Verification of precipitation calculation method based on parameterization of distribution function evolution. I.V.Akimov Hydrometeorological Centre of Russia 11-13 Bolshoy Predtechensky per., 123242, Moscow-242, Russia. E-mail: [email protected]

The new precipitation calculation method presented in [1] was evaluated. The method is based on different approach, than used in Kessler-type parameterizations. In Kessler approach [2] the precipitation beginning from cloudy layer is permitted after cloud water content reaches some value, named as autoconversion threshold. But the value of this threshold could not be obtained from the measurements and the hypothesis about such threshold contradicts with the observed facts that precipitation starts from clouds when cloud water content rises up to quite different values [3,4]. Therefore the new method is based on consideration of precipitation beginning process as formation of part of cloud spectra consisting of large drops which falls out as a precipitation. This is realized by parametric description distribution function evolution during precipitation formation process. The main processes that influence on distribution function evolution are gravitational and turbulent coalescence in liquid clouds and sublimation of water vapour on ice crystals in mixed phase clouds. The proposed method was included as precipitation parameterization module in global spectral model of the Hydrometcentre of Russia T85L31. The precipitation calculation module that was used in old version of model is based on Kessler parameterization. Therefore the comparison of precipitation fields obtained from new version of model with results obtained from old version allows to evaluate the distinction between two different parameterization approaches. Two scores were chosen for verification analysis. The accuracy of precipitation spatial distribution forecast was characterized by Pearcy-Obychov criterion T (also known as Hanssen and Kuipers discriminant). The value of this score vary from T=1 for ideal forecast to T=-1 for absolutely incorrect forecast. The accuracy of precipitation quantity calculation was characterized by root-mean-square error σQ. The evaluation of precipitation forecasts obtained from two versions of model was conducted during time period March-December 2002. The values of T and σQ scores are presented on the figure. The estimation was performed using precipitation observation data from four regions in northern hemisphere. Figure shows that T criterion values resulted from new model version precipitation forecast is greater during all estimations period than values obtained from old model version integration. So for European part of Russia in period April-July criterion T value rises from level 0.2-0.3 to level 0.3-0.5. The usage of new precipitation calculation method also reduces forecast error σQ in all regions except North America. The values σQ is became smaller on 10-20% in comparison with errors resulted from old version of model. In North America at time period June-August the new version of model gives σQ values that overcame error level of old model version. The reason of this fact that new precipitation calculation method uses empirical relations and values of parameters, which were obtained from aircraft measurements data performed in middle latitudes region. But the weather conditions in south coast of North America in summer season is close to tropical weather. Therefore the usage of such parameters in this season is not quite correct. The possible way to eliminate this is to select different values of parameters included in new method for different climatic regions. In common figure shows that incorporation of new method in global spectral atmospheric model resulted in improvement of precipitation forecast in midlatitude region. a)

b)

c)

d)

Pearcy-Obychov criterion T and forecast root-mean square error σQ for 24 hours precipitation forecast, calculated for verification period March - December 2002. Versions of model: 1- old version, 2- version including new method of precipitation calculation. Regions of estimations: European part of Russia (a), Central Europe (b), West Europe (c), North America (d).

References:

1. Akimov I.V., 2003. Method of precipitation intensity calculation based on microphysical processes parameterization in liquid and mixed phase clouds. Izvestia. Atmospheric and oceanic physics. v. 39. 2. Kessler E., 1969. On the distribution and continuity of water substance in atmospheric circulations. Meteor. Monogr., vol. 10. 3. Matveev L.T., 2000. Physics of atmosphere. S.Pb. Hydrometeoizdat. 4. Mason B.J., 1971. The physics of clouds. Clarendon Press, Oxford.

¢¡¤£¦¥¨§ © ¥  ¦¥   ! "$#%&'¥¨£)(+*,¡.-0/1¥2 3

465$798$:<;=3765?>A@3BDCFEG;IH$JK=?LM5N>PO C¦ST@3U3VXWY5?Z[J]\\C!S

Q3RR

^<_a`cb'dfe'g!_ah i_jbbk_ah'lcmn_ao!dbIprqNsutvs wyx0z6{}|~|~€¦p€‚~||ƒYt „_aoF !†e'g‡†!s ˆAs‰p‹ŠŒ_Ihf9†~oŽ

‘’M“”]•—–™˜Iš›'œ‘Ÿž›f D¡ ¢~£}¤¦¥§2¢ƒ¨ ¢!¡ ¢›}©¢~›kŸª«›'¬c¡ ¤[­D®¯›f¨¨¥K°~±‰§2¢ƒ¨ ¢!¡ ¢›}©­Iš~ Ÿ›'ª«±²­~¡ ¤Œš›'ª«ª²›f f§2¢~¨ ¢!¡ ¢~›}©F³c£ƒ¢~£¡ Ÿ±uKŸ›k f§2¢~¨ ¢¦¡ ¢~›

ˆ.xdb<´hfm]lFµKdfeI†ƒ¶n_[ej¶nx~`!l·dfe'g¦_IY_Id<`!d_}l.mno¤¸"i+q¨µ¹Yxclc_I¶ndŒdx¶‰º_[x~o!¶‰Ž‡x~o¦_v»¦hkx~´~o!xdbfm]e¦_I¼`!†ƒbfmnx~o.½¾xh

lc_ao‹df†ƒbf_~s"À _ao!ea_~p‹bfg¦_vlcmndbfmno!ejbfmnx~o¿x~½?bkg¦_ÂÁ2†‘bk_ah †~o!l‡bfg!_Âm]ej_[»¦g‹†~df_[g!†~d bfxà r_vlc_abf_IhfYmno¦_Il.lcm]†ƒ´o¦xdbfm]ea†~¶‰¶nŽÃ½¾xh

bf_IÄ»r_ah'†‘bk`¦hk_IdÄ ‹_I¶‰x‘ÁÅbkg¦_‡½¾hf_I_aÆIm‰o¦´¤»‹xm‰obYÇÈ¿ÉÊ|Ë}ÌÄsAÍ2g¦m]dvm]dÄ`!df`!†ƒ¶n¶‰ŽAl¦x~o¦_‡ FŽÏÎÐm]Ñv»¦hk_Idkejhkmn ¦m‰o!´¿bfg!_M¶nmn¼`¦m]l

½¾hk†eDbkm‰xo¿m‰o·bfg¦_vbfx~bk†~¶?eax~o!lc_Io!dk†‘bf_Ć~dŒ†9½¾`¦o‹eDbfmnx~oÓÒ0Ô¨x~½™bf_aY»r_ah'†‘bk`¦hf_ƃo!lXξmnm]Ñ<†~dkdf`¦Ym‰o¦´Gbfg!†ƒb< rxƒbkg¤ej¶nx~`‹l¿m]ej_

†ƒo!lTea¶‰x`!lÓÁ2†‘bf_Ih[†ƒhk_9m‰oÓbkg¦_ahkYxclcŽFo!†ƒYm]e9_I¼`¦mn¶‰mn ¦hkm‰`!ÕÁ m‰bfgThk_Idf»‹_}eDb[bkx¿†.gFŽF»‹x~bfg¦_abfm]ea†ƒ¶ydf†ƒbf`¦h'†‘bkm‰xo¤º‘†~»‹x`¦h

»¦hk_Idkd`¦hk_ ´~mnº~_IoY ŽÄ֑×2ÉØÒ0ԑÖ0٠Ο{?ÛMÒ0Ô¾ÑcÖ0Ü pÁ g¦_ahk_"Ö0Ù †ƒo‹l9Ö}Ü݌†ƒhk_2bfg!_ df†ƒbf`¦h'†‘bkm‰xoYº0†~»‹x`¦h¨»¦hk_Idkdf`¦hf_ x‘º_ah™Á2†‘bk_ah

×[Ú × ×

È

†ƒo!l·mnea_~pÞhk_Idf»‹_}eDbkm‰º_a¶nŽ~s[Í2g¦_v½¾`¦o!ejbfmnx~oTÒ0ԙ½¾x~hŒbkg¦_v¶nm]¼`!mnl·½¾hk†eDbkm‰xo¿mndŒ`!df`!†~¶‰¶nŽ¿e'g¦xdf_ao¿bfx‡ ‹_6{v½¾x~hŒÇàßáÇ †ƒo!l

|Y½¾x~h bf_IÄ»r_ah'†‘bk`¦hk_Id" r_a¶nx‘ÁP†Ybfg!hf_}dg¦x¶nl6Ç Á m²bkg·†9¶nm‰o!_I†ƒh

Ü]âKã

È

hk†~o¦´~_ŒÇ Ç Ç s™å¨†ƒhkmnx~`!dyº‘†ƒ¶n`¦_Id[Û |ËaÌ Ç Û¦{}|ËIÌæ†~hf_¦†~dkdf`¦Y_IlMm‰o.lcm‰„r_Ihf_Iob dke'g¦_IÄ_}das

ÜnâKã<ä ä ä ÜnâKã<ä

ç

ej¶]†~dkd™x~½ea¶‰x`!l9dfe'g!_aY_Id™dfm‰Y»¦¶nŽÄo¦_a´¶‰_}eDbkd™bfg¦_ ea¶‰x`!lÄmnea_"»¦g!†d_[Î¹Ò Éá{2½¾xh †ƒ¶n¶¦bk_aY»‹_Ihk†ƒbf`¦hk_Id'ÑN†~d™½¾x~h¨mno!dbk†ƒo‹ej_

Ô

lcx~o!_Âmno¿bfg!_vx»‹_Ihk†ƒbfmnx~o!†~¶$dke'g¦_aY_Id

dŸbkhk†ƒbf_I´~Ž~p‘g!x‘Áî_Iº~_ah}p}hf_}d`¦¶‰bkdNmnov†ŒÁ hkx~o¦´ bfg¦_IhfYxclcŽFo!†~Äm]eydŸb'†‘bf_yxƒ½ejmnhfhk`!d?ej¶nx~`!l!d ÎІƒ¶nÁ2†0ŽFd3Áy†ƒbf_IhNea¶‰x`!l¦dN†‘b™Á2†‘bk_ah

df†ƒbf`¦h'†‘bkm‰xo‹Ñ3Á m²bkgÄeax~hkhf_}d»rx~o!l¦m‰o¦´<_ahkhfxhkd3bfg!_2ea¶‰x`!lFµ¯hk†lcm]†‘bfmnx~o[½¾_a_Il¦ !†~e'ïp~†~o!l[mnoƌ¶]†ƒhk´~_xdm‰bfmnº~_2 ¦m]†~d?xƒ½r`¦»¦»r_ahfµ

¶‰_Iº~_I¶¦gF`¦Ymnlcm‰bŸŽ~s t

g!†0º~_膃¶]dfxv†vo`!Â r_ah2xƒ½Nejx~o‹ej_a»¦bfmnx~o!†~¶rlch'†0Á !†e'ïFdIsNð$mnh'dŸb}pbkg¦_¦†~dkd`!Ä»¦bfmnx~oGx~½3bkg¦_ahkYxFl¦Žo‹†ƒYmne<_I¼`¦mn¶nm‰ ¦hkmn`¦ñxƒ½

È

‹x~bfg6Á2†‘bf_Ih †ƒo!lMmnea_¦†‘b bf_IY»‹_Ihk†ƒbf`¦hk_Idy r_a¶nx‘ÁØÇ mnd2o¦x~b"m‰o6†~eIejxhkl¦†~o!ej_<Á m²bkg6»¦gFŽFdfm]ea†ƒ¶»¦hkmno!ejmn»¦¶n_IdIs òF_Ieax~o!lÞpF½¾xh

Ç Ç †Gdk†‘bk`¦h'†‘bfmnx~o¿†l0óŸ`!dbfY_aobŒmndŒlcx~o!_[½¾x~h bfg¦_Äea†~¶nea`¦¶n†ƒbfmnx~o¿x~½¨ejx~o‹lc_ao!dk†‘bk_~ôÞdfm‰o!ea_[bkg¦_ÂoF`¦v ‹_Ih

ä Ü]âKã

mnea_veahfŽcdbk†ƒ¶]d"m]d"º_ahkŽ‡df9†ƒ¶n¶—pdf`!e'g·†ƒo.mno!dbk†~ob'†ƒo¦_Ix~`!d<†~l0óŸ`‹dŸbkÄ_Iob

bfg¦_

_I¼`¦mn¶‰mn ¦hkm‰`!‡s™ð!x~`¦hfbfgëpbfg!_ŒòF_I_Ilc_IhµKð¦_I_Ilc_Ih¨Y_Ie'g!†~o¦mndfàmndîo¦xƒbîhf_I»¦hf_}d_Iobf_Ilëõ$lc_I_a»Gea¶‰x`!l¦d †~hf_ Yxhf_"¶nmnï~_a¶nŽvbfx[ r_

ç

´~¶]†~eamn†ƒbf_IlGbfg‹†ƒo6bfg!m‰o.ej¶nx~`‹l¦d"†‘b bkg¦_Âdk†ƒY_Œbf_IÄ»r_ah'†‘bk`¦hk_~s o!lMö¦½÷bfg3p‹m]ej_è½Ð†~¶‰¶nm‰o!´Y½¾hfxø†ƒ rx‘º~_èmnobfxÃdf`¦ cµ¹½¾hf_I_aÆamno¦´

¶n†0Ž_ah'dymndy½¾xhkea_IlGbfx9Y_a¶‰b"m‰o6xhklc_IhybfxY9†ƒmnob'†ƒmno‡bfg¦_¦»¦hk_Idkejhkmn ‹_}lG¶nmn¼`¦m]lG½¾hk†eDbfmnx~oGùÃbfg!mnd2m]d o¦x~b"º~_ahkŽ9hk_I†ƒ¶nm]dŸbkmne~s

wy_I†ƒhkmno¦´.m‰oTYmno!lÓbkg¦_Idf_Glcm‰úÃej`¦¶‰bfmn_IdIp™†6o¦_aÁÅ»!†~hk†~Ä_abf_IhfmnÆI†ƒbfmnx~oÓdfe'g¦_IY_ÃÁy†d[lc_}dmn´~o¦_}l¤bfx.b'†ƒï~_Gm‰obfx¿†eaeax~`¦ob

ç

ej¶nx~`!lGmnea_Œ†~o!lGea¶‰x`!lÃÁ2†‘bk_ahy FŽÃ†vdf_a»!†~hk†ƒbf_<»¦hfx´~o¦xdŸbkmne" ¦`!lc´_jb _I¼`!†ƒbfmnx~o!dIs d †[o¦x‘º_a¶‹½¾_I†ƒbf`¦hk_

Áî_·½¾x~hkÂ`¦¶]†‘bk_¿bkg¦_¤lc_I»‹xdm‰bfmnx~o‹†ƒ¶Œ´~hkx‘Á2bfg)xƒ½Âea¶‰x`!l)m]ej_ӆ~dM†Ao¦xocµ¹_}¼`¦m‰¶nmn ¦hfmn`¦û»¦hkxcej_IdkdG†~o!l)hk_I¼`¦mnhf_p †‘b6†ƒ¶n¶

bf_IÄ»r_ah'†‘bk`¦hk_IdIp!df†ƒbf`¦h'†‘bkm‰xoGÁ m‰bfg.hk_Idf»‹_}eDb bkx9Áy†ƒbf_ah2½¾xh"ej¶nx~`!l‡¶nmn¼`¦m]l‡Áy†ƒbf_Ih2bfx9_jzcm]dŸb}s Í2g¦_¦_azF»!¶‰m]ejm‰b ea†~¶nea`¦¶n†ƒbfmnx~o

xƒ½ ej¶nx~`!lm]ej_9lc_I»‹xdm‰bfmnx~o!†~¶¨´~hkx‘Á2bfg¤mnd[ !†~df_Il¤x~obfg!_9Y†dfdµ¯´~hkx‘Á2bfg¿_}¼`‹†‘bfmnx~oÓx~½ydfm‰o¦´¶‰_9»!hfm]dŸbkm‰o!_ÃejhkŽFdbk†~¶ndIs9wyŽ

»¦hk_Idkejhkm‰ ¦mno¦´Gea¶‰x`!l.m]ej_v»!†ƒhfbfm]ej¶n_Id<†~d bkg¦mno·g¦_azc†~´~xo!†ƒ¶ë»!¶n†ƒbf_Id<Á m‰bfg¤†ÃÄxo¦xclcm]d»r_ah'd

´~hkx‘Á2bfgMhk†ƒbf_èxƒ½™ej¶nx~`!lMm]ej_¦Ym²zcmno¦´Yhk†ƒbfmnx~o6ü lc`!_èbfxÃlc_I»‹xdm‰bfmnx~oÓÎÐdf`¦ ¦¶nm‰9†ƒbfmnx~o‹Ñy9†0Ž9bkg¦_ao6 r_[lc_ahkm‰º_Il‡†~d

Ü

×

Ü

Î!ýü Ñfþ É¡ Σ¢ Ñ¥¤§¦§¨ Σ©ü ѧ ¥¦§¨ëü ÝÂÎ ÛX{ kÑ ÎŸ{0Ñ

Ü ã—ÿ Ü Ü Ü Ü

×

Ý  Ý

m‰o¤bf_Ihf9d

Ý Ü

Ü

×

ü ÝømndÄm‰bkdÄdf†ƒbf`¦h'†‘bkm‰xoAº‘†ƒ¶n`¦_Gx‘º_ahÂbfg¦_‡mnea_~s m]dƤd¶nx‘Á ¶nŽ¤º‘†ƒhkŽFm‰o!´¿½¾`¦o!ejbfmnx~oXxƒ½"bk_aY»‹_Ihk†ƒbf`¦hk_M†~o!lA»!hf_}dfdf`¦hk_~p

