Download presentation

Presentation is loading. Please wait.

Published byEmma McDougall Modified over 2 years ago

1
Simulations of the atmospheric circulation on a water-covered Earth Group on Numerical Experimentation - WGNE David Williamson NCAR Mike Blackburn University of Reading Peter Gleckler (PCMDI) Brian Hoskins (Reading) Richard Neale (CDC, now NCAR) APE Modelling Groups - Space & Atmospheric Physics, Imperial College 28 February

2
Group on Numerical Experimentation - WGNE Outline Motivation and context Experimental design + participants Aspects of climate – mean and variability Conclusions + next steps Compare idealised climates simulated by global atmospheric circulation models (AGCMs) being developed and used for numerical weather prediction and climate research. Provide a benchmark of current model behaviour and stimulate research to understand the causes of inter-model differences.

3
IPCC (2001) Climate changes over the next few decades are predicted to be much larger than we have seen so far… Uncertainty in climate predictions - IPCC TAR (2001) -

4
Group on Numerical Experimentation - WGNE Evaluation of Atmospheric GCMs - an experimental hierarchy - 1D / 2D idealised flows Dynamical core Idealised moist core Aqua-planetAMIP Full complexity GCM - unique dynamics - unique moist parameterizations Difficult to isolate reasons for differences Aqua-planet idealises the planet, not the model! Dry dynamics - linear relaxation to climatology - Rayleigh friction boundary layer Unrealistic sensitivities Aim for - single idealised moist parameterization - minimal complexity to represent processes Use in all dynamical cores Sensitivity of a moist atmosphere to dynamical formulation

5
Group on Numerical Experimentation - WGNE Moist processes are replaced by linear temperature relaxation + drag Sensitivities to numerical options different from the complete GCM Highlights moist feedbacks in climate Dynamical Core behaviour zonal mean Temperature semi-Lagrangian versus Eulerian advection Chen & Bates (1996); Chen et al (1997) moist GCM dynamical core

6
Group on Numerical Experimentation - WGNE Berlin Academy competition (1746): To determine the order and the law which winds would have to observe if the Earth were surrounded everywhere by an ocean, so as to find at all times the direction and the velocity of the wind for every place Historical aside …. Egger and Pelkowski (2006) Led to the first mathematical models of atmospheric motion 11 entries, including dAlembert and Bernoulli Tidal oscillations only (rotation + gravitational attraction) Expressly excluded effects of radiational heating, though recognised as important for the complete problem Won by dAlembert: 2 layer model of atmosphere + ocean

7
Group on Numerical Experimentation - WGNE The Experiment Complete GCMs but idealised planet More constrained experiment than real-world benchmark (AMIP) Facilitate understanding of model differences and sensitivities No land / orography 8 idealised sea surface temperature (SST) distributions 5 symmetric SSTs span a range of tropical climates Local and global-scale SST anomalies 3-year climate (following spin-up) Following Neale & Hoskins (2000) Symmetric SST profiles Latitude SST (degC) SST anomaly experiments 3KW1 1KEQ 3KEQ

8
Group on Numerical Experimentation - WGNE APE Modelling Groups GroupLocationModelResolnFeatures AGU for APE Japan (consortium)AFES v.1.15T39 L48Spectral, eulerian CGAM Reading, UKHadAM3N48 L303.75º x 2.5º grid CSIRO Aspendale, AustraliaCCAM-4-12C48 L18~220km conformal cubic grid DWD Mainz, GermanyGME ni=64 L31~1º icosahedral-hexagonal grid ECMWF Reading, UKIFS Cycle 29r2T L 159 L60Spectral, semi-lagrangian FRCGC Yokohama, JapanNICAM7km L54icosahedral grid, non-hydro. GFDL Princeton, USAAM2p14N72 L242.5º x 2º grid (IPCC) GSFC Maryland, USANSIPP-1N48 L343.75º x 3º grid K1-Japan Japan (collaboration)CCSR/NIES 5.7T42 L20s-l moisture and cloud LASG Beijing, ChinaSAMILR42 L9Spectral, eulerian MGO St. Petersburg, RussiaMGO-gcmT30 L14Spectral MIT Cambridge, USAMIT-gcmC32 L40~280km cubed sphere NCAR Boulder, USACCSM-CAM3T42 L26Spectral, eulerian UKMO Exeter, UKpre-HadGAM1N96 L º x 1.25º grid, s-lagrangian

