© Crown copyright Met Office Future Changes in Tropical Cyclone Genesis Parameters in Hadley Centre Coupled Climate Models: A Low Resolution Model Study

© Crown copyright Met Office

Future Changes in Tropical Cyclone Genesis Parameters in Hadley Centre Coupled Climate Models:A Low Resolution Model StudyJoo-Hong KimClimate Extremes Research Team, Climate Prediction Group, Met Office Hadley Centre, Exeter, UK

Outline

<First Part>

• Greenhouse Gases-induced Climate Change

• QUMP (Quantifying Uncertainty in Model Predictions) Project in the Hadley Centre

• Surface Temperature Responses in the Hadley Centre Coupled Climate Models

<Second Part>

• General Introduction of Topics on “Tropical Cyclones and Climate Change”

• Description of Tropical Cyclone Genesis Parameters

• Future Changes in Tropical Cyclone Genesis Parameters in Low Resolution Hadley Centre Coupled Climate Model Suites

• Northern Hemisphere season (JJASON)

• Southern Hemisphere season (DJFMAM) – if time is available

Additionally,

• About the Met Office Hadley Centre – if time is available


Greenhouse Gases-induced Climate Change – “Hot issue Under Debate”

• Global mean temperature rise has been observed during 20C, and is projected to continue during 21C, due to the radiative forcing (CO2 and other GHG gases) from the atmosphere

• Recent flattening of temperature rise since 2000 onwards despite of continuous CO2 increases

• No doubt 2000s the warmest decade since 1850

• Met Office Decadal Prediction System (DePreSys) has a skill to predict a weak period of warming

• Arising from natural variability of climate system

• However, uncertainty remains due to the model deficiency –Quantifying the uncertainty in model predictions (QUMP) is required!


Lindzen (2008)

Smith et al (2007)

QUMP(Quantifying Uncertainty in Model Predictions)

• QUMP how to?

• Multimodel ensemble – sampling uncertainties arising from structural differences between GCMs

• Perturbed physics ensemble (PPE) – sampling uncertainty arising from ill-constrained physics parameters whilst tuning a model

• Ideal: Combination of two approaches – IPCC AR5

• PPE approach in the Met Office Hadley Centre,

• Began in early 2000s

• Main purpose: to estimate a PDF of the global mean temperature response to CO2 doubling (i.e. climate sensitivity) in Hadley Centre coupled models, Version 3 (HadCM3; Gordon et al 2000, ClimDyn)

• Each member differs from control by perturbing physical parameters to values elicited from experts

• Though not exploring structural uncertainties, this metric is useful for producing a large ensemble in a ‘systematic’ way

• Perturbing the physics detunes the model, destroys TOA radiative balance, leading to unstable climate, i.e. drift → SST and sea surface salinity “flux adjustment” required (Collins et al 2006, ClimDyn)© Crown copyright Met Office

Timeline: Hadley Centre Models

HadSM3(‘slab’ ocean)


HadCM3

• Standard version: “No climate drift without flux adjustment”

• Atmospheric component (called HadAM3)

• N48L19: 2.5° (in latitude) × 3.75° (in longitude) and 19 vertical levels (~ comparable to a spectral resolution of T42)

• Fully dynamic ocean component

• 1.25° × 1.25° and 20 levels

• Known to show the best simulation of ENSO variability in current climate amongst coupled GCMs (Joseph and Nigam 2006)

cf. HadSM3: HadAM3 coupled to thermodynamic mixed layer ‘slab’ ocean

o

o o p net

dTh c F Q

dt

ρo, cpo: density and specific heat at constant pressure of seawater

ho: annual mean mixed layer depth (typically 50m, chosen to be sufficient to allow for the seasonal heat storage in the ocean without unduly prolonging the thermal response time of the ocean)T: mixed layer temperatureFnet: net atmosphere to ocean heat fluxQ (Q flux): internal ocean mixed layer heat flux, mimics deep water heat exchange and ocean transport

: Ocean Temperature Equation


QUMP with HadSM3 and HadCM3

• What are perturbed?