Ü

Ü

Á g¦m]e'gTmndv†ƒ»¦»¦hkx0zcm‰9†ƒbf_Il FŽ¤†·ejx~o‹dŸb'†ƒob[º‘†ƒ¶n`¦_Ãxƒ½ x~½è{ ¦{}|.ξmno ejxhfhk_Idf»‹xo!lcmno¦´¿ò¯µ¯`¦o¦m‰bkd'ÑDs¿Í2g¦_ÃoF`¦v ‹_Ih

lc_ao‹dm‰bŸŽ¢ ξÇ<Ñ xƒ½$ea¶‰x`!lFµ¯mnea_ »!†~hbkmnea¶‰_}dymndy»!†~hk†~Ä_abf_IhfmnÆa_}l9†~dy†[½¾`¦o‹eDbfmnx~o‡xƒ½?†ƒmnh bk_aY»‹_Ihk†ƒbf`¦hk_ `‹dmno¦´vbfg¦_Œhf_I¶n†ƒbfmnx~o

Ü

È

¢ ξÇ<љÉ¢ _azc»$| Î¾Ç Û·ÇŒÑ"!¦Á m²bkg#¢ Éá{ |$ƒ{}| pcÁ g¦m]e'g6mnd †~oM_IÄ»!m‰hkmneI†ƒ¶ö¦b bkx9†0º0†~m‰¶]†ƒ !¶‰_èl¦†‘b'†v½¾hfx

Ü Ü 

¤&% ¨

È È

Ü

†ƒmnhkeahk†ƒ½÷bŒY_}†~df`¦hf_IY_aobkdŒmnoÓdbfh'†‘bkm²½¾xhfÊea¶‰x`!l¦dIs('x`!l·m]ej_Ym]dèm‰o!m²bkmn†~¶‰¶nŽ¿½¾xhfY_Il¤ FŽ¿g¦_abf_ahkx~´_ao¦_Ix~`!d

x~h2g¦xÄx´~_Io¦_ax`!dî½¾hf_I_aÆamno¦´Äxƒ½™d`!»‹_Ihkeaxx¶‰_}lGlchkx~»!¶‰_abkdIsNÍ2g!_Œ¶]†‘bfbf_Ih »¦hfxcea_Idkdîm]d2»!†~hk†~Ä_abf_IhfmnÆa_}là FŽÃmno!dŸb'†ƒobk†~o¦_ax`!d

½¾hf_I_aÆIm‰o¦´¦x~½ea¶‰x`!lÄÁy†ƒbf_IhN½¾x~h™bk_aY»‹_Ihk†ƒbf`¦hk_Id¨ ‹_I¶‰x‘ÁÛ"‚*)~ËjÌÄs™ð¦x~h g¦_jbk_ahkx~´~_Io¦_ax`!d?oF`!ej¶n_I†ƒbfmnx~oëpÁî_"dfm‰Y»¦¶nŽv†dfdf`¦Y_

bfg!†ƒb+¢ ξÇ<Ñ3mnea_y½¾x~hkÄmno¦´ŒoF`!ej¶n_amcÁ m‰bfgY†Œº~_ahkŽ¦d9†~¶‰¶cmno¦m‰bfm]†ƒ¶c9†~dkdN†ƒhk_y†eDbfmnº‘†‘bk_Il[Á m‰bfg!m‰o9† bfmnÄ_ dbf_I»Ä†~o!l[bkg!†‘bNbkg¦_

Ü

ç

bf_IÄ»r_ah'†‘bk`¦hk_ mnd ‹_I¶‰x‘ÁφèoF`!ej¶n_I†ƒbfmnx~oÄbfg¦hk_Idfg¦x¶nl6ÎÐdf_jb bfx9Û,)ƒË}̦ÑDs eaejxhkl¦m‰o¦´Œbkx[hf_}d`!¶²b'dN½¾hfxö!_I¶nlY_azc»‹_IhfmnY_aobkdIp

Áî_¦hk_I¼`¦mnhk_<Á2†‘bk_ah"dk†‘bf`!hk†ƒbfmnx~oM½¾x~h2bfg!_¦x~o!df_jb"x~½Nea¶‰x`!lGm]ej_è½¾xhf9†ƒbfmnx~o6†ƒ rx‘º~_¦†vbf_aY»r_ah'†‘bk`¦hf_èbkg¦hk_Idfg¦x~¶]lGÇ Î—d_ab

þ

bfxÓÛ ~Ëj̦ÑDsÄð¦xh

ç

¶n¶¦xƒbkg¦_ah™eax~oFº~_Ihkdfm‰xo[h'†‘bf_}dN†ƒhk_î»!†~hk†~Y_jbf_IhfmnÆa_}l¦mnoY†Œdfm‰Ymn¶n†~h?Á2†0Ž[†~dNmnoYdŸb'†ƒo!l¦†~hklvej¶nx~`!lvYmneahfx»¦gFŽcdm]ead?dfe'g!_aY_IdIs

Ÿej_2ejhkŽcdŸb'†ƒ¶]d3Á g¦m]e'gÂoF`!ea¶‰_}†‘bk_îmnov†<Áy†ƒbf_Ih?dk†‘bk`¦hk†ƒbf_}l¦_aoFºFm‰hkx~o!Ä_Iob$Á mn¶‰¶c´hfx‘ÁÓº~_IhfŽ¦¼`¦m]e'ïF¶‰Žè FŽ¦º‘†ƒ»rx~`¦h™lc_I»‹xdm‰bfmnx~o

lc`¦_ÂbfxM†Gg¦mn´~g¿m]ej_Äd`¦»r_ah'dk†‘bf`!hk†ƒbfmnx~o- †~eaeax~h'lcmno¦´YbfxΟ{}Ñjs<^<_a»r_ao!l¦m‰o¦´‡x~o·¶‰xcea†~¶ëbkg¦_ahkYxFl¦Žo‹†ƒYmnevejxo!lcm‰bfmnx~o!dIp

Ü bfg¦_y_jzcm]dŸbkm‰o!´

ea†d_2m]dN_jzc»‹_}eDbk_Ilv½¾x~hNÁy_I†ƒïvlcŽFo!†ƒYm]ea†ƒ¶½¾xhkeam‰o¦´dap~Á g¦_ahk_ bkg¦_ mno¦m²bkmn†~¶¦Ym²zc_}lFµ¹»!g!†~df_yea¶‰x`!lvÁ m‰¶n¶¦h'†ƒ»¦m]lc¶nŽ[´~¶]†~eamn†ƒbf_îbfx

‹_}ejxÄ_"†~oYmnea_"ej¶nx~`!lÄ_jzcm]dŸbkm‰o!´[†‘b¨x~h™o!_I†ƒh mnea_"df†ƒbf`¦h'†‘bkm‰xoës$Í2g¦_"df_Ieax~o!lYeI†~df_ym]d™_azF»r_Iejbf_}lĽ¾x~h¨dbfhkx~o¦´¦lcŽFo!†~Äm]ea†~¶

½¾x~h'ejmno¦´d<†ƒ¶nx~o¦´G½¾hfxobkd"xh

ea†~oA ‹_‡9†ƒmnobk†ƒmno¦_}lÞsÍ2gF`!dIpNbfg!_M¶nmn¼`¦m]lT½¾h'†~ejbfmnx~oAÁ mn¶n¶î»¦gFŽcdfmneI†ƒ¶n¶‰ŽT†~l‘óŸ`!dŸbY†ƒo!lX†ƒoA_IY»¦m‰hkm]ea†ƒ¶2df»‹_}ejm‰ö‹ea†ƒbfmnx~oAm]d

o¦xƒb

òF_a_}lc_ah"ð!_a_}lc_ah"»!hfxcej_}dfd2†~hf_èhk_a»¦hk_Idf_aobf_}lÃ_azF»!¶‰m]ejm‰bf¶nŽM†~o!lM†l¦lcm‰bfmnx~o!†~¶Þ»!†ƒh'†ƒY_abf_ahkmnÆI†‘bkm‰xo!d2†ƒhk_Œ†0ºx~m]lc_Ilës

ç

ö!h'db[bf_Iob'†‘bkm‰º_Gº‘†ƒ¶nmnl!†‘bfmnx~oTÁ m‰bfgAbfg¦_M´~¶nx~ !†~¶¨YxFl¦_a¶ Šèˆ¿éPhk`¦o¦o¦mno¦´¿mno ej¶nm‰9†‘bk_ YxFl¦_Ãhf_Iº~_}†ƒ¶n_Il¤bfg!†ƒbÂbkg¦_

m‰o‹ej¶n`!dmnx~o‡xƒ½™ej¶nx~`‹lGm]ej_è´~hk_I†ƒbf¶nŽ9m‰Y»¦hkx‘º~_}dîbfg¦_èx~`¦bf´~xm‰o!´v¶nx~o¦´Áy†0º_ h'†~lcm]†‘bkm‰xoMlc`!_èbfx9lcm‰„r_Ihf_Iob"h'†~l¦mn†ƒbfmnº~_Œ»¦hfx»cµ

_ahfbfmn_Idîxƒ½3ea¶‰x`!l9m]ej_

g!†~dè r_a_Io¤_}dŸb'†ƒ ¦¶nm]dg¦_}lÞsYð$mnh'dŸb¦º~_Ihfm‰ö‹eI†‘bfmnx~o¤hf_}d`!¶²b'dŒdfg¦x‘Áædf¶nm‰´gbk¶‰Ž·m‰Y»¦hkx‘º~_}l·dkejx~hk_Id ½¾x~h ƒÄµ—bk_aY»‹_Ihk†ƒbf`¦hk_Y†ƒo!l

ƒ||Mg‹q™†6´~_ax»‹x~bf_Iobkmn†~¶™g¦_amn´~gbÂeax~hkhf_I¶n†ƒbfmnx~oëp?†ƒo!lӆ6 r_jbbk_ahvhf_I»¦hf_}d_Iobk†‘bkm‰xo¤x~½îbkg¦_Ã`¦»!»‹_Ihµ¯¶‰_Iº~_I¶¨gF`¦Ymnlcm‰bŸŽÓ†ƒo!l

ç

ej¶nx~`!l¤dŸbkhf`‹eDbf`!hf_}das d¦†ƒo¤_jz¦†ƒY»¦¶n_~pëð?mn´~`¦hk_M{YejxY»!†ƒhk_Id"bfg¦_Yhk_a¶]†‘bfmnº~_ÄgF`¦Ymnl¦m²bŸŽ.x‘º_ahŒÁy†ƒbf_ah膑b ƒ|G†~o!l¤~|~|

ç

g!q™†¦hk_Idf`¦¶‰bfmno¦´è½¾hfx ƒƒµ¯gGŠèˆ¿éThk`¦o!d Á m²bkg9bfg¦_

¶‰_Iº~_I¶—pÞbkg¦_9hkx~`cbkm‰o¦_Ãdke'g¦_IÄ_Yhk_Idf`¦¶‰bkdèmno¤`¦o!hf_}†ƒ¶nmndbfm]eÄg¦m‰´gÓg`!Äm]lcm‰bfmn_Id[ea¶‰xd_vbfx.Á2†‘bk_ah¦dk†‘bf`!hk†ƒbfmnx~o·mnoÓ_I¼`!†ƒbfx~hkm]†ƒ¶

†ƒo!lӍYm]lFµ¯¶n†ƒbfm‰bf`!lc_Mhf_I´~mnx~o!d†~dkdxcejm]†‘bk_IlÓÁ m‰bfgTbkhfx»¦m]ea†ƒ¶îejx~oFº_IeDbkm‰xoT†ƒo!lÓ½¾hkx~ob'†ƒ¶ dfŽcdŸbk_a9dapNhk_Idf»‹_}eDbkm‰º_a¶nŽ~s.i)m‰bfg

bfg¦_Mej¶nx~`!lFµ¯m]ej_Ãdke'g¦_aY_~p3bfg¦_Ghf_I¶n†ƒbfmnº~_9gF`¦Ymnl¦m²bŸŽ¤mndvdmn´~o¦m‰ö‹eI†ƒobf¶nŽ¿hk_Ilc`!ea_IlmnoÓbfg¦_}d_Ãhk_a´m‰xo!d<ù¤lcx‘Á obfx.º‘†~¶‰`¦_}d

†ƒhkx~`¦o‹l€|‘µ¥)‘|¢¡Ãp$Á g¦m]e'gAm‰o‹lcmneI†‘bk_Ãmnea_jµKdf†ƒbf`¦h'†‘bkm‰xoës¿t

ç

½¾hfxobk†ƒ¶ej¶nx~`‹l¦dyÄm‰zc_IlFµ¯»¦g!†d_膑b2Á2†‘bf_Ihydk†‘bk`¦h'†‘bfmnx~oM†ƒhk_

lcm]dŸbkhfmn ¦`cbkm‰xo¤x~½ hf_I¶n†ƒbfmnº~_ÄgF`¦Ymnlcm‰bŸŽ.½¾hkx~ bkg¦_ÄbŸÁîxM½¾x~hk_IeI†~dbkd

hf_I´~mnx~o!dIpFÁ g¦_ahk_¦ej¶nx~`!lcµ¹m]ej_¦dke'g¦_aY_¦†ƒ´†ƒmnoMhk_Idf`¦¶‰bkd2mno6†Ylch'†~dbfm]eŒhf_}lc`!ejbfmnx~o‡xƒ½$hk_a¶]†‘bkm‰º_èg`!Äm]lcm‰bŸŽ~s

ŠŒm‰º_ao†Gdf`!eaea_IdkdŸ½¾`¦¶™eax~Y»¦¶n_jbfmnx~o·xƒ½¨bkg¦_Âbk_Idb[d`¦m‰bf_p놃o¿x»‹_Ihk†ƒbfmnx~o!†~¶3mnobfhkxFl¦`!eDbkm‰xo·xƒ½ bfg¦_Yej¶nx~`!lcµ¹m]ej_Ädfe'g¦_IY_vm]d

ç

dfe'g!_Ilc`¦¶n_Il¤½¾xhˆ¿†0Ž ~|~|‚¦s b[bkg¦_Gdk†ƒY_YbfmnY_~p$bfg¦_Ão!_aÁ dfe'g!_aY_9Á m‰¶n¶ †~¶ndfx6 ‹_GdfÁ m‰bke'g¦_}lÓx~oÓm‰obkg¦_Ãhk_a´~mnx~o‹†ƒ¶

礣 ç £

YxFl¦_a¶ ì$ˆ †~dèÁy_a¶n¶ †~dèmnoAì3ˆ¿d¦†ƒbˆ._jbk_axFòFÁ mndkdÃιòFÁ m²bkÆa_Ihf¶]†ƒo!l‹ÑDp q µKòcˆ Σ¯bk†~¶‰Ž¦ÑDpNÀ ¸<ˆ·òÏNJŒhk_a_}ej_0Ñ膃o!l

£

Ÿˆ·ŠŒi ÎÐqNx~¶]†ƒo‹l!ÑDp¦†~o!l‡m‰o‡bf_Io.À ˆ¿d2hk`¦o¦o!m‰o¦´9Áyx~hk¶]lÃÁ m]lc_~s

ð?m‰´`¦hf_M{~õ¦¥

›kª‰­jŸ±«¬0›Œžƒš¤¦±²¢~±«¨§Y­j¤© žÞ­÷ª«›‰î­aœF¢   vž¢Þ­] Ÿ±«˜}ž‘n£I 2Ÿž›'£ c›f ­aŸ±«£}œ­Iª'ª«£}š¢Ã¥ž›k¤¦›]Ÿ£î­aœF¢

Ÿž›2œ›k¨ 'ª«£IšF¢! б"'› ¥ž›'¤¦›#÷³c£IKŸ£}¤$f¡%©'& О(¦)+*,]£I Ÿ›D­I¥¯Ÿ¥î¥¯­a KŸ±«œ˜#‰ Ÿ£}¤-© /.›'³ Ÿš­a §0©  1023©547698î¡ The SNOWMIP project : validation of albedo and short wave radiation budget simulated by several snow models.

