Retrieval of effective ice crystal size and IWP of large-scale semi-transparent cirrus & their correlations with the state of the atmosphere Gaby Rädel Claudia Stubenrauch, Fadoua Eddounia Laboratoire de Météorologie M Dynamique Ecole Polytechnique, France Gaby Rädel CIRAMOSA, Paris, 20 Nov 2003 1
Overview Cirrus from TOVS Path-B: Evaluation of TOVS cloud height using LITE retrieval of effective ice crystal size, D e, and IWP from TOVS & sensitivity study correlation of De and IWP with atmospheric parameters from ERA-40 analyses horizontal extend of cirrus Gaby Rädel CIRAMOSA, Paris, 20 Nov 2003 2
High cloud frequency from TOVS Path B 1987-1995 p cld < 440 hpa boreal winter boreal summer ITCZ moves into summer hemisphere more cirrus over land more cirrus over ocean Gaby Rädel CIRAMOSA, Paris, 20 Nov 2003 3
Evaluation of TOVS cloud height with LITE (newest LITE inversion by L. Sauvage) measured vertical spectrum Lidar In-space Technology Experiment backscatter lidar on Discovery 10 19 Sep 1994 resolutions: vertical: 15m horizontal: 740m z t LITE: low clouds high clouds z top -z base : 1.3 km 2.7 km Gaby Rädel CIRAMOSA, Paris, 20 Nov 2003 4
Colocation of LITE and TOVS Path-B data 1 1 Spot LITE Spot HIRS t +/- 3 hours : 2750 events ~ after cuts: 1300 cloudy events t [h] Gaby Rädel CIRAMOSA, Paris, 20 Nov 2003 5
Evaluation of TOVS cloud height with LITE RESULTS h[m] h[m] LITE TOVS LITE TOVS signal signal Gaby Rädel CIRAMOSA, Paris, 20 Nov 2003 6
Evaluation of TOVS cloud height with LITE RESULTS 796 TOVS low clouds 495 TOVS high clouds 560 LITE single- 236 multi-layer 161 LITE single- 334 multi-layer very thin LITE high clouds z TOVS 0.5(z top +z base ) LITE (km) single layer: z <1 km: 70% peak at 0 heterogenous scenes z TOVS 0.5(z top +z base ) LITE (km) z < 2 km : 60 % peak at 0.5 km P cld (TOVS) p cld (mid-cloud) better agreement for low large-scale cirrus clouds Gaby Rädel CIRAMOSA, Paris, 20 Nov 2003 7
Effective ice crystal size (D e ) and IWP retrieval of semi-transparent cirrus based on: spectral difference of cirrus emissivities at 11-8 µm Observations: NOAA10 1987 1991, 60 N 60 S, θ v < 25 large-scale cirrus: 1 x1 overcast, p cld <440 hpa T cld < 263K,T meas B (8µm), T meas B (11µm), T cld, T surf ε surf (SARB), closest TIGR H 2 O/T profiles 3R radiative transfer (ε 8µm,ε 11µm ) Method: simulate ε(λ,d e,iwp,θ) hom. cloud, β abs (D e ),<ω 0 (D e )>,<g(d e )> planar polycrystals (mod. ADA) bimodal size distribution radiative transfer (Streamer (J.Key)) vary De, IWP ε(λ,d e,iwp,θ) Gaby Rädel CIRAMOSA, Paris, 20 Nov 2003 8
D e and IWP retrieval (cont.) produce look-up tables: D e = f(ε 8µm,ε 11µm ), IWP =f (D e,ε 11µm ) ε 11 ε 11 ε 8 ε 11 D e IWP for 0.3 < ε 11µm < 0.85 0.7 < τ VIS < 3.8 sensitivity up to D e 80µm Gaby Rädel CIRAMOSA, Paris, 20 Nov 2003 9 IWP
Retrieval of effective ice crystal diameter D e NOAA10 global average : <D e > = 55µm <IWP>= 30 g/m 2 December 1990 D e [µm] Gaby Rädel CIRAMOSA, Paris, 20 Nov 2003 10
D e distributions under different ice crystal shape assumptions median D e : 34.5/37.5/27.5 36.5/39.5/27.5 51.5/55.5/38.5 µm NH midlatitude tropics SH midlatitude D e (tropics) > D e (midlat) D e ( ) > D e ( ) > D e ( ) Gaby Rädel CIRAMOSA, Paris, 20 Nov 2003 11
