© Crown copyright Met Office Implementation of a new dynamical core in the Met Office Unified Model Andy Brown, Director of Science.

© Crown copyright Met Office

Implementation of a new dynamical core in the Met Office Unified ModelAndy Brown, Director of Science

Met Office science strategy:Unified science and modelling for unified prediction


DePreSys

TIGGE ensemble

GloSea4

MOGREPS-Rensemble

Coupled atmos/ocean

Earth System

Timescale36hrs 48hrs 5 days 15 days 6 months 10 years 30 years >100 years

1.5km

4km

12km

24km

40km

80km

150km

300km

Comple

xity

UKV

UK4 NAE

Global

HadCM3

HadGEM1

HadGEM2

Global atmosphere-only

Regional atmosphere-only

HadGEM3-RA regional HadGEM3

Atmospheric grid length


A new dynamical core: ENDGame Atmos Physics 1

Advection

Atmos Physics 2

HelmholtzSolver

x2

x2

• Evolution of existing dynamics• Same equation set & variables• Same horizontal staggering (Arakawa C-grid)• Same vertical staggering (Charney-Phillips)• Iterative Semi-implicit semi-Lagrangian• Simpler solver• Less damping ( = 0.55)• Fully 3D advection for θ • Reformulated continuity equation

More scalable More stable More accurate

A possible timeline for a major model upgrade........

Quality

Time(years)

Operational

Development Model


Model changes alongside ENDGame

• Package of physics changes• ENDGame+physics=

• GA6.0 (climate)• GA6.1 (weather)• GA6.0 and 6.1 extremely similar (more seamless than

ever)

• For global NWP• Model N512 (≈25km) → N768 (≈17km) • 4D-Var inner loop N216 (≈60km) → N320 (≈40km) • Satellite package

ENDGame results


Eddy Kinetic EnergyT+72hr EKE from 10 x ECMWF cases

Paul Earnshaw

9.0

7.0N96 N216 N512 N768


Extra-tropical circulationEquivalent plots from PS32-based trials

See poster by David Walters

July

201

2


Tropical circulationSignificant increases in variability

Equatorial precip frequency/wavenumber spectrum from 20yr climate runs:

Prince Xavier

GA4.0GA6.0TRMM obs


Tropical circulation Major improvements in Tropical Cyclone forecasts

Impact of science upgradeN512 GA6.1 vs N512 GA3.1

Impact of science + resol’nN768 GA6.1 vs N512 GA3.1

Mean abs. PMSL error reduction 3.0 hPa 3.6 hPa

Central pressure reduction 7.1 hPa 11.1 hPa

Mean abs. |v| error reductions 6.7 kt 9.0 kt

Mean max. |v| increase 8.9 kt 13.4 kt

850hPa vorticity increase 79% 155%

Track error reduction 7.3% 8.6%*

Track skill score increase 3.8% 4.5%

Julian Heming, Keith Williams

* Biggest reduction in TC track error in a single UM upgrade for 20 years!

Average impacts from PS32-based trials:


New global versus 12km regional model

Surface weather impacts(Temperature, cloud cover, cloud base, ppn, visibility)

New global versus old global

Significant improvement in near-surface weather measures

Global outperforms 12km Limited Area Model – now retired

Global significantly behind 1.5km model for UK


Outgoing SW radiation

New New minus Old

Old minus climatology New minus climatology

Dan Copsey

ENSO



Kilometre-scale modellingCloser to a centred semi-implicit approach

Impact in UKV Lee Wave test: New DynamicsENDGame

Simon Vosper


DePreSys

TIGGE ensemble

GloSea4

MOGREPS-Rensemble

Coupled atmos/ocean

Earth System

Timescale36hrs 48hrs 5 days 15 days 6 months 10 years 30 years >100 years

1.5km

4km

12km

24km

40km

80km

150km

300km

Comple

xity

UKV

UK4 NAE

Global

HadCM3

HadGEM1

HadGEM2

Global atmosphere-only

Regional atmosphere-only

HadGEM3-RA regional HadGEM3

Atmospheric grid length

Dynamics matters!

Extremely significant model upgradeachieved in seamless framework


Questions?


Physics changes

Significant physics differences between GA3.1 and GA6.1:

Total of 66 tickets (including ENDGame) in GA4, GA5 and GA6


ENDGame in a nutshell Example of scalability vs New Dynamics

• ENDGame required for 17km in PS34

• N768 GA6.1 on 96 nodes currently runs in ≈ 40 mins

• Allows further resolution upgrades over next 10 years

Andy Malcolm, Paul Selwood

© Crown copyright Met Office Implementation of a new dynamical core in the Met Office Unified Model Andy Brown, Director of Science.

Documents

© Crown copyright Met Office Implementation of a new dynamical core in the Met Office Unified Model Andy Brown, Director of Science.