P. Etchevers1, E. Martin1, R. Brown2, C. Fierz3,Y. Lejeune1E. Bazile4, A. Boone4, Y.-J. Dai5, R. Essery6, A. Fernandez7 , Y. Gusev8, R. Jordan9, V. Koren10 , E. Kowalczyk11, N. O. Nasonova8, R. D. Pyles12, A. Schlosser13, A. B.Shmakin14, T. G. Smirnova15, U. Strasser16, D. Verseghy2, T. Yamazaki17, Z.-L. Yang18

1Centre d'Etudes de la Neige, CNRM /Météo-France, Grenoble, France 2Canadian Meteorological Service, Dorval, Qc, Canada 3Swiss Federal Institute for Snow and Avalanche Research (SLF), Flüelastrasse 11, CH- 7260 Davos Dorf, Switzerland 4Météo-France, CNRM, Toulouse, France, 5Institute of Atmospheric Physics; Chinese Academy of Sciences, Beijing, China, 6Met Office, Bracknell, Berks. U.K., 7Instituto Nacional de Meteorologia, Madrid, Spain, 8Laboratory of Soil Water Physics, Institute of Water Problems, Russian Academy of Sciences, Moscow, Russia, 9CRREL, Cold Regions Research and Engineering Laboratory,Hanover, U.S.A., 10NOAA/NWS/OH1/HRL, Silver Spring, MD 20910 U.S.A., 11CSIRO Atmospheric Research, Aspendale, Australia, 12Cooperative Institute for Research in the Environmental Sciences, Boulder, U.S.A., 13COLA/IGES, Calverton, U.S.A., 14Laboratory of Climatology, Institute of Geography, Russian Academy of Sciences, Moscow, Russia, 15Forecast Systems Laboratory, Boulder, USA, 16Institute of Hydromechanics and Water Resources Management; ETH-Hönggerberg, Zürich, Switzerland, 17Frontier Observational Research System for Global Change, Tohoku University, Sendai, Japan, 18Dept of Hydrology and Water Resources, The University of Arizona, Tucson, U.S.A.

Email : [email protected] Introduction Over the last thirty years, many snow models have been developed and have been used for various applications such as hydrology, global circulation models, snow monitoring, snow physics and avalanche forecasting. The degree of complexity of these models is highly variable, from simple index methods to multi-layer models simulating snow cover stratigraphy and texture.

In the project SnowMIP ( www.cnrm.meteo.fr/snowmip/ ), four sites were chosen for the comparison of the snow models. As albedo and snow surface temperature measurement are not available for all of them, only the Col de Porte (CDP) and Weissfluhjoch (WFJ) data are used in this paper. The Col de Porte is a middle elevation site located in the French Alps (1340 m). Weissfluhjoch is a more mountainous site that lies at an altitude of 2500 m in the Swiss Alps.

Albedo parametrization The incoming short wave radiation is the same for all the models and the short wave radiation budget simulation depends on the fraction of the radiation which is reflected by the snowpack, i.e. the snow albedo. The albedo parametrizations of the 23 models are based on temperature (6 models), snow types and/or grain size (6 models), or age (13 models). Four models use either no albedo or a fixed albedo (index based, or a constant albedo, or an albedo depending only on vegetation fraction or shading). Some models use two parameterisations, e.g. age can account for all the aging processes of snow or can be used in conjunction with a parameterisation based on snow grain size or type.

The albedo increases are generally determined by snowfalls. The albedo decreases are more complex because they depend on the snow micro-structure and grain types. As stated above, the decrease of albedo is generally calculated by the models as a function of the snow age, surface temperature, grain size and other parameters provided by the model itself. These parametrizations play a major role in the model performances because albedo is a key factor for calculating the snowmelt. Thus, it appears interesting to examine the accuracy of parametrizations for some particular periods. These selected periods cover at least eight days and do not include snow fall events (table 1). The three CDP periods correspond to snowmelt events and the albedo decrease (-1.63×10-2 per day on average) is due to the appearance of liquid water in the snowpack. At WFJ, the decrease is 5 times weaker (-3.17×10-3 per day on average) because it is due to dry snow evolution (without melting or rain). The quality of each simulation is estimated by comparing the change in observed and simulated albedos between the beginning and end of each period. The albedo decrease averaged for all models is pretty accurate for all the episodes (table 1), but the extreme values show that some models are far from the reality. For the 6 episodes, the rms error of the albedo variation ranges from 0.07 to 0.13, but average results are very different for the two sites.

Site Period Number Observed Observed Simulated Simulated of albedo albedo variation albedo albedo days (beginning and (per day) variations: variations: end) average (per min-max (per day) day) Episode 1 CDP9697 28/02/97- 15 0.63-0.5 -0.0087 -0.01 -0.02/0.005 14/03/97 Episode 2 CDP9798 24/01/98- 28 0.77-0.59 -0.0064 -0.006 -0.01/0.001 20/02/98 Episode 3 CDP9798 19/04/98- 8 0.8-0.53 -0.0338 -0.015 -0.03/0.00 26/04/98 Episode 4 WFJ 13/12/92- 21 0.9-0.86 -0.0019 -0.005 -0.01/0.005 02/01/93 Episode 5 WFJ 29/01/03- 17 0.91-0.8 -0.0065 -0.008 -0.02/0.00 14/02/93 Episode 6 WFJ 21/04/93- 9 0.75-0.74 -0.0011 -0.009 -0.03/0.009 29/04/93 Table 1 : The six periods selected to validate albedo decreases. No precipitation occurred during these episodes. The two last columns contain the average for all models and the minimum/maximum values of the simulated albedo variations.

Particular episodes The rms error is generally larger for the CDP site (0.08 to 0.16) than for the WFJ site (0.03 to 0.11). The performance of a given model depends a lot on its albedo parametrization. For the CDP site, the best models use parametrizations based on the snow age and/or the snow type. As the albedo regularly decreases during the episodes, the age parametrization is adequate to correctly simulate this evolution. This is illustrated by figure 1, where the daily simulated and observed albedos are plotted for episode 3 (CDP9798): many models properly reproduce the albedo decrease from 0.8 to 0.65. Most of these models use a parametrization based on the snow grain size or the snow age. On the contrary, the less accurate models generally include a parametrization using the snow surface temperature. As this parameter does not vary a lot during the melting period, this is not an accurate predictor for the albedo decrease. Finally, if one considers the whole set of episodes, three of the four best models use albedo parametrizations based on snow grain characteristics and snow age. They are able to simulate both the slow albedo decrease in winter and the fast decrease during the melting period. The other models are generally able to simulate properly the albedo for one type of episode but not the other. Only one model using a parametrization based only on the snow age and surface temperature succeeds in simulating the different albedo evolutions.

REF_CDP_9798 : daily albedo ACA_CDP_9798_ES_REF CLA_CDP_9798_ES_REF CLD_CDP_9798_ES_REF 100 COL_CDP_9798_ES_REF CRO_CDP_9798_PF_REF CSI_CDP_9798_ES_REF ESC_CDP_9798_PF_REF IAP_CDP_9798_ES_REF 90 INM_CDP_9798_PF_REF ISB_CDP_9798_ES_REF ISF_CDP_9798_ES_REF ISO_CDP_9798_ES_REF MAP_CDP_9798_ES_REF

) 80 MAT_CDP_9798_ES_REF NOH_CDP_9798_ES_REF (% S17_CDP_9798_PF_REF do SPO_CDP_9798_ES_REF

be SWA_CDP_9798_ES_REF Al 70 SNO_CDP_9798_ES_REF SNT_CDP_9798_ES_REF SNT_CDP_9798_PF_REF TS1_CDP_9798_PF_REF TSM_CDP_9798_ES_REF 60 TSM_CDP_9798_PF_REF UKM_CDP_9798_PF_REF UKM_CDP_9798_ES_REF VIS_CDP_9798_ES_REF Observations 50 0 20 16 12 8 4 0 20 16 19 20 21 22 23 24 25 26 Figure 1 : daily albedo observed (large black line) for episode 3 (CDP site, from 19 to 25 April 1998) The other coloured lines correspond to the 26 albedo simulations.

Development of a New Land-surface Model for JMA-GSM Masayuki Hirai Numerical Prediction Division, Japan Meteorological Agency (e-mail:[email protected]) Mitsuo Oh’izumi Meteorological Research Institute, Japan Meteorological Agency (e-mail: [email protected])

In 1989, a SiB (Simple Biosphere) model was developed as a land-surface scheme for JMA's operational global NWP model (JMA-GSM). However, the SiB model has not been modified substantially after that. Recently, problems considered to be relevant to the land-surface process were pointed out as follows: * Snow-melting is overestimated in spring. In the Northern Hemisphere, the southern edge of the snow cover area retreats faster comparing with an analysis or observations. * Warming bias in the lower troposphere on ice sheet area in summer.

In order to solve such problems, we have developed a new land-surface model, which treats soil and snow processes more accurately. Major changes are as follows: * A conventional force restore method for predicting soil temperature is abandoned and heat conductivity among multiplied soil layers is explicitly calculated. * Phase change of soil water is considered. * Multiplied snow layers are introduced and phase change of snow water is considered. * Snow cover is classified into two categories, “partial snow cover” and “full snow cover”. * A sophisticated snow process is introduced. (Aging of snow albedo, temporal changes of snow density and heat conductivity, keeping water in snow layers and so on.) We carried out preliminary forecast experiments for April 2002 with JMA-GSM T213L40. Initial condition of soil water for the experiments is set to a model-climate state obtained from a long-term integration of the model, while the operational SiB uses a climate state by Willmott et al. (1985). The snow cover area for 72 hour forecast (FT72) initiated at 12 UTC April 15, 2002 (validtime is 12UTC April 18, 2002) is shown in Fig. 1. A snow cover area at the Tibetan Plateau is well predicted by the new land-surface model, while not by the operational one. Comparing with the snow cover area estimated from the brightness temperature of SSM/I (See Fig. 2.), the new land-surface model simulates better than the operational one in this region. Time series of forecast results of sensible and latent heat fluxes and snow depth at a point in the Tibetan Plateau (35N, 90E) are shown in Fig. 3. The new land-surface model well simulates that sensible heat flux decreases reasonably after snow cover forms on the day 2. Instead, latent heat flux increases because of snow sublimation. Net short-wave radiation decreases about 150 W/m2 and net long-wave radiation increases due to snow cover (not shown). As a result, it is shown that the new land-surface model properly represents snow process. However, southern edge of the snow cover area still retreats faster than an analysis or observations. Moreover, RMSE (Root Mean Square Error) of the temperature at 850 hPa in the Northern Hemisphere is not improved and a distinct negative bias appears by the new land-surface model. We go on with development of the land-surface model, planing to put it into operation within a few years. References JAPAN METEOROLOGICAL AGENCY, 2002: Land-surface processes, OUTLINE OF THE OPERATIONAL NUMERICAL WEATHER PREDICTION AT THE JAPAN METEOROLOGICAL AGENCY, 50-52. Willmott, C. J., C. M. Rowe, and Y. Mintz, 1985: Climatology of the terrestrial seasonal water cycle. J. Climatology, 5, 589-606.

Fig 1. The snow cover area at 72 hour forecast (FT72). Initial time is 12UTC April 15, 2002 (validtime is 12UTC April 18, 2002). The left figure shows a result from the new land-surface model, and right one shows that from the operational one.

Fig 2. The snow cover area for April 18, 2002 estimated from the brightness temperature of SSM/I.

New SiB (ini 20020415 35N90E) Routine SiB (ini 20020415 35N90E) SH LH Snow Depth SH LH Snow Depth 300 0.10 300 0.10

] 200 0.08 ] 200 0.08 2 2

100 0.06 100 0.06

0 0.04 0 0.04 F lux [W/m F lux [W/m

-100 0.02 Snow Depth [m] -100 0.02 Snow Depth [m]

-200 0.00 -200 0.00 0 6 12 18 24 30 36 42 48 54 60 66 72 FT [hour] 0 6 12 18 24 30 36 42 48 54 60 66 72 FT [hour] Fig 3. The time series of sensible and latent heat fluxes and snow depth at a point in the Tibetan Plateau (35N, 90E). The result from the new land-surface model is shown in left, and the operational model is in right.

Application of RRTM longwave radiation to NCEP GFS Y.-T. Hou1, S. Moorthi, K. Campana, NCEP and M. Iacono, AER

The longwave (LW) radiative transfer parameterization used in the current NCEP operational Global Forecast System (GFS) atmospheric model was developed at the Geophysical Fluid Dynamics Laboratory (GFDL) (Schwarzkopf and Fels, 1991). Recently, a Rapid Radiative Transfer Model for longwave radiation (RRTM) has been developed at Atmospheric and Environmental Research, Inc. (AER) (Mlawer et al. 1997). The radiative transfer calculation in RRTM is highly accurate with very high computational efficiency, especially in a large model with high vertical resolution. In a 64-layer model test, the RRTM runs in about one half the time as the operational code does. In addition to the major atmospheric absorbing gases included in the operational LW scheme, such as carbon dioxide, water vapor, and ozone, the RRTM also contains numerous minor species such as nitrous oxide, methane, and up to four types of halocarbons. In water vapor continuum absorption calculations, the operational model uses coefficients derived by Roberts et al. (1976), while RRTM uses a more advanced CKD_2.4 model (Clough et al. 1992).

Comparisons of clear-sky cooling rates between the two parameterizations for typical atmospheric profiles show that, in general, differences between the two radiation schemes are relatively small, except in the upper stratosphere, where RRTM produces more cooling effect than the operational LW model. Since December 24, 2002 we have been running a parallel T254-L64 GFS experiment with RRTM. Figure 1 shows comparison of 500 mb geopotential height anomaly correlation scores of 5-day model forecasts between operational model (PP1) and the parallel model (PP2). Zonally averaged 5-day forecast temperature biases (not shown) indicate that RRTM improves tropospheric temperature bias while enhancing the cold bias in the stratosphere. Two five-year climate runs on a T62-L64 model were also conducted for the two LW radiation schemes. Table 1 shows outgoing LW radiation at the top of the model atmosphere (TOA) for the two LW radiation parameterizations and the five-year climatological average from ERBE observations. It shows that the RRTM produces a closer agreement between simulations and observations than the operational model. Comparison of OLR fluxes at the TOA among the two radiation parameterizations and ERBE observation for a five-year average over the months of July and January (not shown) shows that RRTM improves OLR distribution over tropical and summer mid-latitude regions.

Table 1. Five-year averaged monthly OLR (W/m**2) comparisons among the operational LW scheme, the RRTM, and the observations from ERBE data.

Month Operational-LW RRTM-LW ERBE ______

JAN 241.06 235.90 232.48 APR 242.64 237.45 234.47 JUL 247.89 242.96 239.40 OCT 241.06 235.90 235.30 ______

1E-Mail address: [email protected]

References: Clough, S.A., M.J. Iacono, and J.-L.Moncet, 1992, J. Geophys. Res., 97, 15761-15785. Mlawer, E.J., S.J. Taubman, P.D. Brown, M.J. Iacono, and S.A. Clough, 1997, J. Geophys. Res., 102, 16663-16682. Roberts, R.E., J.E.A. Selby, and L.M. Biberman, 1976, Appl. Opt., 15, 2085-2090. Schwarzkopf, M.D., and S.B. Fels, 1991,. J. Geophys. Res., 96, 9075-9096.

Impact of a Cloud Ice Fall Scheme Based on an Analytically Integrated Solution

Hideaki Kawai Climate Prediction Division, Japan Meteorological Agency (e-mail: [email protected])

1. Background Cloud Ice Fall in global NWP model or climate model is very bothersome to treat because the product of the fall speed (vice ) and the time step (Dt ) often exceeds the thickness of a vertical layer (Dz ) of the model. Since explicit time integration cannot be used practically, the process in our Global Spectral Model (GSM) has been treated as follows:

l In cases viceDt £ Dz , cloud ice content (qice ) in a layer is distributed into the layer itself and a layer just below the original layer exactly.

l In cases Dz < vice Dt £ 2Dz , a part of cloud ice content is distributed into only a layer just below the original layer exactly and the other part of it falls through into ground surface within a time step.

l In cases vice Dt > 2Dz , all the cloud ice content falls into ground surface within a time step.

The treatment works appropriately as far as viceDt is smaller than Dz . Time step in T213 Euler model, which is our operational short-range to 1-week forecast model, meets this condition in almost all cases. But in T106, which is our 1-month forecast model, and in T63, with which we will start 3-month dynamical forecast from March 2003, the condition is not satisfied and a large part of cloud ice falls into ground surface within one time step. Moreover, long time step will become a serious problem for cloud ice fall process in T213 after an adoption of semi-Lagrangian time integration scheme in T213 (TL319) scheduled in 2003.

2. Scheme based on Analytically Integrated Solution To avoid the above problem, time integration for cloud ice fall is implemented following analytically integrated solutions below (Rotstayn 1997, Jakob 2000). C q (t + Dt) = q (t)e -DDt + (1 - e -DDt ) ice ice D R a v a f = <100 ice + >100 C = + C gnrt , D rDz Dz Dt ( ) where Rf is the cloud ice flux from the above layer, r is the air density, a <100 a>100 is the ratio of cloud ice particles whose sizes are smaller (larger) than 100 micrometer. Cloud ice content is separated into two categories following Jakob (2000) and McFarquhar and Heymsfield (1997). Since GSM adopts the cloud distribution concept by Sommeria and Deadorff (1977), cloud water content has been distributed into cloud water (ice) and water vapor at first step (to produce * ) and then, cloud ice fall process has been q ice (t + D t ) calculated at second step (to produce ). But this two-step procedure leads to an q ice (t + Dt ) unreasonable overestimation of amount of cloud ice fall and consequent lack of cloud ice * content. Therefore cloud ice generation rate Cgnrt( ) is = (q ice (t + Dt ) - q ice (t )) / Dt introduced in order to treat generation and fall of cloud ice simultaneously. This procedure is more correct theoretically and avoids abrupt change of cloud ice content.