9
Group on Numerical Experimentation - WGNE APE Modelling Groups GroupLocationModelResolnFeatures AGU for APE Japan (consortium)AFES v.1.15T39 L48Spectral, eulerian CGAM Reading, UKHadAM3N48 L303.75º x 2.5º grid CSIRO Aspendale, AustraliaCCAM-4-12C48 L18~220km conformal cubic grid DWD Mainz, GermanyGME ni=64 L31~1º icosahedral-hexagonal grid ECMWF Reading, UKIFS Cycle 29r2T L 159 L60Spectral, semi-lagrangian FRCGC Yokohama, JapanNICAM7km L54icosahedral grid, non-hydro. GFDL Princeton, USAAM2p14N72 L242.5º x 2º grid (IPCC) GSFC Maryland, USANSIPP-1N48 L343.75º x 3º grid K1-Japan Japan (collaboration)CCSR/NIES 5.7T42 L20s-l moisture and cloud LASG Beijing, ChinaSAMILR42 L9Spectral, eulerian MGO St. Petersburg, RussiaMGO-gcmT30 L14Spectral MIT Cambridge, USAMIT-gcmC32 L40~280km cubed sphere NCAR Boulder, USACCSM-CAM3T42 L26Spectral, eulerian UKMO Exeter, UKpre-HadGAM1N96 L º x 1.25º grid, s-lagrangian

10
Group on Numerical Experimentation - WGNE APE Modelling Groups GroupLocationModelResolnFeatures AGU for APE Japan (consortium)AFES v.1.15T39 L48Spectral, eulerian CGAM Reading, UKHadAM3N48 L303.75º x 2.5º grid CSIRO Aspendale, AustraliaCCAM-4-12C48 L18~220km conformal cubic grid DWD Mainz, GermanyGME ni=64 L31~1º icosahedral-hexagonal grid ECMWF Reading, UKIFS Cycle 29r2T L 159 L60Spectral, semi-lagrangian FRCGC Yokohama, JapanNICAM7km L54icosahedral grid, non-hydro. GFDL Princeton, USAAM2p14N72 L242.5º x 2º grid (IPCC) GSFC Maryland, USANSIPP-1N48 L343.75º x 3º grid K1-Japan Japan (collaboration)CCSR/NIES 5.7T42 L20s-l moisture and cloud LASG Beijing, ChinaSAMILR42 L9Spectral, eulerian MGO St. Petersburg, RussiaMGO-gcmT30 L14Spectral MIT Cambridge, USAMIT-gcmC32 L40~280km cubed sphere NCAR Boulder, USACCSM-CAM3T42 L26Spectral, eulerian UKMO Exeter, UKpre-HadGAM1N96 L º x 1.25º grid, s-lagrangian

11
Range of tropical states Precipitation (mm/day) Group on Numerical Experimentation - WGNE

12
Zonal Average Hydrological Cycle Precipitation (mm/day) Evaporation (mm/day) different scale Group on Numerical Experimentation - WGNE

13
Convective / stratiform precip. Group on Numerical Experimentation - WGNE Convective (mm/day) Stratiform (mm/day)

14
Hydrological Cycle: NCAR model Courtesy of David Williamson, NCAR Precipitation: contributions to resolution dependence, T42 / T85 params. timestepgrid Trunc n.diffusion params. at T42 Working Group on Numerical Experimentation - WGNE

15
Group on Numerical Experimentation - WGNE Inter-tropical Convergence Zone When does convection break through the trade-wind inversion? Many interacting processes ITCZ location sensitive to all these processes in models Emanuel (1994) [ Evap – Precip ] Surface Wind ECMWF - APE control (time average) (mm/day) Cool 30º lat. Eq. Warm

16
average 5º N -5º S Tropical Variability (precipitation) Group on Numerical Experimentation - WGNE (mm/day)

17
Tropical Variability (precipitation) average 5º N -5º S Group on Numerical Experimentation - WGNE Higher resolution models NICAM Icosahedral L54 7km grid non-hydrostatic no convective param. IFS Cy29r2 T L 159 L60 ~125km grid pre-HadGAM1 N96 L ° x 1.875° GME Icosahedral L31 ~100km grid (mm/day)