• Six key climate processes: large-scale cloud, convection, radiation, boundary layer, dynamics, land surface processes and sea ice

• Thirty-one key physical parameters that are thought to be crucial for climate sensitivity but ill-constrained due to our poor understanding (Murphy et al 2004, Nature)

• How are their uncertainty ranges decided?

• Many experts’ elicitation all around the UK

• How are they perturbed?

• Monte-Carlo sampling for parameters

• First, HadSM3 were chosen to produce hundreds of PPE: Why? Computationally cheaper (Murphy et al 2004, Webb et al 2006)

• Next, sixteen of them, which show a good present-day climate as well as wide-range of climate sensitivity are selected to study transient (i.e. time-dependent) climate response in HadCM3 (Collins et al 2006, ClimDyn)

• Note that ocean parameters or carbon cycle parameters are not perturbed in this first project

?

?

?


Ensemble Experiments

• Various experiments: control, 1% yr-1, 20C3M, A1B, A2, B1, A1FI, etc

• Used experiments in this study

• Control experiments with CO2 and other forcing agents fixed at pre-industrial values (‘1CO2’ for HadSM3, ‘CTRL’ for HadCM3)

• CO2 increased at a rate of 1 % per year compounded, run to four times pre-industrial concentrations (‘2CO2’ for HadSM3, ‘1PCT’ for HadCM3) (Note: equilibrium response is not available for this model)

• Future change is defined by the response to CO2 doubling

• HadCM3: Ensemble-mean difference in 20 yr (model yr 61-80; a span of CO2 doubled years) average between 1PCT and CTRL PPE runs –> Transient response

• HadSM3: Ensemble-mean differences in 20 yr (after reaching equilibrium) average between 2CO2 and 1CO2 PPE runs –> Equilibrium response (: Climate Sensitivity)

• Why are the two idealized experiments selected?

• Mainly to examine the impact of different ocean representation

• To dispel doubts on the different definition of the response to CO2 doubling,

• 20 yr periods in the HadCM3 transient runs in which the global mean temperature responses would be about equivalent to those of the corresponding HadSM3 equilibrium runs are chosen and compared with the periods of CO2 doubling


Global Mean Temperature Responses

ΔT1.5m(Yr 61-80 ~2×CO2)

0 adigc adigd 3.53 aenwp aenwq 2.17 096-115 3.511 adrea adrfa 2.58 afcra afcqa 1.81 086-105 2.572 adtlg adrvg 2.42 afcrb afcqb 1.81 081-100 2.393 adumd adund 2.82 afcrc afcqc 1.79 095-114 2.834 adumf adunf 2.88 afcrd afcqd 2.06 084-103 2.895 adrhf adrif 3.75 afcrf afcqf 2.07 099-118 3.746 adseo adsfo 3.44 afcrh afcqh 2.2 094-113 3.457 adsbd adscd 4.42 afcri afcqi 2.21 112-131 4.48 adrhj adrij 3.9 afcrj afcqj 2.02 103-122 3.919 adseb adsfb 4.44 afcrk afcqk 2.44 105-124 4.4210 adrhl adril 4.88 afcrl afcql 2.52 103-122 4.8611 adsbb adscb 4.54 afcrm afcqm 2.57 098-117 4.5512 adsbh adsch 4.62 afcrn afcqn 2.27 109-128 4.6213 adrya adrza 4.76 afcro afcqo 2.56 099-118 4.7414 adrye adrze 5.4 afcrp afcqp 2.32 115-134 5.415 adsea adsfa 7.11 afcrq afcqq 2.66 131-150 6.9916 aduvb aduvc 2.19 afcrr afcqr 1.62 079-098 2.16

