Alexander Stickler 1, A. M. Fischer 2, S. Brönnimann 1. Oeschger Centre for Climate Change Research University of Bern, Switzerland

Transcription

1 11 th EMS Annual Meeting, September 2011, Berlin, Germany Vertical structure of 20th century temperature trends in a GCM run, reanalyses, statistical reconstructions and observations Alexander Stickler 1, A. M. Fischer 2, S. Brönnimann 1 1 Oeschger Centre for Climate Change Research University of Bern, Switzerland 2 Federal Office of Meteorology and Climatology MeteoSwiss, Zürich, Switzerland

2 Outline > Motivation > Data and methods > Results > Conclusions 2

3 Motivation > Vertical structure of T trends attribution of trends to anthropogenic influence > Theory, models, observations: amplified warming in upper tropical/subtropical troposphere (Santer et al. 2005, Haimberger et al. 2008, Allen & Sherwood 2008 (from thermal wind balance), Sherwood et al. 2008) lower polar troposphere (e.g. Vinnikov et al. 1980, Manabe & Stouffer 1980, Holland & Bitz 2003, Serreze & Francis 2006, Serreze et al. 2009, Bekryaev et al. 2010) 3

4 Motivation > Amplified warming in tropical upper troposphere should result from enhanced convective latent heat release in the Inner Tropics increased moist static stability due to increased specific humidity in the Subtropics in a warmer climate has not been detected in many radiosonde/satellite/reanalyses datasets (e.g. Santer et al. 2005, Karl et al. 2006), but this has been ascribed to artificial biases (Fu et al. 2004, Fu & Johanson 2005, Mears & Wentz 2005, Sherwood et al. 2005, Randel & Wu 2006, Santer et al. 2008) in case of radiosondes/satellites 4

5 Motivation > Polar amplification: several feedback mechanisms such as ice-albedo feedback (mainly autumn-early winter, incl. radiation and insulation effect, modulated by ice thickness) feedbacks involving circulation (incl. ocean & polar vortex, mainly winter-spring) and/or clouds/humidity/aerosols (shortwave/longwave radiation, summer/year-round) snow-albedo feedback (land surfaces, mainly spring) Arctic T trends should be studied at a seasonal resolution 5

6 Motivation > Reanalyses have given diverging results: polar amplification aloft in ERA-40 (zonally averaged reanalysis data, Graversen et al. 2008, study heavily criticised) vs. surface polar amplification in ERA-Interim (Screen & Simmonds 2010) reanalysis data quality needs to be assessed direct comparison of reanalyses at locations of observations besides zonal averages: regional averages from these extracted profiles to characterise spatial variability 6

7 Data and methods > 20 th century forced model run: SOCOL (Fischer et al. 2008) CCM with high top appropriate for global simulations including the stratospheric chemico-dynamical feedbacks 3.75 x levels up to 0.01 hpa, ens. mean (8 members) > Reanalyses: ERA-40 (Uppala et al. 2005), NCEP/NCAR (NNR, Kistler et al. 2001), Twentieth Century (20CR, Compo et al. 2011) 2.5 x2.5 (ERA-40,NNR), 2 x2 (20CR) 17 (NNR), 23 (ERA-40) and 24 levels (20CR) up to 10 (NNR, 20CR) and 1 (ERA-40) hpa > Statistical reconstruction (REC1, Griesser et al. 2010), 2.5 x2.5 > Calculation of vertical long term trend profiles: For the full time span of each dataset 7

8 Data and methods > Observational data (reference for regional mean intercomparisons): CHUAN (Stickler et al. 2010) Comprehensive, global upper-air station data back to early 20 th century Monthly mean version of data validated against ERA-40 and corrected for lag and radiation errors First radiosondes back to 1930s, but earliest of studied 20-yr periods with sufficient coverage for trend calculation starts 1951, exceptions: Alaska (1941) and E Siberia (1961) > Calculation of regional mean bidecadal linear trends: Regions were formed by selecting min. 3 stations in contiguous areas that show similar bidecadal variability in observations For each year, seasonal means are required to have all 3 monthly means available > 13 out of 20 years with data, no gap in the first and last 2 years Model and reanalyses data extracted at locations of regional stations 8

9 Results long term trend profiles Tropics K/month SOCOL ERA-40 NNR 20CR REC1 Note different periods! Tropical upper tropospheric trend amplification (all seasons) in - SOCOL: hpa - ERA-40: sharper 200 hpa - NNR: higher 100 hpa, partly weaker than surface hpa - 20CR: weaker than SOCOL - 2 nd 850 hpa in ERA-40 and NNR - no amplification in REC1 (only ) 9

10 Results long term trend profiles Arctic K/month SOCOL ERA-40 NNR 20CR REC1 Note different periods! All datasets except REC1 show lower tropospheric trend amplification - SOCOL: only in DJF/MAM - ERA-40: all seasons except JJA, strongest surface warming in MAM - NNR: only in SON/DJF - 20CR: all seasons, >> trends than SOCOL and even ERA-40/NNR in lower troposphere/stratosphere - largest differences between ERA-40 and NNR in DJF 10