3. Result The results of one-month integration using T63 for December 1988 are shown. Figure 1 shows zonal means of cloud water content (including both the liquid and solid state) of analytical and original schemes. Cloud water content is considerably increased at an altitude where the state is solid. The increase is preferable because cloud water content of analytical scheme is closer to equilibrium one ( * ) provided by the cloud q ice (t + Dt ) distribution scheme. Fig.2 shows an impact and a current scheme error from Earth Radiation Budget Experiment (ERBE) data on outgoing longwave radiation (OLR). Analytical solution scheme reduces the positive error on OLR to some extent, for example, over Amazon. It has also been confirmed that new scheme improves the overestimation on net downward shortwave radiation at the top of the atmosphere a little bit (not shown). The remaining biases are probably resolved by sophisticated treatments of radiation processes (Kitagawa and Yabu 2002) and by increasing cloud amount based on a review of a probability distribution function in the cloud distribution scheme.

Fig. 1: Zonal means of cloud water content in unit of 10-6[kg/kg] for December 1987 calculated by (a) analytical solution scheme and (b) original scheme. Vertical axis shows pressure [hPa].

Fig. 2: Outgoing longwave radiation [W/m2] (a) impact and (b) error for December 1987. Error is calculated based on ERBE observation data.

References Jakob, C., 2000: The representation of cloud cover in Atmospheric General Circulation Models, ECMWF. Kitagawa, H., and S. Yabu, 2002: Impact of a revised parameterization of cloud radiative forcings on earth's radiation budgets simulation. Research Activities in Atmospheric and Oceanic Modelling, 32, 04-15–04-16. McFarquhar, G. M., and A. J. Heymsfield, 1997: Parameterization of tropical cirrus ice crystal size distribution and implications for radiative transfer: Results from CEPEX. J. Atmos. Sci., 54, 2187–2200. Rotstayn, L. D., 1997: A physically based scheme for the treatment of clouds and precipitation in large- scale models. I: Description and evaluation of the microphysical processes. J. Roy. Meteor. Soc., 123, 1227–1282. Sommeria, G., and J. W. Deardorff, 1977: Subgrid-scale condensation in models of non-precipitating clouds. J. Atmos. Sci., 34, 345–355.

Numerical Simulations of a Tropical Cyclone using the Arakawa-Schubert Scheme with Vertically Variable Entrainment Rate

Akihiko MURATA and Mitsuru UENO Meteorological Research Institute / Japan Meteorological Agency, Tsukuba, Ibaraki 305-0052, Japan (corresponding author: [email protected])

1. Introduction Numerical simulations for tropical cyclones provide an extremely useful test for the depth of knowledge about cumulus parameterization (Ueno, 2000). We therefore have investigated effects of some cumulus parameterizations on tropical cyclone simulations. As a consequence it was found that the Arakawa-Schubert scheme guaranteed relatively good performance (e.g., Murata and Ueno, 2000). It has been pointed out, however, that fractional entrainment profiles in the scheme did not agree with those of observations and numerical experiments. Lin (1999) investigated entrainment profiles using simulated data from a slab-symmetric cloud-resolving model. He found that entrainment for each cloud type, which was categorized in terms of cloud top height, tended to be significant near cloud base and cloud top. In this study, on the basis of the results from the three-dimensional cloud-resolving model (CRM), vertically variable entrainment rate is applied to a version of the Arakawa-Schubert scheme. The impact of the modified scheme on the simulations of typhoon Saomai (2000) is investigated.

2. Estimate of entrainment rate with a cloud-resolving model The Meteorological Research Institute / Numerical Prediction Division nonhydrostatic model (MRI/NPD-NHM; Saito et al., 2001) with 200 m horizontal resolution is used as CRM. The model includes bulk cloud microphysics by Ikawa et al. (1991), which predicts the mixing ratios of six water species (water vapor, cloud water, rain, cloud ice, snow and graupel) and the number concentration of cloud ice. A domain (280×280) is set on a spiral rainband of Saomai. A 20-min simulation is conducted by using a one-way nest with output from the 1-km grid MRI/NPD-NHM. Before the calculation of entrainment rate, a cumulus area, where a single cumulus is contained, is extracted from the CRM output. The calculation is conducted on the basis of the definition, dM/Mdz, where M is mass flux averaged over cumulus grids and z is height. The cumulus grids are defined as those at which the sum of mixing ratios as to cloud water and cloud ice is larger than 0.1 g/kg and vertical velocity is positive. The result of the calculation clearly shows high entrainment rate just above the cloud base and just below the cloud top. Figure 1 displays the characteristic entrainment profiles that have larger amounts between 1.0 and 1.5 km high and between 6.5 and 7.0 km high. The layers in between are marked by relatively low entrainment rate.

3. Experimental design MRI/NPD-NHM with 20-km horizontal grid length is used in the simulations of Saomai. We incorporate a version of the Arakawa-Schubert (AS) scheme (Kuma et. al. 1993), which has the prognostic equation that predicts the cloud base mass flux and has a downdraft and a mid-level convection, into MRI/NPD-NHM. A profile of entrainment rate, λ, in the AS scheme is modified so that they can qualitatively represent the feature revealed in the CRM result. The assumed profile is as follows:

2∆ π zT + zB λ = A{1− sin( z + − ∆)}+ S zT − zB 2 zT − zB A > 0, S > 0, ∆ < π / 2 -3 where zB and zT are cloud base and top height, respectively. Among unknown constants, ⊿ is set at 5×10 . The remaining constants A and S are determined so that following two conditions are satisfied. One of them is that the cloud top level is located where buoyancy in the cloud disappears. The other condition is that the entrainment rate at the cloud base, λB, is determined on the basis of the CRM result, which indicates that λB falls within the range between the order of 10-3 to 10-2 m-1. As for the AS scheme, λ ranges from the order of -4 -3 -1 10 to 10 m . λB therefore is set at λB=kλ, where k is constant that controls the degree of the cloud-base entrainment rate. With k varying among 2, 3, 5 and 10 (referred as the experiments of AS2, AS3, AS5 and AS10, respectively), 36-hr integrations are conducted to examine the sensitivity of k.

4. Results of numerical experiments Difference in the entrainment rate profile has drastic influence on simulated precipitation patterns. The axisymmetric eyewall is reproduced in AS5, whereas more scattered feature is shown in AS (Fig.2). The precipitation in the former tends to lie in the location relatively near the storm center. The radial profiles of accumulated rainfall amount are shown in Fig.3. From the figure, it is found that the rainfall amount over the outer area (outside the 200 km radius) decreases with increases in the entrainment rate (i.e., increases in k). The radial distribution of precipitation seems to play a role in the size of the simulated tropical cyclone. The radial profiles of the axisymmetric tangential wind velocity change (Fig.4) clearly shows that the quantities over

the outer+ area decrease with increasing+ the cloud-base entrainment rate, which is the same aspect as the precipitation. The result indicates that the storm grows as the entrainment rate drops. The less rainfall over the outer area in the larger entrainment case is accounted for by relatively little mass flux calculated by the modified scheme. The vertical profiles of the cloud mass flux, averaged over the area ranging from 200 to 500 km radii, display that the mass flux is weaker in the larger entrainment cases (Fig.5), suggesting that vertical static instability in the outer area is not eliminated so much as those of the smaller entrainment cases.

8 AS AS5 PSEA 1 * 1 X-Y Z= 0.02 KM 1440 MIN PSEA 1 * 1 X-Y Z= 0.02 KM 144 0 MIN 7 1000 30N 30N 995 1200. 1200. 1000 6 995

] 1100. 1100. m 5 k [ t 4 1000. 940 1000. h 945 g i 950 900. 955 900. 940 3 960 945 He 965 950 970 955960 800. 975 800. 965 2 970 980 975 985 980 1 700. 700. 990 25N 985 25N 995 995 0 600. 600. 990 -0 .01 0 0.01 0.02 0.03 1000 0.04 500. 1000 500. 1000 -1 En 800.0trainme1000.0nt rate1200.0 [m ] 1400.0 800.0 1000.0 1200.0 1400.0 125E 130E 125E 130E Fig .1 Vertical profiles of entrainment rate Fig.2 Horizontal distributions of 1-hr accumulated rainfall calcu lated by the CRM output. Lines are amount at 24 hr. Contours are drawn at 1, 5, 10 and 15 mm/hr. drawn every 5 sec from 6 min to 6 min 30 sec.

200 4 20 18 ] e r 3

h

ng 16 6 150 3 AS 14 ] ]

cha 2 m/ AS AS m y -1

m 12 t

AS2 k

y AS2 [

[ AS2 ci t t

da 10 o

n 100 h AS3 l 1 AS3 AS3 g u -1 e i

o 8 s

AS5 v AS5 AS5 l He a [m i 0 AS10 6 AS10

ll am AS10

a 50 f ent 4 g

n -1 2 a Rain T 0 0 -2 0.000 0.002 0.004 0.006 0.008 0.010 0.012 0.014 0 100 200 300 400 0 100 200 300 400 Cumulus mass flux [kgms-1] Radius [km] Radius [km]

Fig.3 Radial profiles of Fig.4 Radial profiles of azimuthally Fig.5 Vertical profiles of areally azimuthally averaged 36-hr averaged tangential wind velocity change (200-500 km radii) and temporally accumulated rainfall amount. (between the initial and the end of the (36 hr) averaged cumulus mass

integration) on the surface. Unit is flux.

transformed to ms-1 day -1.

References Ikawa, M., H. Mizuno, T. Matsuo, M. Murakami, Y. Yamada and K. Saito, 1991: Numerical modeling of the convective snow cloud over the Sea of Japan. -Precipitation mechanism and sensitivity to ice crystal nucleation rates-. J. Meteor. Soc. Japan, 69, 641-667. Kuma, K., C.-H. Cho, C. Muroi and M. Ueno, 1993: Improvement of typhoon track forecast by the JMA global model: An impact of cumulus parameterization. Papers presented at the third Technical Conference on SPECTRUM. Report No. TCP-33, WMO/TD-No. 595. World Meteor. Org., Geneva, Switzerland, Ⅳ 1-8. Lin, C., 1999: Some bulk properties of cumulus ensembles by a cloud-resolving model. Part II: Entrainment profiles. J. Atmos. Sci., 56, 3736-3748. Murata, A. and M. Ueno, 2000: The Effects of different cumulus parameterization schemes on the intensity forecast of typhoon Flo (1990). J. Meteor. Soc. Japan, 78, 819-833. Saito, K., T. Kato, H. Eito and C. Muroi, 2001: Documentation of the Meteorological Research Institute / Numerical Prediction Division unified nonhydrostatic model. Tec. Rep. MRI, 42, 133pp. Ueno, M., 2000: Numerical prediction for tropical cyclones. Meteorology Research Note, 197, 131-286 (in Japanese).

Development of a Cumulus Parameterization Scheme of the Operational Global Model at JMA

Masayuki Nakagawa Numerical Prediction Division, Japan Meteorological Agency 1-3-4 Otemachi, Chiyoda-ku, Tokyo 100-8122, JAPAN [email protected]

The cumulus parameterization scheme implemented in the Japan Meteorological Agency (JMA) Global Spectral Model (GSM) follows the scheme proposed by Arakawa and Schubert (1974) with modifications by Moorthi and Suarez (1992), Randall and Pan (1993) and Pan and Randall (1998). JMA revised the cumulus parameterization scheme of operational GSM in March 2001 to include reevapolation effect of the convective precipitation (GSM0103). The aim of this revision is to improve performance in the medium- and long-range predictions of tropical precipitation and associated circulation. However, this revision produced a systematic negative error in temperature and geopotential height fields at lower troposphere over the western portion of the North Pacific high in early forecast days. The main reason for this error is the excessive cooling by the evaporation of convective precipitation. At the same time, it is known from a forecast experiment that the GSM can not maintain the atmospheric general circulation over a long period without cooling and moistening effects of the reevapolation. JMA has been developing a new cumulus parameterization scheme which includes the entrainment and detrainment effects between the cloud top and the cloud base in convective downdraft instead of reevapolation of convective precipitation (GSM0206). To test the impact of the new scheme, two data assimilation experiments using GSM0103 and GSM0206 were conducted for the period 1-31 July 2001. Nine days forecasts were performed for 15 cases starting from 12UTC 8-22 July 2001. The results were compared with each other. Fig. 1. Mean forecast error of 850 hPa Figure 1 shows the mean forecast error temperature (K) over the tropics (20S- of 850 hPa temperature over the tropics 20N) averaged for the 15 cases. +: (20S-20N) averaged for the 15 cases with GSM0103, o: GSM0206. respect to the forecast time. It is seen that the GSM0103 forecast shows systematic negative bias throughout the forecast period up to - 0.42 K. The bias of the GSM0206 forecast is about half as much as that of the GSM0103 forecast. This difference is caused by the change of the cumulus parameterization scheme. Root mean square errors for the both experiments were also calculated and found to be comparable (not shown). The 24-hour forecast and its mean error for 500 hPa geopotential height field are shown in Fig. 2. The field simulated by GSM0103 is systematically lower than the analysis field in wide area extending from the North Pacific high to the Asian monsoon region. GSM0206 makes much better forecast than GSM0103 due to the smaller temperature bias at lower troposphere. In conclusion, the new cumulus parameterization scheme suppresses the systematic errors of GSM0103 in temperature and geopotential height fields at lower troposphere. JMA plans to implement this scheme in operational GSM in April 2003.

References. Arakawa, A. and W. H. Schubert, 1974: Interaction of a cumulus cloud ensemble with the large-scale environment, Part I. J. Atmos. Sci., 31, 674-701. Moorthi, S. and M. J. Suarez, 1992: Relaxed Arakawa-Schubert: A parameterization of moist convection for general circulation models. Mon. Wea. Rev., 120, 978-1002. Pan, D.-M. and D. Randall, 1998: A cumulus parameterization with a prognostic closure. Quart. J. Roy. Meteor. Soc., 124, 949-981. Randall, D. and D.-M. Pan, 1993: Implementation of the Arakawa-Schubert cumulus parameterization with a prognostic closure. Meteorological Monograph/The representation of cumulus convection in numerical models. J. Atmos. Sci., 46, 137- 144.

Fig. 2. Twenty four-hour forecast (contour) and mean forecast error (shade) of 500 hPa geopotential height (m) averaged for the 15 cases by GSM0103 (left) and GSM0206 (right). The contour interval is 60 m. Areas with mean forecast errors larger than 5 m are shaded. Implementation of the NCAR Community Land Model (CLM) in the NASA/NCAR Finite-Volume Global Climate Model fvGCM)(

Jon Radakovich1, Guiling Wanf, Shian-Jiann Lin, Sharon Nebuda and Bo-wen Shen

1. INTRODUCTION Amazon region. We performed several experiments designed to improve the CLM2 representation of surface In this study, the recently developed stateof- of-the-art hydrologic processes and the model's computational NCAR Community Land Model (CLM) version 2.0 land- performance. surface model (Dai et al. 2002; Zeng et al. 2002) was integrated into theNASA/NCAR finite-volume Global The experiments (Table 1), each of which included only Climate Model(fvGCM; Lin and Rood 2002). The CLM2 one of the modifications, were run for 5 years starting in provides a comprehensive physical representation of January 2000. All of the experiments were intercompared soil/snow hydrology and thermal dynamics and with the Control (the initial test case) based on a 2000- biogeophysics. The CLM2 was developed collaboratively 2004 average. The following experiments were by an open interagency/university group of scientists, and completed: the exponential csoilc scheme (Experiment I), based on well-proven physical parameterizations and the leaf heat capacity scheme (Experiment II), the implicit numerical schemes that combine the best features of leaf temperature scheme (Experiment III), the revised three previous land surface models: Biosphere- interception scheme (Experiment IV), the revised Atmosphere Transfer Scheme (BATS; Dickinson et al. interception with sub-surface runoff turned off 1993), the NCAR Land-surface Model (LSM; Bonan (Experiment V), and an experiment including all of the 1996), and the IAP94 snow model (Dai and Zeng 1996). modifications (Experiment VI). The Data Assimilation Office (DAO) has collaborated with NCAR to produce the NASA/NCAR fvGCM, which is a For Experiment I, csoilc was considered a unified climate, numerical weather prediction, and function of vegetation density as represented by the LAI chemistrytransport model suitable for data assimilation, (Leaf Area Index), in order to correct the warm bias with the DAO’s finite-volume dynamical core and NCAR’s resulting from the decoupling. Analysis of the results suite of physical parameterizations. revealed that there was a substantial impact, and the warm and dry bias in the fvGCM CLM2 was significantly 2. RESULTS reduced. The global annual mean bias and standard deviation for the intercomparisons of skin temperature The fvGCM coupled CLM2 was run at 2 x 2.5° horizontal with NCEP reanalysis presented in Figure 1, show a resolution with 55 vertical levels for a 15-year period from reduction in the standard deviation and the bias for 1991-2006 with initial conditions based on AMIP Experiment I compared to the Control. Experiment II, the (Atmospheric Model Intercomparison Project) and fixed leaf heat capacity scheme, which was shown to improve sea-surface temperatures based on an annual the memory of skin temperature and impact its diurnal climatology. The 10-year climate from the fvGCM CLM2 cycle, had only a marginal impact on the annual mean Control run was then intercompared with the climate from (Figure 1). Experiment III included changes to the fvGCM LSM, the European Center for Medium-range numerical scheme that solves the water and energy Weather Forecasting (ECMWF) reanalysis and the balance of the vegetation canopy. An implicit scheme, National Centers for Environmental Prediction (NCEP) which is scientifically accurate and computationally more reanalysis. We concluded that the incorporation of CLM2 efficient, replaced the explicit scheme previously used in did not significantly impact the fvGCM atmospheric CLM2 (Wang et al. 2002a). While the implicit scheme climate circulation from that of LSM. The most striking saves computation time, it does not cause noticeable difference was the warm bias in the CLM2 surface skin changes in the model results (Figure 1) temperature over desert regions, which was equal and opposite to the LSM cold bias (Figure 1). We determined For Experiment IV, the change involved that the warm bias can be partially attributed to the value incorporating precipitation sub-grid scale variability into of the drag coefficient for the soil under the canopy the canopy interception scheme, which causes a (csoilc), which was too small for sparsely vegetated decrease of interception loss and subsequent increase in regions resulting in a decouplingbetween the ground the canopy throughfall (Wang et al. 2002b). The results surface and the canopy. We also found that the canopy from the 5-year run show that the new interception interception was high compared to observations in the