3.98 2.18 3.97

HadSM3 HadCM31CO2 2CO2 ΔT1.5m CTRL 1PCT Model Yr ΔT1.5m

© Crown copyright Met Office Climate Sensitivity Transient Response

Scatter Plots: Global Mean Temperature Responses

1CO2 2CO2 CTRL 1PCT0 adigc adigd aenwp aenwq1 adrea adrfa afcra afcqa2 adtlg adrvg afcrb afcqb3 adumd adund afcrc afcqc4 adumf adunf afcrd afcqd5 adrhf adrif afcrf afcqf6 adseo adsfo afcrh afcqh7 adsbd adscd afcri afcqi8 adrhj adrij afcrj afcqj9 adseb adsfb afcrk afcqk

10 adrhl adril afcrl afcql11 adsbb adscb afcrm afcqm12 adsbh adsch afcrn afcqn13 adrya adrza afcro afcqo14 adrye adrze afcrp afcqp15 adsea adsfa afcrq afcqq16 aduvb aduvc afcrr afcqr

HadSM3 HadCM3

r=0.88 r=0.80

Ocean [30S-30N]


Climate Change in the Tropics

• Tropical SST has warmed up, and is projected to increase in this century

• The warming would not be uniform, but

• SST warming pattern

• Viewpoint 1: zonal SST gradient

• El Nino- or La Nina-like response?

• Viewpoint 2: meridional SST gradient

• Enhanced equatorial response (EER) as an El Nino-like warming? (e.g. Liu et al 2005; Xie et al 2010)

• The SST warming pattern is crucial for determining precipitation change and naturally tropical cyclone change as well


Climate Change –“El Nino- or La Nina-like?”

Collins (2005, ClimDyn)

SAT trend [K/century] from CMIP2 1%/yr CO2 increase

El Nino-likeLa Nina-like Neutral

Models that have the largest ENSO-like climate change Models that have the largest ENSO-like climate change also have the poorest simulation of ENSO variability!also have the poorest simulation of ENSO variability!

HadCM3

HadCM3

SST Warming Pattern –“Is There a Robust Response Amongst Models?”


SST trend [K/century] from CMIP2 1%/yr CO2 increase

Trend in HadISST (1940-2000)

Liu et al (2005, JC)

‘‘Enhanced Equatorial Response (EER)’ is more Enhanced Equatorial Response (EER)’ is more robust model response than ENSO-like change!robust model response than ENSO-like change!

- El Nino-like: GFDL, IPSL, MIROC_medres, MPI, MRI, UKMO_hadcm, UKMO_hadgem- La Nina-like: CSIRO, FOAM, GISS- Central Pacific: CCCMA, CNRM, MIROC_hires- Zonally uniform: IAP

HadCM3

HadCM3

SST Warming Pattern:HadCM3 and HadSM3 PPEs

Clear equatorial ridgeEl Nino-like change

Lack of equatorial ridge

-HadCM3: Yr 61-80, spanning CO2 doubled year (~ yr 70)

-HadSM3: Equilibrium response to CO2 doubling© Crown copyright Met Office

HadCM3 Paradox

• Collins “No ENSO-like change”

• Liu et al “El Nino-like according to the zonal SST gradient change but less discernible enhanced equatorial response”

• Kim “Ensemble-mean pattern looks El Nino-like in terms of the zonal and meridional gradient of SST change together”

- Are they really contradict each other?© Crown copyright Met Office

Beat the Paradox – Get Ready!

• Collins versus Liu et al

•Just a minor difference in their perspectives on the term “ENSO-like”

• Collins & Liu et al versus Kim

•Standard alone vs. ensemble-mean of various versions with perturbed atmospheric parameters

•Non-flux adjusted vs. flux adjusted


Beat the Paradox – Action!

??

Model version w/ standard parameter setting Ensemble-mean


What to Keep in Mind?