18
average 5º N -5º S Tropical Variability (precipitation) Group on Numerical Experimentation - WGNE (mm/day)

19
Tropical Variability (precipitation) average 5º N -5º S Group on Numerical Experimentation - WGNE Higher resolution models NICAM Icosahedral L54 7km grid non-hydrostatic no convective param. IFS Cy29r2 T L 159 L60 ~125km grid pre-HadGAM1 N96 L ° x 1.875° GME Icosahedral L31 ~100km grid (mm/day)

20
Group on Numerical Experimentation - WGNE Observations + Theory Observed variability (OLR) Hierarchy of convective organisation Time Time-longitude section of transient OLR averaged between the equator and 5N from May to July in (Nakazawa, 1988) Zonal Wavenumber Frequency (CPD) UKMO_n96 sym. spectrum (precip) Observed sym. spectrum (OLR) Images from Yoshi-Yuki Hayashi; Yukiko Yamada; NOAA CDC

21
Tropical rainfall: spectra APE control 10°S – 10°N 6hour grid-box averages LASG FRCGC HadGAM1 (umet) N48 N96 AGU GSFC CSIRO MGONCAR Group on Numerical Experimentation - WGNE

22
Tropical rainfall: Stratiform fraction APE control 10°S – 10°N Group on Numerical Experimentation - WGNE AGU CGAM DWD ECMWF GFDL GSFC K1JAPAN LASG MGO MIT NCAR UMET_48 UMET_96 CSIRO_a CSIRO_b FRCGC Some correlation with spectral shape

23
Wider SST maximum in tropics Stronger SST gradient : displaced poleward Response of zonal climate to SST qobs-control Working Group on Numerical Experimentation - WGNEhttp://www.met.reading.ac.uk/~mike/APE/ cntl qobs latitude SST (degC)

24

25

26

27

28
m=1 SST anomaly generates planetary waves Expect stationary momenum fluxes to alter zonal flow 3kw1-control Zonal mean differences SST anomaly Working Group on Numerical Experimentation - WGNEhttp://www.met.reading.ac.uk/~mike/APE/

29

30
Working Group on Numerical Experimentation - WGNE Storm-track statistics Tracking of storm features using 6 hourly sea-level pressure NCAR model for all 8 SSTs Courtesy of Kevin Hodges, ESSC, Reading Mean Intensity latitude Pressure anomaly (hPa) Track Density latitude Number per month in 5º radius peak cntl qobs flat cntl5n 1keq 3keq 3kw1 Zonal speed Zonal speed (ms -1 ) latitude

31
Group on Numerical Experimentation - WGNE Storm-track statistics Tracking of storm features using 6 hourly sea-level pressure 6 models for flat SST Courtesy of Kevin Hodges, ESSC, Reading Track Density latitude Number per month in 5º radius Mean Intensity latitude Pressure anomaly (hPa)

32
Group on Numerical Experimentation - WGNE Low Frequency Variability Significant zonal wavenumber m=5 in 3-year means Slow propagation, c = 1.7ms -1 Significant correlation with annular mode variability Courtesy of Masahiro Watanabe, Hokkaido University. GRL 32, L (2005) 1-point correlation maps: 10-day low-pass surf. pressure EOFs of 10-day low-pass streamfunction =0.3 Ref 51.6N

33
Global Energy Balance APE control experiment: 3 year averages + temporal variability Net flux (toa; surface) sw_dn TOA sw_up sw TOA lw Group on Numerical Experimentation - WGNE

34
Group on Numerical Experimentation - WGNE Cloud + Albedo

35
Group on Numerical Experimentation - WGNE Summary Documenting a wide variety of model behaviours No convergence for Δx>100km – basic tropical features not resolved? Attempts to understand sensitivities in individual models Additional experiments needed to understand model differences (e.g. no cloud-radiative feedback; fixed radiation; SCM) Diagnostics focus: Tropical wave activity Diurnal cycle Mid-latitude variability & storm-tracks Issues: Reference solution is unknown Resolution convergence? (HPEs + parameterizations)

36
Group on Numerical Experimentation - WGNE Blank text

37
Group on Numerical Experimentation - WGNE Blank text

Similar presentations

© 2016 SlidePlayer.com Inc.

All rights reserved.

Ads by Google