1 1 Corresponding Author Address: Jon D. Radakovich, Science Applications International Corporation, Data Assimilation Office, NASA GSFC Code 910.3, Greenbelt, MD 20771, [email protected] run show that the new interception scheme causes about 0.5° in warming, which in turn increases the 3. REFERENCES CLM2 warm bias when compared to NCEP (Figure 1). The positive impacts were an increase Bonan, G. B., 1996: A land surface model (LSM in the low-level moisture and a significant version 1.0) for ecological, hydrological, decrease in the interception loss ratio (canopy and atmospheric studies. NCAR Tech. evaporation to precipitation). Experiment V Note NCAR/TN-417+STR, 150 pp. included the modified interception scheme but with Z. -L. Yang and G. -Y. Niu's sub-surface runoff Dai, Y., X. Zeng, 1996: A land surface model scheme turned off. This was done to correct some (IAP94) for climate studies. Part I: overestimation of lateral sub-surface runoff, which Formulation and validation in off-line may have resulted from not considering the impact experiments. Adv. Atmos. Sci., 14, 433- of topography in the runoff scheme. This 460. produced a more realistic runoff ratio (runoff to precipitation). Inhibiting the sub-surface runoff _____, and _____, R. E. Dickinson, I. Baker, G. also reduced the warming caused by the revised Bonan, M. Bosilovich, S. Denning, P. interception scheme (Experiment IV) and the Dirmeyer, P. Houser, G. Niu, K. Oleson, A. results from Experiment V did not deviate much Schlosser, and Z.-L. Yang, 2002: The from the Control (Figure 1). Common Land Model (CLM). (Submitted to Bull. Amer. Meteor. Soc.) In Experiment VI, all of the modifications were incorporated and the largest and most beneficial change was attributed to the exponential Dickinson, R. E., A. Henderson-Sellers, and P. J. csoilc scheme, which considerably decreased the Kennedy, 1993: Biosphere-Atmosphere warm bias in the CLM2 when compared to the Transfer Scheme (BATS) version 1e as Control. This result was expected based on coupled to the NCAR Community Climate Figure 1, which shows Experiment I having the Model. NCAR Tech. Note NCAR/TN- most substantial impact. Also, the standard 387+STR, 72 pp. deviation from Experiment VI does not differ greatly from that of the fvGCM LSM (Figure 1). Lin, S.-J., and R. B. Rood 2002: A “vertically Lagrangian”finite-volume dynamical core Table 1: Description of experiments. for global models. (Submitted to Mon. Wea. Rev.) Experiment Description Control Initial fvGCM CLM2 run Experiment I Exponential csoilc Experiment II Leaf heat capacity Wang, G. L., M. G. Bosilovich, J. D. Radakovich, Experiment III Implicit leaf temperature and P. R. Houser, 2002a: Incorporating Experiment IV Revised interception the implicit scheme for vegetation water Experiment V Exp. IV w/o subsurface runoff and energy balances into the coupled Experiment VI All of the modifications (I-V) fvGCM CLM2. Presented to the Land

Working Group at the CCSM Workshop, Breckenridge, CO.

_____, _____, _____, and _____, 2002b: A precipitation interception scheme for CLM2: considering the impact of sub-grid variability. Presented to the Land Working Group at the CCSM Workshop, Breckenridge, CO.

Zeng, X., M. Shaikh, Y. Dai, R. E. Dickinson, and R. Myneni, 2002: Coupling of the Common Land Model to the NCAR Figure 1. The global annual mean standard deviations of Community Land Model. J. Climate, 15, surface skin temperature between the Control, Experiments I- 1832-1854. VI (Table I), and fvGCM LSM versus NCEP. Implementation and testing of a multi-layer soil model in the NWP models of DWD Bodo Ritter, Reinhold Schrodin, Erdmann Heise, Martin Lange and Aurelia Mueller Deutscher Wetterdienst, Research and Development PO-Box 100465, 63004 Offenbach, Germany E-mail: [email protected] A multi-layer soil model, originally developed for the high resolution limited area model LM of DWD has been implemented in the global forecast model of DWD. The model comprises 7 active soil layers ranging in thickness from 1 cm close to the surface to roughly 5 m for the lowest layer. Below the active soil layers a climate layer with prescribed annual mean temperatures provides boundary conditions for the thermal heat flux computations. The hydrological part of the soil model operates on the same vertical grid and a partial or total freezing of soil water is accounted for in all layers (cf. Heise and Schrodin, 2002). Since the scheme differs considerably from the currently operational 2 layer scheme, it has to be tested and validated thouroughly before operational implementation, which is scheduled for autumn 2003. This validation exercise included a participation in the ELDAS RhoneAGG model intercom- parison experiment (cf. Boone et al., 2001). In the course of this exercise it was revealed that snow melting processes, which were based on the same single snow layer formulation as the previously soil scheme of DWD, were not simulated with sufficient accuracy. The problems were caused by an overestimation of snow density and the fact that ageing of the snow layer with a corresponding de- crease of the snow surface albedo for solar radiation was not considered in the original formulation. Adjustment of the snow density to more realistic values and the introduction of an empirical formula- tion describing the ageing of snow as a function of time since the last significant snow fall solved this problem. A one year integration based on initial data from January 1st 2002 of the full forecast model was performed in order to study potential bias problems in the evolution of the the soil layers, if, in contrast to the RhoneAGG experiment, where the forcing was prescribed, the soil scheme is forced by upper boundary fluxes provided by the atmospheric model. Since data assimilation schemes in operational NWP models introduce very little information directly into the soil scheme, it is essential that the soil model shows as little drift as possible with regard to prognostic variables. However, since a few W/m**2 imbalance in annual mean surface fluxes is sufficient to cause substantial deviations in the subsoil temperatures from their climatological mean values, the evolution of prognostic soil temperatures and water content, at least for high latitudes, is rather sensitive to the empirial relation describing the snow ageing. This demonstrates the need for a careful tuning of the free parameters of the soil model and a thorough validation of the scheme itself. It also supports the specification of a climatological lower boundary condition rather than a zero-flux condition. As an example from this integration Fig.1 compares the simulated global distribution of snow for

1 the end of the first month of the integration and the same month in the following year. The good overall agreement is an indicator of a realistic simulation of the evolution of the snow distribution in the course of one year. Further tests, in particular with regard to the comparison of the new versus the old scheme in operational forecasts and data assimilation will be performed in the near future.

Fig.1: Simulated snow water content at end of first month (top) and one year further (bottom)

References Boone, A., F.Habets and J.Noilhan, 2001: The Rhone-AGGregation Experiment. GEWEX News, WCRP, 11(3),3-5. Heise, E. and R.Schrodin, 2002: Aspects of snow and soil modelling in the operational short range weather prediction models of the German Weather Service. Journal of Computational Technologies, Vo.7, Special Issue: Proceedings of the International Conference on Modelling, Databases and Information Systems for Atmospheric Science (MODAS), Irkutsk, Russia, June 25-29, 2001, 121-140

2

¢¡¤£¦¥¨§©¡ §   ¨  ¦§© !#"$§© %¨&©§(') *§© +&©  *§  ,- /.10324¡ 5 ¦§ ¥

6879;:*<>=?<*@>A7BDCFEG:*H

IJ:*KMLN=?:PO?QRPLTSULVOWKMOWXFOWY?Z[BDLVHMLN=?KM\^]+_JLV

dfeg g5hjilk?mWnWo i

I `pRPLVXFq OW:*KM<*L?`rtsb_¦7uWvWv `*a¦:*HbSUKU=?XFCc=7

Lxw-yz=?CFX#{687p9;:*<|tq OWyP7 YWO¢} 7 =?:

~(^€‚lƒG„† N‡¢€‰ˆ‰„G‡NŠ¨ˆ‰‹!ŒŽ¨b!x‘’$“‡¢„!^ ¢x€‚¢”*x„•‡¢‹G„†G‡¢8–*^x‹†—bƒG‹G‹Gˆ‰‹!Œ5¢”*^—b‡NŠ¨ˆ‚l‹G‡¢€‰€‚˜’‡NŠŠ¨!5™ƒ!—¨x‡¢ƒ

¢šœ›^Š-^¢—¨l€‚¢Œ¢˜ž¨ˆ‰‹GUDŠ¨!Žx‡N—b€‚˜ ŸN¡¢’¢c£ˆ‚¤WƒG^¥©¦xŸ¢Ÿl§¢¨b©$ª«Gˆ‰«¨b!x‘’tƒG-x”;¬˜W¨ˆ‰^‡¢€­”;‡N—b‡¢‘’^Š-^—bˆ‚®x‡NŠ¨ˆ‚l‹G

ˆ‰„!x‹¬Š¨ˆ‰^‡¢€WŠ-DŠ¨!l-ƒG-x„’ˆ‰‹tŠ¨!«™ƒ!—¨x‡¢ƒ’¢š­›^Š-^¢—¨l€‚¢Œ¢˜•¯$€‚¢–;‡¢€!~J¨¨ˆ‰‘fˆ‰€‰‡NŠ¨ˆ‚l‹¦‡¢‹G„f°—¨x„Gˆ‰UŠ¨ˆ‚l‹²±?˜W-Š-x‘

¢³¯$~5±W°¨$‡¢‹G„[Š¨!fšF¢—¨x^‡¢-ŠD´Gx€‰„G$šF—¨l‘µ¯$~5±W°)Š-Œ¢x‹!^—b‡NŠ-†‹G‡N—¨—¨T“¶–;‡¢‹G„[–G—bˆ‚Œl¬Š¨‹!x¨$Š-x‘’”*^—b‡NŠ¨ƒ!—¨x

U¢—¨—¨x-”*l‹G„Gˆ‰‹!ŒtŠ-²‡f‹?ƒG‘Ž–*^—¢š·-”*xUŠ-—b‡¢€ˆ‰‹!šF—b‡N—¨x„bG‡¢‹G‹!x€‰¢š¸Œ¢^l-Š¨‡NŠ¨ˆ‚l‹G‡N—¨˜¨‡NŠ-x€‰€‰ˆ‚Š-xŽ¢F¢© Œ!©«¯5›ž±G¨b©

ª«!¹‘’W„!x€5^€‚lƒG„º‡¢‹G„º—b‡¢„Gˆ‰‡NŠ¨ˆ‚l‹º¨b!x‘’x ‡N—¨¹Š¨!x‹º N‡¢€‰ˆ‰„G‡NŠ-x„,–?˜,Ul‘’”;‡N—bˆ‰-l‹º¢š¦Š¨!¹‘’W„!x€‰€‚x„

–G—bˆ‚Œl¬Š¨‹!x¨Š-x‘’”*^—b‡NŠ¨ƒ!—¨5“ˆ‚Š¨•Š¨!D—¨x‡¢€¨‡NŠ-x€‰€‰ˆ‚Š-Dˆ‰‘f‡NŒ¢^—¨˜¢©

‹!¦”;‡N—¨Š¨ˆ‰^ƒG€‰‡N—”G—¨¢–;€‚x‘¼ˆ‰‹Š¨!²¯$~5±W°º-˜W-Š-x‘½ˆ‰„!x‹¬Š¨ˆ‚´Gx„“ˆ‚Š¨ Š¨Gˆ‰J¨b!x‘’’ˆ‰Š¨G‡NŠJŠ¨!²¯$~5±W°

»

‘’W„!x€*„!?x«‹!¢Š”G—¨W„GƒGU$x‹!lƒ!ŒlUl€‰„•–G—bˆ‚Œl¬Š¨‹!x¨Š-x‘’”*^—b‡NŠ¨ƒ!—¨xˆ‰‹†Š¨!5Š-—¨¢”;ˆ‰^^¾WŠ¨!D¨‡NŠ-x€‰€‰ˆ‚Š-5ˆ‰‘f‡NŒ¢

¨!T“f‡¢‹¿‡N”G”G—¨x^ˆ‰‡N–;€‚†”*^—bUx‹¬Š¨‡NŒ¢ ¢šJ–G—bˆ‚Œl¬Š¨‹!x¨tŠ-x‘’”*^—b‡NŠ¨ƒ!—¨–*x€‚T“ÁÀN¡¢¡¢Â “Gˆ‰€‰-Š¦Š¨!‘fˆ‰‹Gˆ‰‘¦ƒG‘

‘’W„!x€ –G—bˆ‚Œl¬Š¨‹!x¨«Š-x‘’”*^—b‡NŠ¨ƒ!—¨x$‡N—¨¦l‹G€‚˜‡N—¨lƒG‹G„žÀ¢ÀN¡¢Â’©­~šFŠ-^—5‡†-^—bˆ‚xJˆ‰‹¬ ¢x-Š¨ˆ‚Œl‡NŠ¨ˆ‚l‹zŠ¨!¦”G—¨¢–;€‚x‘

ˆ‰’šFlƒG‹G„+Š->–*„Gƒ!Š-PŠ¨!zˆ‰U“‡NŠ-^—•Ul‹¬Š-x‹¬Š„Gˆ‰‡NŒl‹!l-x„+šF—¨l‘Š¨!z^€‚lƒG„+Š-x‘’”*^—b‡NŠ¨ƒ!—¨“Gˆ‰b¿ˆ‰

Š-?†€‚T“ÇNŠŠ¨!¦Ul€‰„ ^€‚lƒG„Š-x‘’”*^—b‡NŠ¨ƒ!—¨D—¨^Œlˆ‚l‹­©ª«!Žˆ‰‘’”;€‚x‘’x‹¬Š¨‡NŠ¨ˆ‚l‹¢š·‡²‹!^“ÄGˆ‰‡NŒl‹!l-Š¨ˆ‰D¨b!x‘’

”G—¨¢”*l-x„²–?˜’Ň¢‹!Œf‡¢‹G„±W‡¢¨-x‹ž¢ÅÀN¡¢¡lÀl¨8–;‡¢-x„²l‹†Š¨!$~J£J› €‰ˆ‰„G‡N—œ¢–;-^—¨ N‡NŠ¨ˆ‚l‹†€‚x‡¢„G8Š-t‡t¨ˆ‚Œl‹Gˆ‚´;^‡¢‹¬Š

ˆ‰‘’”G—¨T ¢x‘’x‹¬Šˆ‰‹•Š¨!D‘’W„!x€¨ˆ‰‘¦ƒG€‰‡NŠ¨ˆ‚l‹G^©

ˆ‚Œlƒ!—¨Ž¦^‡Vdz–•¨!T“œ‡ŽUl‘’”;‡N—bˆ‰-l‹’–*^ŠÈ“8^x‹²Š¨!J‘’W„!x€‰€‚x„t–G—bˆ‚Œl¬Š¨‹!x¨·Š-x‘’”*^—b‡NŠ¨ƒ!—¨‡¢‹G„²‘’x‡¢¨ƒ!—¨UÇ

Æ

‘’x‹¬Š¨5šF—¨l‘j¯5›ž±!ɕ¨‡NŠ-x€‰€‰ˆ‚Š-’ˆ‰‘f‡NŒ¢^—¨˜ l‹¹ÊËlƒG‹!fÀN¡¢¡lÀ?©’ª«!’‘’W„!x€¸Œ¢x‹!^—b‡¢€‰€‚˜ ^‡N”GŠ¨ƒ!—¨x5Š¨!’‘f‡TÌ-¢—

bG‡N—b‡¢UŠ-^—bˆ‰-Š¨ˆ‰^8¢šŠ¨!J–G—bˆ‚Œl¬Š¨‹!x¨Š-x‘’”*^—b‡NŠ¨ƒ!—¨$„Gˆ‰-Š-—bˆ‚–;ƒ!Š¨ˆ‚l‹t”;‡NŠ-Š-^—b‹†¢–;-^—¨ ¢x„²–?˜fŠ¨!$¨‡NŠ-x€‰€‰ˆ‚Š-¢©·™ƒ!Š

Š¨!J‘’W„!x€‰€‚x„’Š-x‘’”*^—b‡NŠ¨ƒ!—¨Jˆ‰‹!¢ŠUl€‰„fx‹!lƒ!Œl†Ul‘’”;‡N—¨x„f“ˆ‚Š¨²¨‡NŠ-x€‰€‰ˆ‚Š-«—¨x¨ƒG€‚Š¨^¥¬x-”*x^ˆ‰‡¢€‰€‚˜¦T ¢^—Š¨!