• The flux adjustment may not be critical in determining the future SST change pattern at least in standard HadCM3

• Differently perturbed atmospheric parameters do act significantly on the amplitude of the future SST change, but do not modify the gross SST change pattern dramatically amongst PPE members

– likely arising from lack of the structural difference

- The amplified banded equatorial response in the parameter perturbed HadCM3 models, compared to the standard HadCM3 is interesting but it is impossible to quantify the role of each parameter within these seventeen members

• The equatorially banded response is only shown in the HadCM3 ensemble-mean compared to ‘slab’ ocean coupled versions, which is very likely arising from dynamic ocean coupling

• The SST response patterns look similar other than near equatorial region© Crown copyright Met Office

?

Now Move on

Tropical Cyclone-related Diagnostics

IBTrACS (1842-2008)IBTrACS (1842-2008)


Recollection: Main Purpose of This Study

• Use of the QUMP ensemble

•Ensemble approach: inevitable for climate change projections

•The impact of perturbed parameters

•How large the uncertainty in terms of the spatial change pattern, and is it still comparable to the MME approach as for the climate sensitivity?

• Use of the two differently coupled versions (dynamic vs. ‘slab’)

•The impact of ocean representation: material or trivial? global or regional?


Tropical Cyclones and Climate Change

Approaches

1. Quantifying the relationship between tropical cyclones and climate from modern observations as well as via ‘paleotempestology’ (geological proxies, historical documents, etc)

2. Detecting in tropical cyclones in high-resolution GCM or dynamic downscaling simulations and relate them to the simulated climate change

3. Statistical-dynamical downscaling

4. Investigating large-scale diagnostics pertinent to tropical cyclone activity in a low-resolution model ensemble


Approach 1

• Debate heated up in 2005 onwards since two stunning papers (Emanuel 2005, Webster et al 2005) published – “Has the tropical cyclone intensity (or the frequency of intense tropical cyclones) really shown an increasing trend during 20C?”

• Mostly concentrated on the North Atlantic variability

• Pros: Santer et al (2006), Holland and Webster (2007), Elsner et al (2008)

• Cons: Landsea (2005), Michaels et al (2005), Chan (2006), Klotzbach (2006), Landsea et al (2006)

• Uncertainties in modern observational best track records due to changing instruments and skills – “The intensity estimates are less reliable especially for stronger tropical cyclones”

• A medieval peak is reconstructed by the sedimentary evidence and the statistical model estimate using proxy-reconstructed large-scale indices, which is comparable to the recent upswing of the Atlantic tropical cyclone counts

Emanuel (2005)

Webster et al (2005)

Landsea (2005)

Mann et al (2009)

Kossin et al (2006)

•Knutson et al (2010)’s statement: “It remains uncertain whether past changes in any tropical cyclone activity (frequency, intensity, rainfall, and so on) exceed the variability expected through natural causes, after accounting for changes over time in observing capabilities”© Crown copyright Met Office

Chan (2006)

Approach 2

• Towards higher-resolution numerical model

• The main stream of researches

• Direct detection and tracking of tropical cyclone(-like) vortices with the intensity threshold used in the real world

• Restricted by limited computer power

• Still the simulated intensity is weak

- Bender et al (2010) attained realistic intensity simulations by using an operational hurricane prediction model


Approach 3

• Statistical-dynamical downscaling

• developed by Emanuel et al (2008)

• Random seeding of numerous weak disturbances

• Trajectory simulations using the beta-and-advection model

• Intensity simulations using the air-sea coupled intensity prediction model


Approach 4

• Towards a large ensemble

•Inevitable to use computationally cheaper low-resolution models [Δx ~ O(102 km)]

•Tropical cyclones are poorly simulated in such models

• Investigating the large-scale diagnostics related to tropical cyclone activity

•Formulating a tropical cyclone genesis parameter by combining several large-scale dynamic & thermodynamic variables


?

?

Which Approach Here?

• For the diagnostics from low-resolution GCMs

•All approaches are applicable, in practice, but …

• This study confined to “Approach 4”

•Motivated by McDonald et al (2005)

•A good starting point using a large ensemble for investigating tropical cyclone genesis distribution

•My in-house first role


?