11 Results long term trend profiles Arctic K/month Upper tropospheric amplification in JJA in NNR Negative trends in free troposphere in ERA-40 / NNR/20CR SOCOL ERA-40 NNR 20CR REC1 Note different periods! All datasets except REC1 show lower tropospheric trend amplification - SOCOL: only in DJF/MAM - ERA-40: all seasons except JJA, strongest surface warming in MAM - NNR: only in SON/DJF - 20CR: all seasons, >> trends than SOCOL and even ERA-40/NNR in lower troposphere/stratosphere - largest differences between ERA-40 and NNR in DJF 11

12 Results zonal mean bidecadal trends Example: DJF Tropics ERA-40 SOCOL K/dec NNR REC1 20CR 12

13 Results zonal mean bidecadal trends Example: DJF Tropics Even ERA-40 and NNR trends on zonal level show large differences in vertical structure and even sign of trends SOCOL ERA-40 K/dec NNR REC1 20CR 13

14 Results zonal mean bidecadal trends Example: DJF Tropics ERA-40 SOCOL K/dec REC1 NNR Much less vertical structure in 20CR cf. NNR/ERA-40 20CR 14

15 Results zonal mean bidecadal trends Example: DJF Tropics ERA-40 SOCOL Strongest variability/warming in upper troposphere K/dec NNR REC1 20CR 15

16 Results zonal mean bidecadal trends Example: DJF Tropics ERA-40 SOCOL Secondary maxima Emerging global warming, clearer in all other seasons K/dec NNR REC1 20CR 16

17 Results zonal mean bidecadal trends Example: DJF Arctic Note different scale! SOCOL K/dec ERA-40 General vertical structure agrees (also SON, but not MAM/JJA) NNR REC1 20CR 17

18 Results zonal mean bidecadal trends Example: DJF Arctic Note different scale! ERA-40 SOCOL Emerging global warming (all seasons) K/dec NNR REC1 Relatively strong variability, weaker in other seasons in REC1 20CR 18

19 Results regional mean bidecadal trends CHUAN radiosonde stations and regions Tropics Americas Africa Asia Pacific 19

20 Results regional mean bidecadal trends CHUAN radiosonde stations and regions Arctic Novaya Zemlya NE Central Siberia Bering Strait E Alaska Siberia Canada NW E Atlantic Karelia W E Central SW Central Siberia Siberia Siberia Canada SE 20

21 Results general agreement of regional mean trends with observations > Tropics ERA-40: relatively good agreement with CHUAN trends incl. vertical structure, best in Americas sector, less so in Asia and Pacific NNR: agreement with CHUAN is significantly worse than that of ERA-40 for all regions, especially w.r.t. the vertical structure of trends 20CR: temporal behaviour and vertical structure of trends often different from CHUAN, ERA-40 and NNR, especially around the tropopause, and in all regions 21

22 Results general agreement of regional mean trends with observations > Arctic ERA-40: fairly good agreement with CHUAN trends incl. vertical structure Best for NE Atlantic, Karelia, SE Canada, Alaska, Central Siberia Less good in NW Canada, E Siberia, W Siberia, Nov. Zemlya NNR: agreement with CHUAN is significantly worse than that of ERA- 40 for NE Atlantic, SE Canada, SW Central Siberia, E Central Siberia For Karelia, NW Canada, E Siberia, Alaska, Nov. Zemlya, W Siberia, SW Central Siberia, results are equally close to CHUAN than in ERA-40 20CR: some regions show much better agreement with CHUAN compared to Tropics, but less good than ERA-40 and NNR, no good agreement for Karelia, SE Canada, Nov. Zemlya, W Siberia, Central Siberia 22

23 Conclusions > Tropics: upper tropospheric T trend amplification found in all datasets except REC1, but traditional reanalyses and obs. show second maximum in lower troposphere > Arctic: polar lower tropospheric amplification found in all datasets except REC1, differences between regions and seasons deserve closer attention > Comparison with CHUAN radiosonde observations: ERA-40 best of considered reanalyses for trend studies on decadal and longer time scales in Tropics and Arctic (all subregions) NNR less well suited for many regions in Arctic and generally for the Tropics 20-CR seems not suited for such trend studies, even more so in the first half of the 20 th century > Primary observational data should be used whenever the quality of reanalyses for the use in certain research questions is unclear > Good agreement with observations does not imply a reanalysis reliability in less well observed times/regions (c.f. Graversen et al.) 23