Š-—¨¢”;ˆ‰^‡¢€—¨^Œlˆ‚l‹­©·ª«!$‘fˆ‰‹Gˆ‰‘¦ƒG‘¶–G—bˆ‚Œl¬Š¨‹!x¨œŠ-x‘’”*^—b‡NŠ¨ƒ!—¨J¢–;-^—¨ ¢x„†–?˜fŠ¨!5¨‡NŠ-x€‰€‰ˆ‚Š-$ˆ‰J¦xŸ¢ÍNÂΖ;ƒ!ŠœŠ¨!

‘’W„!x€‰€‚x„•‘fˆ‰‹Gˆ‰‘¦ƒG‘Ï N‡¢€‰ƒ!5ˆ‰8l‹G€‚˜•À¢À¢ÀN’©;ª«!5„GˆpÐ*^—¨x‹GUDˆ‰À¢ŸN’©GÄ[ŽG‡x ¢DUÑ!‡¢‘fˆ‰‹!x„•šFlƒ!—¢Š¨!^—«ˆ‰U

^€‚lƒG„¢”GŠ¨ˆ‰^‡¢€”G—¨¢”*^—¨ŠÈ˜¨b!x‘’xJ‡¢‹G„šFlƒG‹G„Š¨G‡NŠšF¢—JŠ¨!¦‘f‡VÑ!ˆ‰‘¦ƒG‘Òˆ‰UŽ“‡NŠ-^—5Ul‹¬Š-x‹¬Š5ƒG-x„ ˆ‰‹Š¨!

^‡¢€‰^ƒG€‰‡NŠ¨ˆ‚l‹G8Š¨!D‘f‡VÑ!ˆ‰‘¦ƒG‘Ó„GˆpÐ*^—¨x‹GU5ˆ‰‹†–G—bˆ‚Œl¬Š¨‹!x¨8Š-x‘’”*^—b‡NŠ¨ƒ!—¨5„Gƒ!$Š-fƒG¨ˆ‰‹!Œt„GˆpÐ*^—¨x‹¬Š¨b!x‘’Dˆ‰

’¥;‹!¢Š–;ˆ‚Œ’x‹!lƒ!ŒlŠ-fUÑW”;€‰‡¢ˆ‰‹•Š¨!D„Gˆ‰¨U—¨^”;‡¢‹GU˜²–*^ŠÈ“8^x‹z‘’W„!x€‰€‚x„†—¨x¨ƒG€‚Š¨‡¢‹G„¢–;-^—¨ N‡NŠ¨ˆ‚l‹G^©

Ô

£«xUx‹¬Š¨€‚˜¢¥Ň¢‹!Œ‡¢‹G„P±W‡¢¨-x‹+¢ÅÀN¡¢¡lÀl¨J”G—¨x-x‹¬Š-x„[‡‹!^“—¨x€‰‡NŠ¨ˆ‚l‹G¨Gˆ‚” –*^ŠÈ“8^x‹¹ˆ‰Ut“‡NŠ-^—ŽUl‹¬Š-x‹¬Š

‡¢‹G„[Š-x‘’”*^—b‡NŠ¨ƒ!—¨²„!^Š-^—b‘fˆ‰‹!x„žšF—¨l‘¤Õ˜¢x‡N—b•¢Ö¦xŸ¢Ÿl§xÇ#ÀN¡¢¡¢¡¬¨Ž¢š€‰ˆ‰„G‡N—-dz—b‡¢„G‡N—$¢–;-^—¨ N‡NŠ¨ˆ‚l‹GŽUl€‰€‚xUŠ-x„P‡NŠ

Š¨!†±?lƒ!Š¨!^—b‹P¯J—¨x‡NŠ¦°œ€‰‡¢ˆ‰‹G^¥­‡NŠDŠ¨!²¨ˆ‚Š-’¢š8Š¨!²^€‚lƒG„[‡¢‹G„[—b‡¢„Gˆ‰‡NŠ¨ˆ‚l‹žŠ-x-Š-–*x„[¢š~Š¨‘’l-”;!^—bˆ‰’£‡VÇ

„Gˆ‰‡NŠ¨ˆ‚l‹›x‡¢¨ƒ!—¨x‘’x‹¬ŠŽ¢c~J£J›ž¨8”G—¨¢Œ¢—b‡¢‘Òˆ‰‹ ¤W€‰‡¢!l‘f‡W©ª«!^˜•Ul‘’”;‡N—¨x„Š¨!xˆ‚——¨x¨ƒG€‚Š¨“ˆ‚Š¨‡¢ˆ‚—bU—b‡NšFŠ

»

¢–;-^—¨ N‡NŠ¨ˆ‚l‹GŠ¨‡N¤¢x‹†T ¢^—Š¨!J‘fˆ‰„WÇȀ‰‡NŠ¨ˆ‚Š¨ƒG„!x·‡¢‹G„²šFlƒG‹G„’Š¨G‡NŠ8Š¨!Jˆ‰U«“‡NŠ-^—Ul‹¬Š-x‹¬Š„!^—bˆ‚ ¢x„fšF—¨l‘¶Š¨!

*Ø

‡¢ˆ‚—bU—b‡NšFŠ©¢–;-^—¨ N‡NŠ¨ˆ‚l‹G¸ˆ‰©Gˆ‚Œl!^—©šF¢—©“‡N—b‘ˆ‰U^€‚lƒG„G¢cª,×¹Ç#ꢃ ¨‡¢‹G„t€‚T“8^—¸šF¢—·Ul€‰„tˆ‰U^€‚lƒG„G¢cª,Ù

*Ø

dzÕl¡ ¨b© ˆ‚Œlƒ!—¨5¦^«¨!T“·Š¨!—¨x€‰‡NŠ¨ˆ‚l‹G¨Gˆ‚”¦–*^ŠÈ“8^x‹²ˆ‰U“‡NŠ-^—8Ul‹¬Š-x‹¬Š8‡¢‹G„tŠ-x‘’”*^—b‡NŠ¨ƒ!—¨«„!^Š-^—b‘fˆ‰‹!x„

Æ

ƒG¨ˆ‰‹!Œ²lƒ!—5¨b!x‘’¢c-l€‰ˆ‰„ž^ƒ!—¨ ¢V¨‡¢‹G„zŠ¨G‡NŠ5„!^Š-^—b‘fˆ‰‹!x„ –?˜ Ň¢‹!Œ‡¢‹G„¹±W‡¢¨-x‹¿¢c„!¢Š-Š-x„[^ƒ!—¨ ¢V¨b©Žª«!

-˜W‘Ž–*l€‰—¨^”G—¨x-x‹¬ŠŠ¨!¦‡¢ˆ‚—bU—b‡NšFŠ«¢–;-^—¨ N‡NŠ¨ˆ‚l‹G«Š-²“Gˆ‰bŠ¨!¦-l€‰ˆ‰„^ƒ!—¨ ¢Ž“‡¢´GŠ-Š-x„­©«ª«!D–G—bˆ‚Œl¬Š¨‹!x¨

Ø

Š-x‘’”*^—b‡NŠ¨ƒ!—¨„!^Š-^—b‘fˆ‰‹!x„>“ˆ‚Š¨ÚŠ¨!~J£J›Û¨b!x‘’ˆ‰’¨!T“‹+ˆ‰‹>´GŒlƒ!—¨[¦^„­© €‚x‡N—b€‚˜¢¥Š¨!ˆ‰‘’”G—¨T ¢UÇ

‘’x‹¬ŠšF—¨l‘Š¨!x-DbG‡¢‹!Œ¢x«ˆ‰¨ˆ‚Œl‹Gˆ‚´;^‡¢‹¬Š^©ª«Gˆ‰8^‡¢‹•–*5-^x‹•šF—¨l‘ÓŠ¨!5^€‚lƒG„G8T ¢^—Š¨!$Š-—¨¢”;ˆ‰^“Gˆ‰b

‡N—¨Ž–G—bˆ‚Œl¬Š-^—«Š¨G‡¢‹Š¨!l-¦¨!T“‹ ˆ‰‹´GŒlƒ!—¨²¦U–©ª«!D¢Š¨!^—$^€‚x‡N—^ Wˆ‰„!x‹GUŽˆ‰Š¨!¦‘’W„!x€‰€‚x„‘fˆ‰‹Gˆ‰‘¦ƒG‘

–G—bˆ‚Œl¬Š¨‹!x¨ Š-x‘’”*^—b‡NŠ¨ƒ!—¨œ“Gˆ‰b¦„!—¨¢”;šF—¨l‘ÎÀ¢À¢ÀNÂ,Š-5ÀN¡ ’©¢ª«Gˆ‰ ˆ‰‘’”G—¨T ¢x‘’x‹¬Š¸ˆ‰‹G„Gˆ‰^‡NŠ-x Š¨G‡NŠ©Š¨!ˆ‰U

Ô

“‡NŠ-^—Ul‹¬Š-x‹¬Š‡NŠ8Ul€‰„fŠ-x‘’”*^—b‡NŠ¨ƒ!—¨xœˆ‰‹’Š¨!$¯$~5±W°Ú-˜W-Š-x‘ψ‰Š-?¦¨‘f‡¢€‰€Å¾¬ˆ‚Šœ‡¢€‰-D—b‡¢ˆ‰-xœ‡ŽÜ?ƒ!x-Š¨ˆ‚l‹fl‹ ‡

ˆ‚Œlƒ!—¨¦¦¡ °¸‡¢‹!x€‡¦¨!T“8Š¨!J–G—bˆ‚Œl¬Š¨‹!x¨Š-x‘’”*^—b‡NŠ¨ƒ!—¨šF—¨l‘ ¯5›ž±!ɦ¨‡NŠ-x€‰€‰ˆ‚Š-l‹•Ê¦ËlƒG‹!$ÀN¡¢¡lÀ?©8°¸‡¢‹!x€

Æ

–¿ˆ‰¦Š¨!‘’W„!x€‰€‚x„>—¨x¨ƒG€‚Š¨tƒG¨ˆ‰‹!ŒzŠ¨!¢—bˆ‚Œlˆ‰‹G‡¢€¯$~5±W°¶ˆ‰U“‡NŠ-^—²Ul‹¬Š-x‹¬Š†¨b!x‘’¢©Ú°¸‡¢‹!x€«„+ˆ‰¦Š¨!

¨‡¢‘’D‡¢”;‡¢‹!x€––;ƒ!ŠƒG¨ˆ‰‹!Œ¦Š¨!$Ň¢‹!Œ²‡¢‹G„ ±W‡¢¨-x‹ˆ‰U$“‡NŠ-^—Ul‹¬Š-x‹¬ŠJ¨b!x‘’¢©”;‡¢‹!x€*$”G—¨x-x‹¬Š¨Š¨!

ˆ‰UD“‡NŠ-^—5Ul‹¬Š-x‹¬Š$‡¢J‡fšcƒG‹GUŠ¨ˆ‚l‹¢š·^€‚lƒG„Š-x‘’”*^—b‡NŠ¨ƒ!—¨Ž„!^Š-^—b‘fˆ‰‹!x„ƒG¨ˆ‰‹!Œ’–*¢Š¨Š¨!t¯$~5±W°,¨b!x‘’

–;‡¢-x„zl‹[‡¢ˆ‚—bU—b‡NšFŠ$¢–;-^—¨ N‡NŠ¨ˆ‚l‹+¢c-l€‰ˆ‰„z^ƒ!—¨ ¢V¨J‡¢‹G„zŇ¢‹!Œ‡¢‹G„¹±W‡¢¨-x‹ž¨b!x‘’¢c„G‡¢¨!x„[^ƒ!—¨ ¢V¨b©Žª«!

-˜W‘Ž–*l€‰ˆ‰‹•”;‡¢‹!x€­5—¨^”G—¨x-x‹¬Š¨«‡¢ˆ‚—bU—b‡NšFŠ«¢–;-^—¨ N‡NŠ¨ˆ‚l‹G^©

Š¨!5‡¢^^ƒ!—b‡¢U˜†¢š ˆ‰U$“‡NŠ-^—«Ul‹¬Š-x‹¬Š‘’x‡¢¨ƒ!—¨x‘’x‹¬Š¨šF—¨l‘Ó‡¢ˆ‚—bU—b‡NšFŠ‡NŠ«Ul€‰„†Š-x‘’”*^—b‡NŠ¨ƒ!—¨J—¨^Œlˆ‚l‹•‡¢Š¨!

„Gˆ‰‡NŒl‹!l-Š¨ˆ‰ˆ‰U“‡NŠ-^—8Ul‹¬Š-x‹¬Š¨b!x‘’ƒG-x„’ˆ‰‹tŠ¨!$¯$~5±W°P‘’W„!x€!ˆ‰¸–;‡¢-x„fl‹f‡¢ˆ‚—bU—b‡NšFŠ‘’x‡¢¨ƒ!—¨x‘’x‹¬Š¨^©

Ø

££¢ ¢œ££¢¥¤ ¢±

Æ

£ˆ‚¤WƒG^¥§¦©‚¥©~”G”;€‰ˆ‰^‡NŠ¨ˆ‚l‹ž¢š‡¨b!x‘’fšF¢—Ž N‡¢€‰ˆ‰„G‡NŠ¨ˆ‰‹!Œ^€‚lƒG„[ˆ‰‹¹‡¢‹[¢”*^—b‡NŠ¨ˆ‚l‹G‡¢€·Œl€‚¢–;‡¢€¨¤Ä,°Ã‘’W„!x€Å¥

¦xÊW¦xÉTÇb¦xÊ¢Íl§¬¥¦xŸ¢Ÿl§¬©

© ¡   ¡ ¡ !

Ø

Ň¢‹!Œ!¥¥"©©‚¥¸‡¢‹G„P’©¸±W‡¢¨-x‹­¥ ˆ‚—¨—bƒG5^€‚lƒG„¹‘fˆ‰U—¨¢”;¬˜W¨ˆ‰^‡¢€·”G—¨¢”*^—¨ŠÈ˜z—¨^Š-—bˆ‚^ N‡¢€œƒG¨ˆ‰‹!Œ€‰ˆ‰„G‡N—D‡¢‹G„¹—b‡¢„G‡N—

‘’x‡¢¨ƒ!—¨x‘’x‹¬Š¨^©•°¸‡N—¨Š$##% ©› ˆ‰„G€‰‡NŠ¨ˆ‚Š¨ƒG„!’^ˆ‚—¨—bƒG5‘fˆ‰U—¨¢”;¬˜W¨ˆ‰^‡¢€·‡¢‹G„[—b‡¢„Gˆ‰‡NŠ¨ˆ‚ ¢f”G—¨¢”*^—-Ç

Š¨ˆ‚x^¥'& À¢À¢ŸW¦MÇ#À¢ÍN¡lÀ?¥ ÀN¡¢¡lÀ?© §(*),+- £./102 3 4!

