Transcript Slide 1
Impact of convection schemes on cloud feedbacks in the tropics Mark Ringer CFMIP/GCSS, Vancouver, June 8-12th 2009. © Crown copyright Met Office Aims and outline • Initial study of structural uncertainty & cloud feedbacks – in this case dependence on convection scheme • Compare with parameter-perturbed (QUMP, climateprediction.net) & multi-model ensembles • Focus on tropical response to 2 × CO2 in Met Office slab model, HadSM3 • Three experiments, using Gregory-Rowntree (“Standard”, STD), Emanuel (EMA), Betts-Miller (B-M) convection schemes • “Usual” diagnostics (cloud radiative forcing, ISCCP clouds) plus analysis methods from three recent papers © Crown copyright Met Office Surface temperature change (K/K) HadSM3 is dominated by minima in warming over Pacific & Atlantic STD B-M © Crown copyright Met Office EMA ΔT(K) STD EMA B-M Mean 2.69 3.47 3.37 Ocean 2.18 3.02 2.89 Land 4.15 4.79 4.76 Global 3.47 4.03 3.77 This is also a feature of the coupled model’s response to warming STD EMA STD: Coupled: A1B:2090s-2000s B-M © Crown copyright Met Office Change in cloud forcing (Wm-2/K) Minimum in the warming is clearly associated with local cloud changes STD EMA B-M Large negative cloud feedback in SE Pacific in STD, weaker in EMA, opposite sign in B-M. (Similar in Atlantic) Larger positive feedback over land, particularly over South America, in STD. © Crown copyright Met Office The cloud forcing changes are also evident in the coupled model STD EMA STD: Coupled: A1B:2090s-2000s B-M © Crown copyright Met Office Change in ISCCP high cloud (%/K) Large responses in STD & EMA, weak response in B-M Thin STD EMA B-M Thick © Crown copyright Met Office Change in ISCCP low cloud (%/K) Large response in STD, weakens in EMA, changes sign in B-M All STD EMA B-M Thick © Crown copyright Met Office 500 hPa vertical velocity (hPa/day/K) STD has a strong large-scale dynamical response STD B-M © Crown copyright Met Office EMA This gives a good indication of the changes to the large-scale circulation – the major areas of increasing/decreasing ascent & descent. Precipitation change relative to control (%/K) Strongly related to vertical velocity STD EMA B-M In general, increased ascent corresponds to increased precipitation and vice versa. © Crown copyright Met Office “Dynamic” precipitation (%/K) Large areas in STD where dynamic changes lead to increased precipitation STD EMA B-M Difference between the precipitation change and the rate of moistening from the ClausiusClapeyron relation (i.e. ~ 7% K-1 of warming) (following Vecchi & Soden, 2007) © Crown copyright Met Office B-M more closely resembles AR4 coupled models (This pattern is also seen in AR4 slab models) STD EMA B-M (Figure 8c from Vecchi & Soden, 2007) © Crown copyright Met Office Links to SST variability in key regions (following Toniazzo et al. 2007) Nino 3.4 SE Pacific 7.5-17.5ºS 80-125ºW © Crown copyright Met Office Nino 3 In STD, increased low-level cloud in SETP associated with increased SST variability locally and over the Equator. This behaviour is suppressed with EMA and B-M schemes. Relationship between lower and upper level moisture over tropical ocean High correlation with all three schemes STD EMA Joint pdfs binned into intervals of lower and upper tropospheric moisture (PWC – UTH) B-M r(UTH, PWC) OBS 0.85 STD 0.79 EMA 0.89 B-M © Crown copyright Met Office 0.84 (following Bennhold & Sherwood 2008) Relation between changes in lower and upper level moisture is much weaker with EMA and B-M schemes STD B-M © Crown copyright Met Office EMA r(ΔUTH, ΔPWC) STD 0.84 EMA 0.32 B-M 0.44 pdfs of changes in UTH & PWC under 2×CO2 Summary – 1 • Changing the convection scheme in HadSM3 has a significant impact on the tropical surface temperature response to CO2 doubling which is linked to different responses of low-level cloud With the standard Gregory-Rowntree scheme (STD): • The increased low cloud in the SE Tropical Pacific => increased SST variability => large-scale pattern of cloud/precipitation changes • Circulation changes act to increase precipitation changes in many areas • Strong connection between moisture changes in the lower and upper troposphere © Crown copyright Met Office Summary – 2 • These relationships are suppressed with the EMA and B-M schemes • With B-M low cloud is reduced in SETP and circulation changes act to reduce the precipitation over most of the tropics – this is more consistent with (the ensemble mean of) AR4 models • Weaker relationship between changes in lower and upper tropospheric moisture with B-M and EMA With the standard Gregory-Rowntree scheme the slab model responses of surface temperature and low cloud, and the hypothesized link between low cloud in the SETP, SST variability and circulation are consistent with the behaviour of the coupled model © Crown copyright Met Office ΔTS in the CFMIP1 models (K/K) HadSM3 is not unique but has the clearest example of SE Pacific warming minimum © Crown copyright Met Office Cloud forcing response to 2 × CO2 The low cloud change in HadSM3 (STD) is a ‘fast’ response © Crown copyright Met Office