33
© Crown copyright Met Office 2009 From “dynamical thinking” to parametrization progress…? Steve Derbyshire Thanks to: Sean Milton, Martin Willett, Rachel Stratton, Hongyan Zhu

From “dynamical thinking” to parametrization progress…?

Embed Size (px)

DESCRIPTION

From “dynamical thinking” to parametrization progress…?. Steve Derbyshire Thanks to: Sean Milton, Martin Willett, Rachel Stratton, Hongyan Zhu. From dynamical thinking to (faster) parametrization progress?. Contents What limits our progress? Illustrations of our experience Perspectives. - PowerPoint PPT Presentation

Citation preview

Page 1: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office 2009

From “dynamical thinking” to parametrization progress…?Steve Derbyshire

Thanks to: Sean Milton, Martin Willett, Rachel Stratton, Hongyan Zhu

Page 2: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

From dynamical thinking to (faster) parametrization progress?Contents

1. What limits our progress?

2. Illustrations of our experience

3. Perspectives

Page 3: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

1. What really holds us back?

Page 4: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

The Advance of Science

Source:Anonymous

Page 5: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

Perhaps we’re limited by...• Detailed convective process knowledge?

• But NB lots more obs, CRMs etc..

• Capabilities of non-cloud resolving systems?• Much-rehearsed arguments• Cf. MetO involvement in CASCADE (~1.5km but aim to

drive conventional parametrization improvements too)

• Difficulties in mapping between process study and “full model” results?

?

Page 6: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

2. Illustrations of our experience

Page 7: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

Illustrations in NWP case-study sampleKnown problems identified in systematic NWP

verification (~ 2005) illustrated in small sample used for preliminary testing of changes:

• Humidity biases• Temperature biases• Excessive tropical circulations

(Following plots from Sean Milton/Martin Willett)

Page 8: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

Relative Humidity biases in 5-day forecast tests

Substantial dry bias in inner-tropical upper troposphere (up to 10% RH in places)

Page 9: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

Temperature biases in NWP 5-day forecast

Warm bias in most of tropical troposphere but cold bias at the top of convection (top too low?)

Page 10: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

Wind spin-up in 5-day forecasts

Page 11: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

Precipitation and wind spinup

S.Milton

Page 12: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

Expectations for convection parametrization sensitivities

Simple arguments:

• T, q biases seen in NWP and SCM results must have some impacts on dynamics and hydrological cycle• E.g. 0.5K convective warm bias cf. SST+0.5K?• Weak Temperature Gradient arguments

• Various other impacts • convective momentum transport • cloud-radiative effects

Page 13: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

Total heat and parametrization sensitivity

Total heat thinking (Emanuel-Neelin): tropical convection broadly response to accumulation of h=cpT+Lq+gz, exported largely via

Surface transfer ~ ch U h

Possible explanations for excessive convection:

Winds too strong? ch too high?Sea too warm? (cf. Emanuel)

BL

Cloud layer

Precipitating downdraught

DD too weak? Or carry insufficiently low h back into the BL (because saturated)? (cf. Raymond)

Convection exports insufficient h for given mass-flux (cf. Neelin’s Gross Moist Stability)?

Convective updraught

( / )h p dp

Page 14: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

EUROCS humidity case/motivation for adaptivity

Mass flux profiles in a range of free-tropospheric humidities

Cloud-Resolving Model SCM (UM4.5)

Derbyshire et al., QJRMS (2004) – EUROCS Special Issue

90%

70%50%

25%

[Numbers are for “humidity parameter RHt”]

Page 15: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

Specific changes Adaptive forced detrainment as extension of

Gregory-Rowntree forced detrainment• Replace 0.2K threshold by more physical scaling

so fractional detrainment ~ fractional decline in buoyancy

Page 16: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

Adaptive Detrainment – Relative Humidity impacts

(Preliminary NWP tests by Sean Milton/Martin Willett)

Page 17: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

Adaptive Detrainment – Temperature impacts

Page 18: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

Preliminary climate tests – J.Rodriguez

1-year annual mean precipitation (1990)