Tropical Cyclone Genesis Parameters

Name Formula

Gray’s seasonal genesis parameter (Gray 1979)

GrayGPGrayGPRoyer’s convective seasonal genesis parameter (Royer et al 1998)

ConvGPConvGPEmanuel and Nolan’s genesis potential index (Emanuel and Nolan 2004)

GPIGPI


1200mb

950mb

950mb

500mb60m

500 700mb

0msfc

5 3

4026 C 5 max ,1

30e

w w

f Vf

f p

Hc T dz

p

1200mb

950mb

950mb

5 3 c

f Vf k P

f p

2 3200mb 3

3pot5 2 600mb

850mb

850mb

10 1 0.150 70

VHVf

p

* f: Coriolis parameter, ζ: relative vorticity, V: horizontal winds, p: pressure level, T: ocean temperature, ρw and cw: density and specific heat capacity of sea water, θe: equivalent potential temperature, H: relative humidity, k: proportionality factor, Pc: convective precipitation, Vpot: potential intensity

IIζζ IIWSWS

EE

CPCP

IIθθ

JJηη JJWSWS JJHH JJpotpot

IIζζ IIWSWS

IIHH

GrayGP

• Thermodynamic potential: ocean thermal energy (E), moist static stability (Iθ) and relative humidity (IH)

• In a warmer climate GrayGP gives a large increase of total cyclogenesis frequency, owing to a large increase in the ocean thermal energy factor

• A criticism about its reliance on the constant value of 26˚C as the threshold for tropical cyclone formation

• With increasing temperatures in the tropics, the vertical lapse rate will adjust such that the warmer atmosphere will experience a shift in atmospheric stability, resulting in an increase in the threshold SST required to support deep convection

• Understanding of how the threshold changes© Crown copyright Met Office

1200mb

950mb

950mb

500mb60m

500 700mb

0msfc

5 3

4026 C 5 max ,1

30e

w w

f Vf

f p

Hc T dz

p

IIζζ IIWSWS

EE IIθθ IIHH

ConvGP

• Modified GrayGP (Royer et al 1998)

•Dynamic potential is the same as the GrayGP

•Thermodynamic potential is totally discarded

• Convective precipitation (Pc) as a single thermodynamic potential parameter

•The measurement of the intensity of convection in the GCM

•Relating it with tropical cyclones would be more direct and physically consistent


1200mb

950mb

950mb

5 3 c

f Vf k P

f p

CPCPIIζζ IIWSWS

GPI

• Thermodynamic potential intensity (Jpot) of tropical cyclones as one of the thermodynamic potential parameters

• Log fit to the best track data using multiple regression, optimized to give the best fit to the spatial distribution and seasonal variation

• Differences in constituent variables, powers and constants may result in a different future change – A comparison is required!


2 3200mb 3

3pot5 2 600mb

850mb

850mb

10 1 0.150 70

VHVf

p

JJηη JJWSWS JJHH JJpotpot

ΔConvGP (JJASON)



-HadSM3: Equilibrium response to CO2 doubling

Unit: # per 2.5x3.75 per 20yr

Contour: Ensemble-mean climatology (control)Shading: Ensemble-mean future change

Contour: Ensemble-mean future changeShading: # of members having same change sign with ensemble-mean future change for each grid box

Absolute values of ensemble-mean future changes less than 0.1 are masked out

ΔConvGP vs ΔGPI (JJASON)



Contour: Ensemble-mean future change (Grey for negative values)Shading: # of members having same change sign with ensemble-mean future change for each grid box

HadCM3

HadSM3

ΔConvGP ΔGPIAbsolute values of ensemble-mean future changes less than 0.1 are masked out

North Atlantic & Northeast Pacific


ΔConvGP

ΔGPI

Menkes et al (2009)

r (TC, NINO3.4)

Vecchi and Soden (2007)

ΔGPI (SRES A1B, 18 models)

