation s of u p d rafts, d ow n d rafts, an d tu rbu len ce in fair -w eath er cu m u li, J. Atm os. Sci., 58, 1750–1766. [Mad on n a et al., 2011] Mad on n a, F., Am od e- o, A., Boselli, A., Co rn acch ia, C., Cu om o, V., D'Am ico, G., Giu n ta, A., Mon a, L., an d Pa p - p alard o, G. (2011). CIAO: th e CN R-IMAA ad van ced observatory for atm osp h eric r e- search , Atm os. Meas. Tech ., 4, 1191-1208, d oi:10.5194/ am t-4-1191-2011. [Min g et al., 2006] Min g, Yi, V. Ram asw am y, Leo J. Don n er, Vau gh an T. J. Ph illip s (2006). A N ew Param eterization of Clou d Drop let Activation Ap p licable to Gen eral Circu lation Mod els, J. Atm os. Sci., 63, 1348–1356, d oi: h ttp :/ / d x.d oi.org/ 10.1175/ JAS3686.1. [Rossow an d Sh iffer, 1999] Rossow , W.B., an d R.A. Sch iffer (1999). Ad van ces in u n d er- stan d in g clou d s from ISCCP. Bu ll. Am er. Meteor. Soc., 80, 2261-2287. [Tu rn er et al., 2007] Tu rn er, D. D., an d Coa u - th ors (2007). Th in liqu id w ater clou d s: Th eir im p ortan ce an d ou r ch allen ge, Bu ll. Am er. Meteor. Soc., 88, 177–190.