Aquaplanet warming experiments with CAM: a tale of the subtropics

Transcription

1 Aquaplanet warming experiments with CAM: a tale of the subtropics AMWG meeting, 12 February 2018, NCAR Thomas Toniazzo Uni Research (?) & Bjerknes Centre Bergen, Norway

2 Plan of this talk A) Intro B) Slab AP ( QS ) (non-)equilibrium simulations with CAM6 and CAM6-Oslo C) Slab AP ( QS ) 4xCO2 sensitivity experiments with CAM6-Oslo D) Prescribed-SST AP ( QP ) +4C sensitivity experiments with CAM5.3 on different dycores

3 Development configuration of NorESM Based on NCAR's CESM2-ε CAM6-L32 finite-volume with tweaks + CAM-Oslo atm chemistry & aerosol modules + MICOM on tripolar 1ºx(1º-1/4º) grid, sigma L53 + HAMOCC 6 ocean biogeochemistry & CLM 5.0, CISM/GLIMMER 2.0, CICE 5.0 MCT coupler

4 A) Why AP 1. Atmosphere only active component, hence no feedbacks from other componets 2. Cheap(er) to run 3. Energy cycle transparent (closed in slab) 4. Offers links to existing theory of the global circulation 5. hence less difficult to understand impact of model or parameter changes

5 A) Slab AP simulation with CAM-Oslo (dev: beta06) CAM-OSLO only, 30m slab, no land/ice/ocn Perpetual equinox, no diurnal cycle (but implied when cosz averaging) No topography, ridge/beljaars scheme off, gw_oro on Initial conditions: aqua_ _1.9x2.5_l32_c nc TRACMIP Qfluxes, symmetrised All other input (F2000) symmetrised (APin/ directory under inputdata/noresm-only/atm/cam/chem/trop_mozart_aero/emis/) clubb_gamma=0.25, zmconv_num_cin=1 PI GHGs, TSI=1350 W/m2

6 A) Other NorESM2 CAM formulation changes wrt CESM2 1. COARE v3 (Fairall et al. 2003) surface-flux coupling with ocean 2. Local conservation of hydrostatic energy under moisture changes¹ 3. Zonal-mean and global conservation of angular momentum 4. Time-step average zenith angle outside radiation module N.B. All of these also implemented in non-dev ''master'' ¹approximate until other components are also updated

7 A) We already used a similar configuration of CAM-Oslo, but based on CESM1.2/CAM5.3, for an AP MIP experiment. A version close to this will be our fall-back model should ε remain finite. TRAC-MIP: a model intercomparison based on ML-AP integrations

8 B) Beta06: CAM-CESM vs CAM-OSLO

9 B) QSC6 cases (5_4_128) CESM2 is cold. Cooling also occurs in energetically closed slabaquaplanet simulations where the only active component is CAM. A globally balance, observationsbased QFLUX climatology is used here. Strong cooling initiates in the subtropics, then spreads. (arbitrary i.e. diffs wrt i.s.) SST 200hPa OLR Outgoing Net Radiation

10 B) QSC6 cases (5_4_128) PBL feedbacks do not seem responsible for the cooling. Short-wave cloud forcing is (if anything) positive downward as is long-wave cloud forcing. So the net atmospheric radiative feedback is negative (stabilizing). Total and especially low cloud cover increases, but lower-tropospheric stability (LTS, SSTT700) decreases. SWCF (W/m²) PRECT (mm/day) LWCF LTS (K) CLDTOT CLDLOW

11 B) Decreasing stability, with cooling strongest at the top of the PBLcloud layer... A probable feedback cooling mechanism

12 B) Decreasing stability, with cooling strongest at the top of the PBLcloud layer, which thickens... A probable feedback cooling mechanism

13 B) Decreasing stability, with cooling strongest at the top of the PBLcloud layer, which thickens, and sustains LW cooling. A probable feedback cooling mechanism

14 B) Decreasing stability, with cooling strongest at the top of the PBLcloud layer, which thickens, and sustains LW cooling. Even as q drops, RH increases. A probable feedback cooling mechanism