¢¡¤£¦¥¤§¨£ ©  ©¤£ © §¥¤ £! ©#" £ %$&©

*£$£ +, $&©!§¨ +¡¤£-*£

')( ( (

.0/2143&56/87%9;:<78=>@?4143

ACBEDGFIHIFIJEFLKM8N+OPJRQ-KGSTHVUCWXZY[WF]\EKGQWFIHVUGDC^<_J`DCDTHVWBba&UCWdceKGQfgM8NPhiUjHkKGBiUTKTDC^

WBic_J`DCDTHVWBblPfdcSmMQ-KjFLKnMSmMXoMnpHVUCWX6_KTDqKnWSmUG\sr

SnKncFLKnUTKGBEDC}@H]HZ~KGS ^ Y[M@DjUCMƒ)_„ hhRAma

uTvZwx-y z|{ z w€ex€‚E€

†‚‡bˆ`‰VŠŒ‹]mŽeŠ ‚L‘’ “+”G’t‡ •–]‡ †j•jŽe‡b–]”q—

˜P™Zš[›iœ mž Ÿ¢ ]¡£¢s¤R¥]¦R§¨¢`¥&¡Tš@©b ªnœ<¢`¤R«G¢`¬Z¤R I¢`¬)­%®°¯±©b¦t²Zš@¥³% Œ´G™µ`¢`«T I¢`§Z¥]šb«Tš@¡T¦R¥]¶i´G ]¦R¬·¢`¡g¢`¬

¦R¸i´G ]¦R¬º¹o»@¼2½ ˜P™Zš ©b¦t²Zš@¥¨ ]¬Z¾@¥]¶Z²Zš@¡;´G™Zšb¸¨¢`«G¢`©bš‚´Gš@«T ]¿¢e´G ]¦R¬À¸¨¢`¾jÁe¢`¤Rš ¦`Â+¡T¶Z§Z¤R«T ]²tm¡T¾¢`¥]šÃ¸Z«T¦t¾@š@¡T¡Tš@¡

Âk«T¦R©Ä´G™ZšÆŨ«Tš@¬Z¾j™Ç¦R¸|š@«G¢e´G ]¦R¬¨¢`¥ ©b¦t²Zš@¥ žÈɯÊ&ËÊ&ÌeÍnŨ›°¹ Îd¼2½·˜P™Zš[²Zš@š@¸Ç¾@¦R¬EµRš@¾‚´G ]¦R¬Ï¡T¾j™Zš@©bš[ ]¡

§¨¢`¡Tš@²Ã¦R¬[¹ Ðe¼i§Z¶i´¨ ]¬Z¾@¥]¶Z²Zš@¡4¬¶Z©bš@«T¦R¶Z¡4²Zš‚µRš@¥]¦R¸Z©bš@¬E´G¡%Ñk¸¨¢`«G¢`©bš‚´Gš@«T ]¿¢e´G ]¦R¬Ò¦`ÂÓ²Z¦³%²Z«G¢`¶Z¤R™E´4Ô¨¶tÕiš@¡@Ö

«Tš@²Z ]¡C´G«T ]§Z¶i´G ]¦R¬Ã¦`Â6©b¦R©bš@¬E´G¶Z©ÆÖEš‚´G¾×q½&Ø ¶Z«T«Tš@¬E´G¥ŒÙôG™ZšÉ«Tš@¡T¦R¥]¶i´G ]¦R¬Ã¦` ´G™ZšÉ©b¦t²Zš@¥¨ ]¡%»R½ ÚEÛRÜEÎRÝÕ »R½Œ»dÎRÝ

²Zš@¤R«Tš@š@¡+¥]¦R¬ÓÌe¥I¢e´Ö¢`¬Z²[Î`ÞµRš@«C´G ]¾¢`¥|¥]š‚µRš@¥]¡@½4Í8´ ³+¢`¡ Âk¦R¶Z¬Z²b´G™¨¢e´ Âk¦R«´G™Z ]¡ÑL¢`¬Z²¾@¦E¢`«T¡Tš@«j׫Tš@¡T¦R¥]¶i´G ]¦R¬6Ö

´G™Zš ß;¶Z¦Æ¾@¥]¦R¡T¶Z«TšgÂk¦R«²Zš@š@¸¾@¦R¬EµRš@¾‚´G ]¦R¬¸¨¢`«G¢`©bš‚´Gš@«T ]¿¢e´G ]¦R¬à³ ¦R«TÁ¡-³ š@¥]¥¨ ]¬¤©b ]²Z¥I¢e´G Œ´G¶Z²Zš@¡@Ö6³%™Z ]¥]š

´G™Zš#¾@¥]¦R¡T¶Z«Tš#§¨¢`¡Tš@²Ç¦R¬ÇØ ž¯Ê¢Ñk¾@¦R¬EµRš@¾‚´G ŒµRšs¢µ`¢` ]¥I¢`§Z¥]šb¸|¦`´Gš@¬E´G I¢`¥ š@¬Zš@«T¤`ÙZ×Ò ]¡£§|š‚´T´Gš@«g¡T¶Z Œ´Gš@²áÂk¦R«

´G«T¦R¸Z ]¾@¡@½

Í8´à ]¡Æ¸Z«T¦R¸|¦R¡Tš@²â´G¦º¶Z¡Tšá¢Ï¾@¦R©£§Z ]¬¨¢e´G ]¦R¬°¦`£´G™Zš@¡Tš·´m³ ¦º¾@¥]¦R¡T¶Z«Tš@¡@Ö-²Zš@¸|š@¬Z²Z ]¬Z¤ã¦R¬¢µRš@«G¢`¤Rš

´Gš@©b¸|š@«G¢e´G¶Z«Tš%¬Zš¢`«4´G™Zš%¡T¶Z«CÂL¢`¾@š`½¨˜P™Z ]¡¨ ]²Zš¢³+¢`¡¨´Gš@¡C´Gš@² ]¬g´G™Zš+Âk«G¢`©bš‚³ ¦R«TÁ£¦`Â|´G™Zš›iœ mž Ÿä©b¦t²Zš@¥2½

˜P™Zš ´Gš@©b¸|š@«G¢e´G¶Z«TšÃ³%™Z ]¾j™á¢`¾‚´G Œµ`¢e´Gš@¡gØ ž¯Êå¾@¥]¦R¡T¶Z«Tšb ]¬Z¡C´Gš¢`²·¦`Â%ß;¶Z¦s¾@¥]¦R¡T¶Z«Tš ]¡-¾j™Z¦R¡Tš@¬·´G¦à§|š

ÐRÛEÎt½Œ»-ßg½Óæ§Eµt ]¦R¶Z¡T¥ŒÙRÖZ´G™Z ]¡+´G™Z«Tš@¡T™Z¦R¥]²Æ¾¢`¬Æ§|š;¶Z¡Tš@²ÆÂk¦R«+´G¶Z¬Z ]¬Z¤g´G™Zš¸¨¢`«G¢`©bš‚´Gš@«T ]¿¢e´G ]¦R¬6½

欣Åç ]¤Z½»³ š ¸Z«Tš@¡Tš@¬E´¨¿@¦R¬¨¢`¥©bš¢`¬´Gš@©b¸|š@«G¢e´G¶Z«Tš´G«Tš@¬Z²g¢µRš@«G¢`¤Rš@²£¦µRš@« »dÎ+èZµRšqm²¨¢Ù-Âk¦R«Tš@¾¢`¡C´G¡

¡C´j¢`«C´G ]¬Z¤áÂk«T¦R©é»dÝd´G™ä¦`ÂҚ¢`¾j™°©b¦R¬E´G™å»êRêRÜiÖÛRÛRÛRÛáëɘÉØìÑLÊ&ØPí[®°Å¢e´j¢E×bÂk¦R«#š¢`¾j™ä´mÙt¸|š¤¦`Â

¾@¥]¦R¡T¶Z«Tš£¢`¬Z²sÂk¦R«¾@¦R©£§Z ]¬Zš@²s¾@¥]¦R¡T¶Z«Tš`½;æ¬ZšÒ¾¢`¬à¡Tš@šÒ´G™¨¢e´´G™ZšÒ¶Z¡Tš£¦`¨¾@¦R©£§Z ]¬Zš@²à¾@¥]¦R¡T¶Z«TšÒ«Tš@²Z¶Z¾@š@¡

¾@¦R¥]²s§Z I¢`¡É ]¬s¶Z¸Z¸|š@«É´G«T¦R¸|¦R¡T¸Z™Zš@«Tš£ ]¬Æ´G«T¦R¸Z ]¾@¡%³% Œ´G™s«Tš@¡T¸|š@¾‚´´G¦ðïC¸Z¶Z«Tš£ß;¶Z¦Rñ ¾¢`¡Tš`Ö §Z¶i´²Z¦tš@¡¬Z¦`´

¸Z«T¦t²Z¶Z¾@š@¡£šqÕi¾@š@¡T¡T ŒµRš#³+¢`«T©ò§Z I¢`¡Ò¬Zš¢`«-´G™Zš ¡T¶Z«CÂL¢`¾@š`Ö¢`¡ÀïC¸Z¶Z«Tš#Ø ž¯ʨñÀ²Z¦tš@¡@½s˜P™Zš@«Tš# ]¡Ò¢`¥]¡T¦à¢

¡T© ¢`¥]¥ ]©b¸Z«T¦µRš@©bš@¬E´É ]¬ÆÈíà›¡T¾@¦R«Tš@¡%Âk¦R«P¾@¦R©£§Z ]¬Zš@²Æ¾@¥]¦R¡T¶Z«Tš`½

ó

£ E£ £-¤ £§

¹o»@¼ í)½ žÒ½ô˜õ¦R¥]¡C´mÙtÁ™6ÖÛtš@©b ªnœ<¢`¤R«G¢`¬Z¤R I¢`¬°™Z ]¤R™«Tš@¡T¦R¥]¶i´G ]¦R¬©b¦t²Zš@¥-¦`Âb´G™Zšá¢e´G©b¦R¡T¸Z™Zš@«Tš·Âk¦R«

¬¶Z©bš@«T ]¾¢`¥4³ š¢e´G™Zš@«£¸Z«Tš@²Z ]¾‚´G ]¦R¬6Ö¨ö Ö¨­ÉÚZÖ&»‚nê·Ñ2Î`ÛRÛi»d×q½

—Ej‚‰Lˆ`÷ø)†‚m†jŽe”GŽeŠ]ŽGù`Àˆ`÷|ú[ûRú`”GŽeŠ]ŽGù`

ÑLÈ%¶Z¡T¡T I¢`¬Æ¢`¬Z²sʨ¬Z¤R¥] ]¡T™Ó×

¹ Îd¼ýüi½ nÅ ½tË;š@¥]š‚Ùt¬6ÖbÊP½iþ+¢`¿@ ]¥]š`Ög¯õ½iþ ¦R¶Z¤Rš¢`¶Z¥Œ´ Öž+´G©b¦R¡T¸Z™Zš@«T ]¾É¸¨¢`«G¢`©bš‚´Gš@«T ]¿¢e´G ]¦R¬Ã¡T¾j™Zš@©bš@¡

†‚ˆ`Š

]¬íƚ‚´Gš@¦enŨ«G¢`¬Z¾@š`ÿ ¡ÀžÈɯÊ&ËÊ ­-½ ®å½ ¯õ½©b¦t²Zš@¥2½%Ím¬¢¡

ˆ`”Tˆ`‡ †‚m†‚”q‰¤£ˆ`L‰2Že÷ýŽ¦¥À‚—¨§mù`”q‰Lú ©2•Gˆ`Š]†

ÖZÊ&ØPí[®°Åº›tš@©b ]¬¨¢`«+¸Z«T¦t¾@š@š@²Z ]¬Z¤R¡ÒÑn»êRê`Ú×qÖ¨ÐRÞEݝmÚEÛEÎt½

¨’t‚‰2•Gˆ`Š |”GŽ•j†jj†j

¹ Ðe¼ ¯õ½£þ ¦R¶Z¤Rš¢`¶Z¥Œ´Öž ¡T ]©b¸Z¥]šÏ¸¨¢`«G¢`©bš‚´Gš@«T ]¿¢e´G ]¦R¬±¦`Â[´G™Zšã¥I¢`«T¤Ršqm¡T¾¢`¥]šÇš š@¾‚´G¡ð¦`Âƾ@¶Z©£¶Z¥]¶Z¡

¾@¦R¬EµRš@¾‚´G ]¦R¬6Ö ö Ît»ÛRÞdCÎt»dÎt»ÃÑn»êRÞEÝR×q½

ø)Že÷|–s†GˆE– †e– 

¢¡¤£¦¥¨§ ©  ¤   !"#$ %¤" & '()+* ,-/. "#0 '1$. "2§/34) ,5,768 "-'(* 9:" ¦;)+ ¦,5%¤" ,2<>=@?$'¤/A¤B An objective method to diagnose the contribution of steering flow at each level to tropical cyclone motion

Mitsuru Ueno Typhoon Research Department, Meteorological Research Institute e-mail: [email protected]

Tropical cyclone (TC) motion is one of the most important forecasting issues both in the numerical weather prediction (NWP) and operational forecasting. Even with the state-of-the-art NWP models significant disagreements can be sometimes observed in TC track forecast among them although it has been believed that TC movement is primarily determined by the large- scale flow field and therefore should be well predicted with such models. A number of numeri- cal studies demonstrate that the evolution of TC vortices is highly sensitive to the formulation of convective heating. So in this study our primary concern lies in (1) how significantly the envi- ronmental flow fields are affected by the formulation of convective heating, (2) whether or not asymmetric convective heating directly and significantly drives the TC vortices, and (3) how much the difference in vortex structure is responsible for the track difference. In order to clarify the role of convective heating in TC motion, we conducted an idealized experiments using an f-plane version of a numerical prediction model, in which a tropical cy- clone is embedded in a vertically sheared environment. A wide variety of TC motion shows up in the presence of the vertical shear, depending upon the formulation of convective heating (Ueno 2000). A preliminary analysis of the experimental results suggests that the difference not only in the steering flow but also in the axisymmetric structure of simulated TCs is responsible for the diversity of simulated TC motions. The latter's contribution can be defined by a set of weighting factors (referred to as steering weight) which is mathematically derived from surface pressure tendency equation and reflects upon the thermal structure of the individual vortex. v Letting the steering weight at k -th model layer W k and the steering flow at the same layer k , the steering component of three-dimensional TC motion v may be expressed as k max STR =≡∆∆σσ ∆ vvSTR ∑ wppkk kkk , srf ∆ k=1 where p k is thickness in pressure of k -th model layer and p srf is pressure at the sea surface. Figure 1 shows the vertical profile of the steering weight averaged over the integration period of 72 hours for two extreme track cases in Fig.1 of Ueno (2000), that is, B&M (experiment with Betts and Miller scheme) and KUR (experiment with Kurihara scheme). The profiles seem rea- sonable in that the weight is very small at the highest two layers near the tropopause, which is consistent with previous observational studies. Note that the two profiles significantly differ each other in the troposphere. Once we get the steering weight W k , the steering motion compo- nent v STR is easily computed using the steering flow v k which is defined as an areal-mean asymmetric flow near the storm center. Figure 2 compares the track reproduced from hourly

v STR with "actual" one which is obtained as successive MSLP center locations. To see the impact of steering weight on the reproduced track, the tracks obtained by using mismatched steering weight (i.e., KUR-weight for B&M and B&M-weight for KUR) are shown in the same figure. The result demonstrates that the long-term vortex motion is well reproduced by the proper combination of the steering weight and steering flow. An additional experiment shows that the steering weight is somewhat inherent to the convection scheme used and robust against the change of environment. Details of the discussions may be found in Ueno (2002).

0.00.0

0.5-0.5

B&M KUR vertical level in sigma

1.0-1.0 -1 0 1 2

steering weight Figure 1: Vertical profile of steering weight. Symbols are put at each model level.

300 (a) B&M-track (b) KUR-track

500

200

300

100 y (km) y (km)

100 0 "actual" "actual" B&M-wgt KUR-wgt KUR-wgt B&M-wgt

-100 -100 -300 -100 100 -100 0 100 200

x (km) x (km)

Figure 2: Vortex tracks reproduced from velocity components for (a) B&M and (b) KUR. The vortex is placed at (0,0) initially and symbols indicate hourly positions up to 72 h.

References Ueno, M., 2000: Impact of upper-level flow on the tropical cyclone motion in a vertically sheared ambient flow. WMO/TD-No.987, 5.39-40. Ueno, M., 2002: Steering weight concept and its application to tropical cyclones simulated in a vertical shear experiment. Submitted to J. Meteor. Soc. Japan.

¢¡¤£¦¥¨§ © © £ § ©¤§¡¤£ ¢¡¤£ "!$# &%'¤()*#+"£,.-/# , #+§

0

#1# !2£3£ "546#+§12§ (87:9. ;< §1=# ?>:©@£A3""B ! 9C§:,¤£

D E*FHGJI3KMLON6KQPSRTJG

UQVXWYV[Z\^]`_a]cb"db"ebcf gihjV3elkmf]XZh'n1oqp?brnb o`UVXWYV[Z

sut`v t`z{z`| vt€`‚ z ‚„ƒ ‚ {†¨‡

R*KMw R*xJxJy E[R*TJG~} I3K NJFHG F

ˆŠ‰Œ‹Žj‰"j‘“’j”•—–j˜š™œ›‘cžŸ‘O 5žŽj‰¢¡5£i¤"”a¥¦”Sj§+ ¤”&‘O 5¥¦`¡¨¤"”a›žŽ£S‰X©¥¦”a›žŽ£{‘'‹~¥¦cc”a‹Mª[ 5¤"”¦¡ 5›i‘O ¡5¨¤”{›5”@‘'‰"8 5¤”

5›j¨*¡5¨¤”{›5” £{‘'‰‰"' :«*” ›5”q‘O 5”q¬žŽ‰¬ 5¤”@¡5‘j¥¦” ¥&‘'‰‰”{›{­œ®¯¡¯‘j‰°‘'‹„ 5ž„ 5©"”=«*”{£)¥¦”{¡œ¤"ž„¤”a›¯ž„‰¬ ¤”S¡ 5›i‘O ¡5¨¤”{›5”ª

5¤"”±j¥¦ž„‰‘'‰` °‘'«"¡5j›¨c 5žŽj‰š‹„žŽ‰”±¨›'²"‹Ž”¢£i¤‘'‰”{¡?§³›5¥´‘µ– j›”a‰` 5¶¢‹Žž„‰”·¡5¤"‘'¨*”•—£)‹„‹ŽžŽ¡5žŽj‰"‘j‹u«"›5`‘jc”{‰ž„‰"`™¦ ¢‘

¸¯j¨¨‹Ž”a› ‹ŽžŽ‰”&¡¤"‘j¨”`•¹¸¯j¨¨"‹„”{› «›5`‘jc”{‰žŽ‰`™)­Sº¤”¦j¨*”a›i‘O ž„‰"‘'‹J‹„«"‘'‹~»œ˜/¼¨¥¦cc”a‹~j§ ½`‘'¨"‘j‰Œ¾¬”a 5”{j›j‹ŽjjžŸ£a‘j‹

®œ”a‰"£a¿[•¹½`¾°®QÀlÁ Âc¾°Ã"Ä{ÃjÅ`™JcX”{¡+‰' 1¡5©cÆÇ£)žŽ”a‰` ‹„¿@ ‘jÈj”œ‘£a£)©‰` +'§H ¤”:‹Ÿ‘O 5 5”a› ‘'«"¡5j›¨c ž„‰S l¿X¨*”j­~º¤”:£{‘'‹Ÿ£)©‹Ÿ‘O ž„‰