24 References Allen RJ and Sherwood SC, 2008: Nature Geoscience, 1, Bekryaev RV et al., 2010: J. Climate, 23, Compo GP et al., 2011: Q. J. R. Met. Soc., 137,1-28. Fischer AM et al., 2008: Atmos. Chem. Phys., 8, Fu Q and Johanson CM, 2005: Geophys. Res. Lett., 32, doi: /2004gl Fu Q et al., 2004: Nature, 429, Graversen RG et al., 2008: Nature, 541, Griesser T et al., 2010: J. Climate, 23, Haimberger L et al., 2008: J. Climate, 21, Holland MM and Bitz CM, 2003: Climate Dyn., 21, Karl TR et al., 2006: Temperature Trends in the Lower Atmosphere: Steps for Understanding and Reconciling Differences, report by the Climate Change Science Program and the Subcommittee on Global Change Research, Washington. Kistler R et al., 2001: Bull. Amer. Meteor. Soc., 82, Manabe S and Stouffer RJ, 1980: J. Geophys. Res., 85, Mears CA and Wentz FW, Science, 309, Randel WJ and Wu F, 2006: J. Climate, 19, Santer BD et al., 2005: Science, 309, Santer BD et al., 2008: Int. J. Climatol., 28, Screen JA and Simmonds I, 2010: Nature, 464, Serreze MC and Francis JA, 2006: Climatic Change, 76, Serreze MC et al., 2009: Cryosphere, 3, Sherwood SC et al., 2005: Science, 309, Sherwood SC et al., 2008: J. Climate, 21, Stickler A et al., 2010: Bull. Amer. Meteor. Soc., 91, Uppala SM et al., 2005: Q. J. R. Meteor. Soc., 131, Vinnikov KY et al., 1980: Soviet Meteor. Hydrol., 6,

25 Results zonal mean bidecadal trends Example: DJF Tropics ERA-40 SOCOL Relatively weak variability, stronger in other seasons K/dec NNR REC1 20CR 25

26 Results CHUAN obs. regional mean bidecadal trends Tropics, ex. SON Lower tropospheric cooling Whole tropospheric cooling Warming in mid to upper troposphere 26

27 Results CHUAN obs. regional mean bidecadal trends Tropics, ex. SON Warming aloft Surface warming only 27

28 Results CHUAN obs. regional mean bidecadal trends Tropics, ex. SON Strong warming 28

29 Results CHUAN obs. regional mean bidecadal trends Tropics, ex. SON Warming Warming aloft, but cooling at the surface 29

30 Results zonal mean bidecadal trends Example: DJF 30

31 Results CHUAN observed regional mean bidecadal trends > Arctic Alaska (only JJA): cooling Alaska, Canada: DJF lower tropospheric warming, MAM/JJA/SON cooling (except SON Canada SE warming) W Siberia and SW/NE Central Siberia (only DJF-JJA/MAM-JJA): cooling All regions except Atlantic: (lower) tropospheric warming in DJF/MAM/JJA (except DJF in E Central Siberia, surface MAM cooling in NW Canada/W Siberia, JJA cooling in Canada/Novaya Zemlya/W Siberia), cooling in SON (except in Alaska/NW Canada) NE Atlantic: surface cooling, weak trends aloft (only MAM) 31

32 Results CHUAN observed regional mean bidecadal trends > Arctic All regions: warming (except Novaya Zemlya/Karelia/NE Atlantic (DJF), NE Atlantic/SE Canada/NE Central Siberia (MAM), NE Central Siberia (JJA), Alaska/NW Canada (SON)) All regions except Nov. Zemlya: free tropospheric warming (except Alaska (DJF), Karelia (MAM), W Siberia (DJF/JJA/SON) and SW Central Siberia (DJF)) Novaya Zemlya: surface warming, slight cooling aloft (only MAM) 32

33 Results general agreement of other datasets with observations > Tropics ERA-40: relatively good agreement with CHUAN trends incl. vertical structure, best in Americas sector, less so in Asia and Pacific NNR: agreement with CHUAN is significantly worse than that of ERA-40 for all regions, especially w.r.t. the vertical structure of trends 20CR: temporal behaviour and vertical structure of trends often different from CHUAN, ERA-40 and NNR, especially around the tropopause, and in all regions SOCOL: Some CHUAN trends outside envelope of ens. members, notably Pacific and Americas cooling around 100 hpa MAM Asia warming Pacific/Americas tropospheric warming Pacific cooling at 1000 hpa 33

34 Results general agreement of other datasets with observations > Arctic ERA-40: fairly good agreement with CHUAN trends incl. vertical structure Best for NE Atlantic, Karelia, SE Canada, Alaska, Central Siberia Less good in NW Canada, E Siberia, W Siberia, Nov. Zemlya NNR: agreement with CHUAN is significantly worse than that of ERA- 40 for NE Atlantic, SE Canada, SW Central Siberia, E Central Siberia For Karelia, NW Canada, E Siberia, Alaska, Nov. Zemlya, W Siberia, SW Central Siberia, results are equally close to CHUAN than in ERA-40 20CR: some regions show much better agreement with CHUAN compared to Tropics, but less good than ERA-40 and NNR, no good agreement for Karelia, SE Canada, Nov. Zemlya, W Siberia, Central Siberia SOCOL: CHUAN trend signs generally inside envelope of ens. members (larger internal variability cf. Tropics), exceptions: MAM SE Canada and JJA Alaska warming, SON Alaska cooling, DJF W Siberia cooling 34