Northwest Pacific


ΔConvGP

ΔGPI

Menkes et al (2009)

r (TC, NINO3.4)

ΔGPI (SRES A1B, 18 models) Vecchi and Soden

(2007)

Yokoi and Takayabu (2009)






Dynamic Variables(JJASON)


ΔV925 & Δζ925

HadCM3

HadSM3


ΔWS



Thermo. Variables(JJASON)

© Crown copyright Met OfficeContour: Ensemble-mean future changeShading: # of members having same change sign with ensemble-mean future change for each grid box



HadCM3

HadSM3

ΔPc ΔPI ΔH

Quantifying Roles of Individual Components


Overbar with e: Ensemble-meanΔ: Future change of individual memberCTRL: Preindustrial control

CTRL CTRL

CTRL CTRL

CTRL CTRL

ConvGP ConvGP

ConvGP

ConvGP

WS

ee e

WS I

e e

WS I

e e

WS CP

f I I CP

f I I CP

f I I CP

δ(ABC)= δA∙(BC) + δB∙(AC) + δC∙(AB) (δ → 0)


ΔConvGPIζ

ΔConvGPIws

ΔConvGPCP

ΔGPIJη

ΔGPIJws

ΔGPIJpot

ΔGPIJH

HadCM3(JJASON)


ΔConvGPIζ

ΔConvGPIws

ΔConvGPCP

ΔGPIJη

ΔGPIJws

ΔGPIJpot

ΔGPIJH

HadSM3(JJASON)

Table summary (JJASON)

RegionCommon change Main driver

ConvGP GPI ConvGP GPI

North Atlantic Ocean

Decrease in the Caribbean Sea Iζ, IWS and CPHadCM3: JWS, JH > Jpot

HadSM3: JWS, JH

Increase to the east of Florida Jpot, JWS

Northeast Pacific Ocean

1. Meridional dipole-like change (decrease to the north, increase to the south) in the NEP-MDR2. Increase along the ITCZ

1. North: Iζ > IWS > CP South: Iζ, CP > IWS

2. Iζ, IWS and CP

1. North: Jη, JWS, JH South: Jη, JWS, Jpot

2. JWS, JH > Jη, Jpot

Northwest Pacific Ocean

1. Increase in the north and east of the NWP-MDR2. Decrease across the Philippines*

1. Increase to the east of the NWP-MDR2. Decrease in the NWP-MDR

1. HadCM3: Iζ

HadSM3: Iζ (east),

CP (north)2. HadCM3: Iζ, CP

HadSM3: Iζ

1. HadCM3: Jη, JWS, JH

HadSM3: Jη, JWS

2. HadCM3: JWS, Jpot, JH

HadSM3: Jη, JH

Relatively more consistent contribution from vorticitywind shear, relative humidity

The uncertain changes are marked with an asterisk (*).North Indian Ocean is not included because future changes are small and uncertain.© Crown copyright Met Office

ΔConvGP (DJFMAM)








ΔConvGP vs ΔGPI (DJFMAM)


Contour: Ensemble-mean future change (Grey for negative values)Shading: # of members having same change sign with ensemble-mean future change for each grid box

HadCM3

HadSM3

ΔConvGP ΔGPIUnit: # per 2.5x3.75 per 20yr


Southwest Pacific & Australia


HadCM3

HadSM3

ΔConvGP

Menkes et al (2009)

r (TC, NINO3.4)



Southwest Indian


HadCM3

HadSM3

ΔConvGP

Menkes et al (2009)

r (TC, NINO3.4)

Ho et al (2006)

El Nino minus La Nina








Dynamic Variables(DJFMAM)


ΔV925 & Δζ925

HadCM3

HadSM3


ΔWS



Thermo. Variables(DJFMAM)

© Crown copyright Met OfficeContour: Ensemble-mean future changeShading: # of members having same change sign with ensemble-mean future change for each grid box