15 B) Cooling in CESM-CAM6 prototype There seems to be an atmospheric cooling problem Most evident without land, ice, ocean components Analysis of the tendencies in slap-ap simulation points at a localised cloud LW feedback PBL can keep suppling moisture through sharp drop in Q at cloud-layer base T there drops through LW cooling enough to keep Q down and RH up Still seen in beta07; has this changed in beta09? All changes in CAM-Oslo! C)

16 C) Beta06: Global SST timeseries with CAM-Oslo and formulation changes 4.2C 6.6C

17 C) Time evolution of potential temperature on the Equator All CAM6O -zenith... -Energy... -COARE Oslo chem only

18 C) Precipitation in subtropics, timeseries

19 C) Zonal-mean precipitation timeseries

20 C) Zonal mean SST and precipitation

21 C) Warming amplification factor

22 C) Cloud Cover and Cloud Radiative forcing in subtropics, timeseries

23 C)

24 C) Meridional mass streamfunction, differences (4xCO2)

25 C) Summary From 4.2C to 6.6C warming under 4xCO2 when formulation changes are used Mostly AM and COARE Big contribution from cloud forcing in the subtropics Mainly «base» state in the subtropics Generally better qualitative consistency with theoretical expectations climatology potentially «bi»-stable, wet/dry subtropics dry / water cloud moist / ice cloud wet to dry under 4xCO2 leading to very different response Is this related with B)? Stronger tropical upper-tropospheric warming Mostly AM

26 D) Things I don't know Q: how does the Hadley circulation respond to warming? SQ1: dependence on atmospheric model and on conserved integrals SQ2: relationship with atmospheric transports and with point of thermal equilibrium (subtropics/subpolar) Experiments: AP (no land, no topography, GWs off) Prescribed SSTs representative for Equinox or Solstice Perturbation of SSTs +4C Diagnostics: Δ meridional streamfunction maxima Δ width of Hadley cell and ITCZ position (0's of MSF) Δ max zonal wind in STJ's and STJ latitude

27 D) CQE and circulation shifts (Schneider et al. 2004) The job of moist convection is to export high sb to regions of lower sb: S N 2 ~ s b (Emanuel et al. 1994) Expected shift in meridional position of ascending branch of HC: y= F 0 df /dy 0

28 D) Held and Soden 2006 and the role of subtropical warming H ~ OLR/Lq N²ITCZ But N²subtr Eq δ(mcq) ~ δolr H ~ OLR/N² δolr ~ 2 %/K and δq ~ 7 %/K Therefore δmc ~ -5 %/K Convective adjustment also implies δ N²ITCZ ~ 2 %/K If H ~ Mc, then for the subtropics δ N²subtr ~ 7 %/K. Hence reduced thermal wind in subtropics. Since also δ N²subpolar is small, there is increased thermal wind in mid-latitudes. The HC expands as it weakens. This is all qualitatively seen in the experiments.

29 D) Schneider (1977), Held & Hou (1980) model a) momentum conservation b) thermal wind balance c) energy balance R (aω) R²(HΔH/CτΔv)(aΩ)

30 D) Schneider (1977), Held & Hou (1980) model a) momentum conservation b) thermal wind balance c) energy balance Effect of increased heat export

31 D) Schneider (1977), Held & Hou (1980) model a) momentum conservation b) thermal wind balance c) energy balance Effect of increased AM export

32 D) Quantitative response in HC diagnostics for +4K experiments Perpetual equinox (sstobs) NH ITCZ SH

33 D) Quantitative response in HC diagnostics for +4K experiments HadISST, MAM NH ITCZ SH

34 D) Quantitative response in HC diagnostics for +4K experiments Perpetual solstice (sst10) NH ITCZ SH

35 D) Quantitative response in HC diagnostics for +4K experiments HadISST, JJA NH ITCZ SH

36 D) Summary +4K sensitivity tests with CAM-AP Big quantitative differences with resolution and dycore, related with the degree of warming in the subtropics Theoretical expectations qualitatively matched, but some H&Htype scalings are more robust (e.g. mass flux) than others (e.g. width, and zonal winds) Important role of (changing) exports to the mid-latitudes Better general agreement for equinoctial circulations Transient (seasonal), zonally asymmetric circulations present a mix of solstitial and equinoctial behaviour Simplified physics may clarify the role of dycore better.