Sensitivity of (ε 11 -ε 8 ) to atmospheric profile (PhD thesis R. Holz, 2001) closest TIGR profile - profiles within σ < 20% (ε 11 -ε 8 ) σ (ε 11 -ε 8 ) * tropics midlat. ε 11 ε 11 no bias for ε 11 >0.3 noise Gaby Rädel CIRAMOSA, Paris, 20 Nov 2003 12
Uncertainties on ice crystal size retrieval Reference: homogeneous cloud, T cld =237K, z=1km,t surf =300K, polycrystals, bimodal-γ size distribution parameter changed to/by (D e standard -D e )/D e standard crystal shape size distribution horiz.heterogeneity vert. heterogeneity + water cloud lapse rat h cld, T surf, z spheres hexagonal columns tropical 90% cloud cover D e low =2 D e high r e =7µm;LWC=0.2g/m 3 6.5 -> 11 /km +4km, -15K, +1km <De> 15µm 60µm 26 µm 36 µm 54 µm 68 µm 80 µm 90 µm 108 µm +15% +15% -20% thin cirrus 15µm 60µm +7% +2% -15% 20% +25% +25% +3% +3% +30% +25% -28% thick cirrus 15µm 60µm +25% +12% -15% 20% +15% +15% +5% +5% -6% to +10% Gaby Rädel CIRAMOSA, Paris, 20 Nov 2003 13
Long term survey of cirrus properties 1987-1991 D e [µm] effective diameter Nε [%] effective emissivity IWP [g/m 2 ] ice water path Gaby Rädel CIRAMOSA, Paris, 20 Nov 2003 14
Regional and seasonal variations of D e and IWP TOVS NOAA10 3-year averages eff. emissivity (%) 75 70 65 60 55 Nε IWP D e win spg sum aut IWP (g/m2) Nε(SHm) > Nε(NHm) > Nε(trop) NH land: Nε(sum) < Nε(win) 35 30 25 20 15 win spg sum aut eff. Diameter (micron) 65 60 55 50 45 40 35 30 win spg sum aut NH ocean NH land trop. Ocean trop. Land SH ocean SH land NH z=2km trop. z=2km SH z=2km IWP(trop) > IWP(NHm) > IWP(SHm) land: IWP(midsum) > IWP(midwin) D e (trop) > D e (NHm) > D e (SHm) land: D e (midsum) > D e (midwin) Gaby Rädel CIRAMOSA, Paris, 20 Nov 2003 15
D e and IWP as function of cloud temperature D e De (micron) 90 80 70 60 50 40 Large-scale semi-transparent cirrus 60 N 60 S 7,5 22,5 37,5 all different IWP 55 IWP IWP (g/m2) 45 35 25 30 20 15 different D e 32,5 47,5 62,5 82,5 all 10 200 210 220 230 240 250 260 270 cold cirrus: Tcld (K) Gaby Rädel CIRAMOSA, Paris, 20 Nov 2003 16 5 200 210 220 230 240 250 260 270 Tcld (K) D e depends more on IWP than on T cld IWP increases with T cld
Regional dependence for thin and thick Cirrus D e 90 80 70 thin cirrus thick cirrus midlatitudes thin cirrus thick cirrus midlatitudes 90 80 70 midlatitudes 55 midlatitudes 32,5 47,5 62,5 82,5 all 7,5 22,5 37,5 52,5 all 45 45 55 midlatitudes IWP De (micron) 60 50 40 De (micron) 60 50 40 IWP (g/m2) 35 25 IWP (g/m2) 35 25 30 20 7,5 22,5 37,5 all 10 200 210 220 230 240 250 260 270 Tcld (K) 90 tropics 80 70 30 20 10 200 210 220 230 240 250 260 270 Tcld (K) 90 tropics 80 70 ->Tcld tropics 15 5 200 210 220 230 240 250 260 270 55 45 Tcld (K) tropics 32,5 47,5 62,5 82,5 all 15 5 32,5 47,5 62,5 82,5 all 200 210 220 230 240 250 260 Tcld (K) 55 tropics 45 270 -> Tcld De (micron) 60 50 40 30 De (micron) 60 50 40 30 15 20 20 7,5 22,5 37,5 all 7,5 22,5 37,5 52,5 all 10 10 5 200 210 220 230 240 250 260 270 200 210 220 230 240 250 260 270 ->Tcld 200 210 220 230 240 250 260 270 Tcld (K) Tcld (K) Tcld (K) different behaviour in midlatitudes and tropics Gaby Rädel CIRAMOSA, Paris, 20 Nov 2003 17 IWP (g/m2) 35 25 IWP (g/m2) 35 25 15 32,5 47,5 62,5 82,5 5 200 210 220 230 240 250 260 thick Ci in tropics: D e and IWP do not depend strongly on T, almost no scatter due to different IWP or D e Tcld (K) all 270 -> Tcld