HadCM3

HadSM3

ΔPc ΔPI ΔH

Quantifying Roles of Individual Components


Overbar with e: Ensemble-meanΔ: Future change of individual memberCTRL: Preindustrial control

CTRL CTRL

CTRL CTRL

CTRL CTRL

ConvGP ConvGP

ConvGP

ConvGP

WS

ee e

WS I

e e

WS I

e e

WS CP

f I I CP

f I I CP

f I I CP

δ(ABC)= δA∙(BC) + δB∙(AC) + δC∙(AB) (δ → 0)


ΔConvGPIζ

ΔConvGPIws

ΔConvGPCP

ΔGPIJη

ΔGPIJws

ΔGPIJpot

ΔGPIJH

HadCM3(DJFMAM)


ΔConvGPIζ

ΔConvGPIws

ΔConvGPCP

ΔGPIJη

ΔGPIJws

ΔGPIJpot

ΔGPIJH

HadSM3(DJFMAM)

Table summary (DJFMAM)

RegionCommon change Main driver

ConvGP GPI ConvGP GPI

Southwest Pacific Ocean

Decrease in the SPCZ Iζ > IWS > CPHadCM3: JWS, JH > Jpot, Jη

HadSM3: JWS, JH, Jη > Jpot

*Increase towards the northeast of the SPCZ (HadCM3) Iζ Jη

South Indian Ocean

Decrease in the SWI-MDR Iζ > IWS > CP JH, Jη > JWS

Northwest Pacific Ocean

*Increase (HadSM3) CP > Iζ, IWSJH > JWS, Jη > Jpot

Relatively consistent contribution from vorticity relative humidity

The model-dependent changes are marked with an asterisk (*).



Uncertainty arising from Genesis Parameter Formula

Which is more credible future in these models?Which is more credible future?

Main Conclusion 1

• The dynamic ocean coupling does affect the near equatorial region, but does not have any material influence on subtropical warming pattern

• The gross change patterns in tropical cyclone genesis parameters are essentially similar between dynamic ocean coupled (HadCM3) and ‘slab’ ocean coupled model (HadSM3)

• The large-scale changes does not look to be driven by the ocean-atmosphere coupled dynamics, though they look El Nino-like for Hadley Centre low-resolution coupled models - which has a thread of connection with the argument “Projected weakening of tropical circulation is not driven by the ocean-atmosphere coupled dynamics” by Vecchi and Soden (2007)


Main Conclusion 2

• Parameter uncertainty gets lower as far as the spatial change pattern is concerned, whilst structural uncertainties remain high

• The atmospheric physical parameter perturbations give a wide range of climate sensitivity (e.g. 2.19-7.11°C for seventeen HadSM3 runs), comparable (or even larger than) to that from the multi-model ensemble,

• the atmospheric physical parameter perturbations do not mess up the model’s standard spatial pattern of the tropical SST change (zonal and meridional SST gradients), which are of fundamental importance for the tropical atmospheric response pattern

• Fundamentally different from its effect on ENSO variability – which is the most dominant factor in determining ENSO variability in CGCMs! (Guilyardi et al 2004; Toniazzo et al 2008)


Minor Conclusions

• The flux adjustment is likely to affect the tropical SST warming pattern not as large as does dynamic ocean coupling (this is confirmed at least for standard HadCM3), whereas it does affect ENSO characteristics (Toniazzo et al 2008)

• The regions of future tropical cyclone changes are identified which appear to have little sensitivity to the sampled uncertainties

• The future changes in HadCM3 are more El-Nino-like than those in HadSM3, indicating differences in regional scale


Minor Conclusions

• Whilst the ConvGP emphasises relative vorticity’s contribution, the GPI brings the contributions of relative humidity and vertical wind shear in relief

• Differently formulated genesis parameters give rise to uncertainty, which is a limitation of this approach

•When it comes to changes in the regional total frequency, large uncertainty is concentrated in the NWP basin

•Need to do inter-comparison, as well as develop a better genesis parameter


Suggestions

• It is possible to attempt a probabilistic prediction based on a large ensemble of ‘slab’ ocean coupled models, with the use of scaling techniques to account for the characteristic differences between equilibrium and transient change patterns arising from different ocean representations (Harris et al 2006)

• The roles of large-scale diagnostics will be further investigated together with the direct detection of tropical cyclones in high-resolution models


The Met Office HQ in Exeter


The Met Office HQ in Exeter


History of the Met Office


• 1854: The Met Office is founded (First head: Vice-Admiral Robert FitzRoy).