Adaptive mod seems to improve Tropical W. Pacific with benefits to Pacific wind-stress

Observations

HadGAM1 control

HadGAM1 with adaptive detrainment

Page 19: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

Impact on wind-stresses

(Rachel Stratton)

Page 20: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

3. Perspectives

Page 21: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

What we found

• We corrected much of the T, q biases using an improved convective detrainment that we wanted to implement anyway

• This also seemed largely to improve the excessive broadscale convection circulations

• Particular benefit to Pacific wind-stresses (and hence ENSO): robust ~30% reduction

• All broadly consistent with WTG thinking (which was part of the original motivation)

• But.. more mixed signals with e.g. MJO (ongoing work including H.Zhu, H.Hendon)

Page 22: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

Qs to dyn-conv community

• Was this a good use of WTG thinking (simple, impressionistic..)?

• How much consensus is there anyway about these feedbacks?

• What about other (e.g. rotational) modes? Handle v as well as u?

• Can the community do more to reach a truly consolidated understanding?

• Can/should parametrization developers more systematically harness dynamical thinking to accelerate progress?

Page 23: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

Questions and answers

Page 24: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

Supplementary slides

Page 25: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

Adaptive entrainment and detrainmentIn CRM study of deep convection Swann (QJ 2001) found:

Plume excess buoyancy ~ half adiabatic value

(schematic)

0

height

Plume buoyancy excess vex

adiabatic (no entrainment)

actual

Seems to imply entrainment and detrainment adapt to control vex

Page 26: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

Impact of shallow-Cu “microphysical” tests

(A.Lock; similar impact shown in aquaplanet by J.Petch)

(1) “No shallow” test: treat as deep convection(2) “Shallow precip” test: reduce precipitation threshold from 1 g/kg to 0.5 g/kg • Similar impacts in NWP tests•Seems to explain some past experience with convection changes via side-effects on shallow-Cu precipitation

latitude

+ +

-

Page 27: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

NWP tests of shallow-Cu precip change

Adrian Lock, using NWP Case Study Suite

Page 28: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

Slides from Australian UM partners (Zhu, Hendon,

CAWCR)

Page 29: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

Mean daily precipitation composited into 5 % bins of saturation fraction

OBS

SP-CAM

CAM

UM

Extension of Zhu et al. (JAS 2009).UM here is a version of UM6.3 (first attempt at a CAWCR version)

Page 30: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

Fig.1

NEW RESULTS! May 2009 with UM6.6 HadGEM3-A job agtgd

Page 31: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

UM-6.3

UM-6.6

Fig. 3

Page 32: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

UM-6.3

UM-6.6

Fig. 4

Page 33: From “dynamical thinking” to  parametrization progress…?

© Crown copyright Met Office

Hongyan’s summary • One year climate run analyzed with UM 6.6• Relationship between precipitation and saturation fraction has been

improved greatly comparing with the results from UM 6.3 (see Fig.1). According to the exponential relationship between precipitation and saturation fraction, I anticipated that the MJO should be better in UM6.6.

• I used the UM 6.6 10-year climate run from my college, Zhian Sun, who is working on radiation in ACCESS. The coherence spectrum between symmetric precipitation and U850 (Fig.2) shows that there is peak coherence for UM6.6 at wavenumber 1 with about 50 days period, consistent with the observed MJO. The symmetric U850 space-time power spectral for U850 (fig. 3) also exhibits a strong peak at wavenumber 1 and 50 periods, a good agreement with observation. For the precipitation (Fig.4), the power spectral peaks at wave number 2 and 3, but in the rather low frequency region (bigger than 50days). Even though, the eastward power is stronger than the westward power in UM 6.6, which was about the equal in UM 6.3 run. Observed MJO is characterized by a distinct spectral peak at eastward wavenumber 1-3 with period of 30-90 days for precipitation.

• Timing of latent heat flux cf. precip also improved (role of BL changes?)