¥¦”) ¤cSj§" ¤” 5›i‘'‰"¡5¥¦žŽ¡¡5ž„‰=§³©‰£É 5žŽj‰¦žŸ¡J«"‘j¡5”{Sj‰S 5¤"”œ¡l i‘O 5žŸ¡ 5žŸ£a‘'‹c«‘'‰"S¥¦X”a‹Ê•—ÁuXcc¿jªŽÄ{Ë`ÌjÍj™)ª'Q¤”{›5”Q‘¤–3›5”{‰ ¶

‹Žž„‰”@¡¤‘'¨*” žŽ¡u‘j¡¡5©¥¦”{[­œº¤”@¸:¨¨‹Ž”a›¯”)ÎH”{£É užŽ¡¯¡žŽ¥¦¨‹Ž¿ÏžŽ‰"£a‹„©"”{¬«X¿Ð›5”a ‘jž„‰žŽ‰Ç‘¦²‰žÑ ”@‘j«"¡5j›¨c 5žŽj‰Ð©‰"”a›:‹ŽO

¨›”{¡¡©"›5”Q£)‰"cž„ 5žŽj‰"¡+Q¤”a›” 5¤"”œ¨©›”œ–3›5”{‰ ¶œ‘'«"¡5j›¨c ž„‰S j©‹Ÿ¦«*”œ‰j›¥&‘'‹Ž‹„¿=‰”a‹„žŽjžŽ«‹Ž”j­~Á Âc¾°ÃÄqÃjŤ¤"‘¡~j›”{‘'

”a››j›¦ž„‰ ”a¥¦¨*”a›i‘O 5©"›5”¬žŽ‰Ò 5¤”8¡ 5›i‘O ¡5¨¤”{›5žŸ£m›”až„‰¨•—¡¤"OQ‰Ò‹Ÿ‘O ”a›É™ÉªQ‘'‰ ¤žŸ¡&ž„‰¡©cÆÇ£až„”{‰ m 5›”{‘O ¥¦”a‰` Ǎ'§u 5¤”

¸¯j¨¨‹Ž”a›œ‘j«"¡5j›¨c 5žŽj‰Ïž„‰? 5¤"” –j˜Ó›‘cžŸ‘O 5žŽj‰Ï¡5£i¤"”a¥¦” žŽ¡ ¤j©"j¤`  5&«*” ‘j¡j‰"”uj§6ž„ ¡Q£{‘'©"¡5”j­

»œ”{Ô¨"‘'›i‘'¥¦”) ”a›ž„¶q‘O ž„‰¢žŸ¡S«”{ž„‰¢c”a’”a‹Žj¨*”{±žŽ‰›c”{› 5Œž„‰` ›5cc©"£a”& 5¤"”?¸¯j¨¨‹Ž”a›¦‘'«¡›5¨c ž„‰·¨›j¨*”a›‹„¿­

ˆÊ žŽ¡u«"‘¡”q°‰° 5¤"”¦–˜Õ›i‘jcžŸ‘O ž„‰8¡£i¤”a¥¦”@'§Y ¤”&»:®¯Â®ÖÁuc"‘'›iŒÂX¨"‘£)”¦×6‹„žŽj¤` SØ1”a‰` 5”{›@•ÊØ1¤j©¢”) ‘j‹—­Žª ÍjÃjÃ`Íj™

‘'‰° ¤”Ǹ:¨¨‹Ž”a› ”aÎ*”q£É žŸ¡ £)j‰¡žŸc”a›”{Œ«X¿°©"¡5ž„‰"? ¤”@ 5›i‘'‰"¡5¥¦žÑ 5 ‘j‰"£)”@ ‘j«‹Ž”{¡¤¨›”{£aj¥¦¨©c ”{ŒQžÑ ¤Œ ¤”Ç‘j£{£)©c i‘O ”

‹Žž„‰”aÀM«X¿`Àʋ„žŽ‰”•—– Ù–™ £a©‹Ž£a©‹Ÿ‘O 5žŽj‰Ð£)‰"¡5žŽc”{›5žŽ‰S 5¤” ¸¯j¨"¨‹„”{›Q«›‘j”a‰žŽ‰"­~º›‘j‰"¡5¥@žŸ¡¡žŽj‰?'§6‘j‰Ï‘' 5¥¦¡5¨¤”{›5žŸ£¯¨"‘O ¤

£a‘j‰Œ«*”Ç£a‘j‹Ž£a©‹Ÿ‘O 5”q8«X¿°›”)ÎH”a›žŽ‰m Ð 5¤”q¡”@ i‘'«‹Ž”{¡¤Qž„ 5¤·‘'‰¢”)ÎH”{£) 5žŽ’j”&¨›”{¡¡5©›5”¦‘'‰° ”a¥¦¨”{›‘' 5©›”&£)¥@¨"©c 5”qŒ«X¿

 ”ažŽj¤` 5žŽ‰@ 5¤” ‘j«"¡5j›«”{›‘j¥¦j©‰` œ‘'‹Žj‰¦‘@¨"‘O ¤·•—Í“Àʨ"‘j›‘j¥@”a 5”{›¡£a‘j‹„žŽ‰@¥¦”) 5¤"X™É­

Ø1Xj‹ŽžŽ‰8›i‘O ”{¡ «X¿Ú–˜Û›‘cžŸ‘O 5žŽj‰µ‘'›”?£)¥¦¨©c 5”q·«X¿± 5¤žŸ¡@‰”aÔ¨‘'›i‘'¥¦”) ”a›ž„¶q‘O 5žŽj‰Ú‘j‰"± ¤”?‰”Ï©"¡”q·žŽ‰

Á Âc¾°Ã"Ä{ÁÅ­JמŽ"­ŽÄ¯‘j‰"?Í=›”{¡5¨”q£É 5žŽ’j”{‹„¿&¡5¤O 5¤"”u›”{¡5©‹„ ¡+§³j› 5¤"”¯¥¦žŸXÀM‹Ÿ‘O žÑ ©"c” ¡5©¥¦¥@”{›Q‘'‰"Ç 5¤”¯ 5›j¨žŸ£a‘j‹*‘' 5¥¦'À

¡5¨¤”a›”¯ ‘'ȁ”a‰?§³›5¥Ü¾°£qØ1‹Ÿ‘O i£i¤”a¿&”a :‘'‹M­Ž•Ä{ËXÝ'Íj™)ªQ¤"žŽ£i¤Ð‘'›”u”a’XžŽ”{?ž„‰` 5@ށÃ'ÀM‹Ÿ‘“¿j”{›¡+'§~Á Âc¾°ÃÄqÃjÅ=ž„‰Ð£a‘j‹Ž£a©‹Ž‘' 5žŽj‰3­

º¤”Q²j©›”{¡JžŽ‰"cžŸ£a‘' 5”Q ¤"‘O ¯•Äq™6 ¤”:£a`‹„žŽ‰ žŽ‰¦ 5¤”:©¨¨*”a›1¡ 5›i‘O ¡5¨¤”{›5”žŸ¡YžŽ‰"£)›”{‘¡”q=«*”{£{‘'©"¡5”Q'§[žŽ‰ ›5cc©£)žŽ‰¤ 5¤”

¸¯j¨¨‹Ž”a›Q‘j«"¡›5¨ 5žŽj‰Ç¨›j¨*”a›‹Ž¿jª[•—́™Y ¤”¤£)Xj‹Žž„‰"SžŽ‰m ¤”u¥@žŸc‹Ž” ¡l ›‘' 5`¡¨¤"”a›”:žŸ¡ ”{£)›”{‘¡”q[ªc‘'‰"Œ•¹Å`™Y ¤” £)Xj‹ŽžŽ‰

žŽ‰= ¤” 5›j¨*¡5¨¤”a›” žŽ¡~‘j‹Ž¡5:žŽ¥¦¨›5O’”{=ž„‰S 5”{›5¥&¡J'§«”a  5”{›+‘'›5”{”a¥¦”a‰` 6QžÑ ¤S ¤”Q– Ù–?›5”q¡©‹„ ¡{­6ˆÊ JžŸ¡~¡5©¨¨*¡5”{ ¤"‘O

•—́™YžŸ¡c©”¯ 5S 5¤”užŽ‰"£)›”{‘¡”¯'§3 ¤”¯©"¨`‘'›i¦ß"©àǞ„‰m 5¤” ¡ 5›i‘O 5`¡¨"¤”a›”œQž„ 5¤? 5¤”¤£i¤"‘'‰"j”:žŽ‰Ç ¤”¯ 5›j¨*¡5¨¤”a›”•5•¹Å`™™É­

»:”)àX {ªc ¤žŸ¡œ¨"‘j›‘j¥@”a 5”{›5žŽ¶{‘' 5žŽj‰mžŸ¡Q”aàc‘j¥¦ž„‰”qm«X¿m 5¤” ‰”¤¥¦j‰` 5¤Ï§³j›”{£{‘j¡ Q”)àc¨”{›5žŽ¥¦”a‰` œ©"¡5ž„‰"Ç½¾°®=á ¡œj‹Žj«"‘j‹

¥¦cc”a‹M­8×6ž„­ Å8¡¤Oœ¡ 5¤”?j‰”m¥@‰` 5¤·¥¦”{‘j‰Ú”a››5› ž„‰¶aj‰‘'‹„ÀM¥¦”{‘j‰¢ 5”a¥¦¨*”a›i‘O ©›5”?ž„‰Ú½©‹Ž¿ÚÍ'ÃjÃ"Äj­¬º¤”Ç›”a’XžŸ¡”q

’j”{›¡5žŽj‰@žŽ¡J ¤”:¡‘'¥¦”œ‘j¡6 ¤”¯Á Âc¾8ÃÄ{ÁÅu”)à£)”{¨c Y§³j›J ¤”œ¨"‘j›‘j¥¦”) 5”{›5žŽ¶{‘' 5žŽj‰@'§* 5¤”¯¸:¨¨‹Ž”a›Y”aÎ*”q£É +žŽ‰@ 5¤”:–˜â›‘cž„À

‘O ž„‰m¡£i¤”a¥¦”j­Jº¤”¯j›”{‘O £)j‰` ›‘¡l +«*”) l ”a”{‰&¨`¡ž„ 5žŽ’j”¯”a››5›q•—‘'«*O’j” 5¤"”¯ÅOÀʤ"¼~‘¤‹Ž”a’”a‹ã™Y‘'‰&‰”a`‘O ž„’”œ”a››5›q•¹«”{‹„O

ž„ i™Y£aj¥¦¨©c ”{¦«X¿S 5¤”œ›5žŽjžŽ‰"‘j‹Á Âc¾°Ã"Ä{ÃjÅužŸ¡Y”{‹„žŽ¥¦ž„‰"‘' 5”q[ª`‘j‰"@ 5¤"”:¥&‘'‰žÑ ©"c”œj§H 5¤"‘' +‰”a`‘O ž„’”Q”a››5›~žŽ¡+›5”qc©"£)”q

žŽ‰¬ 5¤”S‰”aä’”a›i¡žŽj‰ ­º¤"žŽ¡:ž„¥¦¨"‘£É ¯žŸ¡u£)j‰¡žŸ¡l ”a‰` :Qž„ 5¤¬ ¤”S‘j«O’”)ÀÊ¥¦”a‰` 5žŽj‰”q±•lÄq™œ‘'‰"Ú•MÍj™)­œ˜¢¤žŽ‹„”= 5¤”S”a››5›œžŽ¡

žŽ‰"£)›”{‘¡”qÇžŽ‰Ï¡5j¥¦”u›5”{jžŽj‰"¡1'§ 5¤”u 5›j¨*¡5¨¤”{›5”¯«X¿& 5¤”¤›”)ß"”q£É ž„‰m'§ 5¤” ‘j«O’”S•¹Å`™ÉªX ¤žŽ¡žŽ¥¦¨‹Žž„”q¡ 5¤”¤‰”q£)”{¡¡5žÑ l¿

5Sž„¥¦¨›O’j” ¤”:j 5¤”{›+¨¤X¿c¡5žŽ£{‘'‹"¨"›5c£)”q¡5¡5”{¡+ž„‰Ç 5¤”: 5›j¨*¡5¨¤”a›”jªž„‰£)‹Ž©"cžŽ‰S£a‹„©"XÀʛ‘cžŽ‘' 5žŽj‰¦žŽ‰ ”a›i‘j£) 5žŽj‰&¨›c£)”{¡¡{­

åQ”a§³”a›”a‰"£a”{¡

Ø1¤j©3ª ¾·­ ¸=­Žª ¾·­ ½­æÂX©"‘'›”a¶ª çS­æèJ­ – žŽ‘j‰@‘j‰"Ï¾·­ ¾Ú­ ÀŠé ­ ê+‘'‰3ªæÍ'ÁÁÍë6®¨ ¤”a›¥&‘'‹[žŽ‰c§³›‘j›5”qm›i‘jžŽ‘' 5žŽj‰m¨‘'›i‘'¥¦”) ”a›5À

žŽ¶{‘' 5žŽj‰m§³j›Q‘O ¥@`¡¨"¤”a›žŽ£¯¡l ©"cžŽ”{¡{ªº ”q£i¤‰žŸ£a‘j‹›”a¨*j›5 ¡5”a›žŽ”{¡1j‰m‹„«"‘'‹*¥¦cc”a‹Žž„‰"¦‘j‰"m"‘O ‘@‘j¡¡5ž„¥¦žŽ‹Ž‘' 5žŽj‰

ì~‹„©¥¦”¦ícî3ª »:®uÂc®¦Áuc‘j›Ï¡¨‘j£)”¯ß"žŽj¤` :£)”{‰` 5”a›q­

Áu`cc¿ª å ­ ¾·­ŽªŽÄ{ˁ́ÍcëJ®ä¡l i‘O žŽ¡ 5žŸ£a‘j‹[¥@cc”{‹[§³j›Q‘O ”a›’O‘'¨*j©›œ‘j«"¡5j›¨c 5žŽj‰3ªïS­ ½­ åQO¿­ ¾°”) {­æÂXc£'ªñðò3ªŽÄ{ó`̓ÀiÄ{ójË"­

¾°£qØ1‹Ÿ‘O i£i¤”a¿ª å ­ ® ­Žª å ­ ˜ä­ ×”{‰‰3ª ½"­ ôœ­ ® ­æÂX”{‹„«X¿jª × ­ ô:­ ì~‹„¶ ‘j‰"Ð½"­æÂH­æÁ¤‘'›ž„‰ª„ÄqË`ÝjÍcëJõ¯¨ 5žŸ£a‘'‹[¨"›5¨”{› ž„”q¡1j§ ¤”

‘' 5¥¦¡5¨¤”{›5”ªXÅj›?”{3­„ª ®:×JØ åœ–·ôY‰X’Xž„›‰3­ åQ”{¡{­ ¼~‘'¨*”a›i¡1»:"­ ö í*íªŽÄ{Á÷'¨¨ ­ Heating Rate by Longwave Radiation (McClatchy MLS, clear sky ,375co2ppm)

New Current 1

10 p[hPa]

100

1000 -12 -10 -8 -6 -4 -2 0

[K/day]

×6ž„­„āë~º¤”=¤”q‘O ž„‰?›‘' 5” «X¿Ð–j˜ø›i‘jžŽ‘' 5žŽj‰°£)¥¦¨©c 5”qÐ§³j›¤‘Ç£a‹„”q‘'›:¥¦žŸXÀʋŽ‘' 5ž„ 5©"c”@¡©"¥@¥¦”{›u‘' 5¥¦¡5¨¤”{›5” «X¿? 5¤” ‰”a¨¨"‘'›i‘'¥¦”a 5”a›žŽ¶{‘O ž„‰ •¹¡5j‹ŽžŸ&‹„žŽ‰”“™‘'‰"? 5¤"” j‰”¤©"¡5”{ÏžŽ‰ÐÁ Âc¾8ÃÄ{ÁÅ"•—‘j¡5¤”{m‹„žŽ‰”“™

Heating Rate by Longwave Radiation (McClatchy TRP, clear sky ,375co2ppm)

New Current 1

10 p[hPa]

100

1000 -12 -10 -8 -6 -4 -2 0

[K/day]

מŽ"­æÍcëYÂc‘j¥¦” ‘j¡Q×6ž„­„Ĥ”aàc£a”a¨c §³›œ‘&£)‹Ž”{‘'› ›5¨žŽ£{‘'‹3‘' 5¥¦¡5¨¤”{›5”

×6ž„­ Å"ë~º¤”S‰”)ÀÊ¥¦j‰` 5¤¬¥¦”q‘'‰¬§³›5”q£a‘j¡ :”{›5›j›¯'§Y¶a‰"‘'‹„ÀÊ¥@”q‘'‰Ð ”a¥¦¨”{›‘' 5©›”=žŽ‰8½©‹Ž¿¬ÍjÃjÃ"Äj­:º¤”@£)©››”a‰` u½`¾°®QÀ Á Âc¾°ÃÄqÃjÅ•³›ž„¤ ə+‘j‰"? 5¤” ›”a’XžŸ¡”qÇ’”a›i¡žŽj‰?©"¡žŽ‰¦ ¤” ‰”a¨"‘'›i‘'¥¦”a 5”a›žŽ¶{‘O ž„‰ •³‹Ž”)§ã əɭ