Atmospheric properties accompanying large-scale cirrus ERA-40 ECMWF reanalyses: humidity, U, V and W for 23 pressure levels every 6 hours, 1.125 x 1.125 spatial resolution Co-location with TOVS observations: (1989, 1990) Water vapour (cm) mean RMS tropics: largest water vapour, smallest winds midlat. winter: strongest winds Horizontal wind (m/s) mean RMS Frequency of situations with strong strong updraft no wind downdraft NH midlatitude summer 3.0 1.2 14.5 10.9 9% 38% 3% NH midlatitude winter 1.4 0.8 26.1 15.8 13% 29% 7% tropics 5.0 0.9 7.6 6.0 7% 44% 0.1% SH midlatitude summer 2.3 1.0 23.4 13.8 6% 42% 4% SH midlatitude winter 1.5 0.8 22.3 15.2 10% 34% 4% SH: horizontal winds always strong most large-scale semi-transparent cirrus in situations with no vertical wind Gaby Rädel CIRAMOSA, Paris, 20 Nov 2003 18
Vertical updraft for thin and thick cirrus NH midlatitude summer frequency of strong updraft ε<0.55 ε>0.55 6% 11% NH midlatitude winter tropics SH midlatitude summer SH midlatitude winter 9% 7% 4% 7% 16% 9% 7% 11% thick cirrus more often in regions with strong vertical updraft Gaby Rädel CIRAMOSA, Paris, 20 Nov 2003 19
D e and IWP as function of humidity and wind Large-scale semi-transparent cirrus 60 N 60 S, T cld < 233K Stubenrauch et al. 2003, submitted to Atmos.Res. D e and IWP increase with water vapour D e 12 µm smaller in case of strong winds IWP 10 gm -2 larger in case of strong vertical updraft Gaby Rädel CIRAMOSA, Paris, 20 Nov 2003 20
Regional distributions of D e and IWP as function of humidity and wind dry, strong updraft dry, strong hor.wind D e humid, strong updraft humid, strong hor. wind IWP D e larger in case with no winds than strong winds humid tropics: IWP larger in case of strong w than strong u+v Gaby Rädel CIRAMOSA, Paris, 20 Nov 2003 21
Cirrus horizontal extent determine horizontal extent of cirrus clouds (ε > 0.3): a. empty boxes are filled with most likely information on cirrus type b. simple clustering algorithm groups adjacent boxes containing ), cirrus (0.95>ε>0.5 ) deep convection (ε>0.95 or thin cirrus (0.5>ε>0.3 ) Examples: 18/07/1989 and 30/12/1989 7h30 PM very thin Ci clear sky mid and low clouds thin Ci Ci deep conv. Gaby Rädel CIRAMOSA, Paris, 20 Nov 2003 22
Cirrus horizontal extent cirrus clusters: largest in tropics smallest in ML summer D e as fct. of distance to convective centre: D e small if very close to convective centre and in smaller clusters -> dynamics? (1 box = 100 x 100 km) small clusters (<20boxes) large clusters (>500boxes) Gaby Rädel CIRAMOSA, Paris, 20 Nov 2003 23
Conclusions and Outlook p cld (TOVS) p(mid-cloud) D e and IWP can be retrieved on a global scale for large-scale semi-transparent cirrus: <D e >=55µm <IWP>=30g/m 2 IWP increases with T cld - D e increases with IWP TOVS Path-B & ECMWF reanalyses D e and IWP increase with atmospheric water vapour, increase depends on vertical updraft, hor. wind, formation processes? Find parameterizations IWP =f (q,w,t), D e =f (IWP,q,w,u+v,T) using also cluster information Gaby Rädel CIRAMOSA, Paris, 20 Nov 2003 24