• 1914: Lewis Fry Richardson imagines 64,000 people carrying out calculations in a vast hall and comments: "Perhaps some day in the dim future it will be possible to advance the computations faster than the weather advances. But that is a dream".

• 1922: Forecasts are first broadcast by BBC radio.

• 1944: The D-Day landings are planned but are postponed due to bad weather.

• 1962: Her Majesty the Queen performs the official opening ceremony of the new Headquarters at Bracknell. The Met Office takes delivery of its first electronic computer so that numerical forecast techniques can be applied operationally.

• 1981: The Met Office's first supercomputer — the Cyber 205 — is installed.

• 1982: The first global operational forecasting model is introduced to assist in operations for the Falklands War.

• 1990: The Met Office becomes an Executive Agency of the Ministry of Defense. The Met Office Hadley Centre for Climate Prediction and Research is opened.

• 2004: The Met Office's new headquarters in Exeter are officially opened and fully operational.

The Met Office Hadley Centrefor Climate Prediction and Research

• Opened in 1990, by the Prime Minister Margaret Thatcher

• The UK's foremost climate change research centre

• Largely co-funded by Defra (the Department for Environment, Food and Rural Affairs), the MOD and DECC (Department of Energy and Climate Change)

• One of the world's leading centres for climate change research

• About 200 staffs working in the headquarters

Happy 20th Birthday!© Crown copyright Met Office

Some Pictures around Southwest


“Hey, ya fergot La Yogurt”

“Eye-yaff -yalla-yok-ell”

“EY-ya-fyat-lah-YO-kut”

Thank you!Thank you!

Apri 17, 2010, Image taken by the MODIS on NASA’s Aqua SatelliteApri 17, 2010, Image taken by the MODIS on NASA’s Aqua Satellite(http://earthobservatory.nasa.gov/NaturalHazards/view.php?id=43690)(http://earthobservatory.nasa.gov/NaturalHazards/view.php?id=43690)

How to Pronounce?


Large Ensemble and Sampling Uncertainty

• A large (> 200) QUMP ensemble with HadSM3

•What can we do using this data?

• In principle, the contribution of a specific parameter to the model response (especially for climate sensitivity) can be back-traceable – a powerful point of the PPE approach

•But, ideally, tens of thousands of ensemble member is required to simulate all sets of parameter perturbed models - nearly impossible, but has been done and still doing (e.g. http://climateprediction.net)

•Sampling uncertainty arising from model formulation (i.e. ill-constrained physical parameters)


ΔConvGP: Results from 228 HadSM3 PPE Runs







ΔGPI: Results from 228 HadSM3 PPE Runs







Metric to Quantify the Uncertainty arising from Model Formulation


total model formulation internal variability of 2 CO2 internal variability of 1 CO2

model formulation

total model formulation internal variability

2 2 2 2

2

internal variability of 1 CO

X X X X

X

X X X

total internal variability of 2 CO2 internal variability of 1 CO2

2 2

2 2 2

2 2internal variability of 1 CO

X X X

Δ: Future changeX: Random Variableσ2: Variance

Uncertainty arising from Model Formulation

ConvGP

GPI

Absolute values of ensemble-mean future changes less than 0.1 are masked out© Crown copyright Met Office

Documents

© Crown copyright Met Office Future Changes in Tropical Cyclone Genesis Parameters in Hadley Centre Coupled Climate Models: A Low Resolution Model Study