BESM, looking into the future…

Chair: Paulo Nobre FAPESP, 19 February 2013

MOZART driven by BESM offline mode

OZONE

TOLUENE

T + 1 DAY T + 1 MONTH T=20060101, 00z

1/120

Competing demands of resolution, complexity, uncertainty, and long integrations in Climate System

Modelling:

Resolution

Complexity

Uncertainty

Long simulations

Courtesy: J. Shukla, IGES/COLA

Why a move to higher resolution is necessary

I: Complexity in the atmosphere

Water vapour and window channel radiances from Meteosat-7

Same fields from HadAM3 at climate resolution

Courtesy: J. Shukla, IGES/COLA

Why a move to higher resolution is necessary II: Complexity in the ocean

SSTs in the Gulf stream from infrared measurements aboard MODIS

10 …. and as simulated by OCCAM

1/40

1/120

Courtesy: J. Shukla, IGES/COLA

Equatorial Pacific sea surface temperatures (0C) from UK Ocean Model

1/12o

1/4o

1o

Courtesy: J. Shukla, IGES/COLA

BESM Ocean Model SST 1/10° global grid

HAND Model-Landscape Descriptor, A. D. Nobre, 2010

Weather Prediction Model of ~2020

Coupled Ocean-Land-Atmosphere Model

~1 km x ~1 km

100 levels

Unstructured, adaptive grids

Landscape-resolving

(~100 m)

~1 km x ~1 km

50 levels

Unstructured, adaptive grids

Assumption: Computing power enhancement by a factor of 103-104

Courtesy: J. Shukla, IGES/COLA

MCT/INPE-REDE CLIMA-FAPESP Supercomputer for Climate Change Research

Sustained

Throughput

15 Tflops

Main Memory 20 TBytes

Primary Storage 3 PBytes

Aquisition

Installation

Late 2010

Total budget US$ 25 M

and will make it possible to run global climate model simulations at high

spatial resolutions to grid sizes of 20 km !

BESM Supercomputer Performance

BESM-OA2.3 CMIP5 would require the whole

30,000 processors of Cray EX6 during two

years to be completed

. AGCM2009SSIB

. MOM4p0

River Discharge

. AGCM2009SSIB

. MOM4p1

Topaz; SIS,

. AGCM2012IBIS

. MOM4p1

Topaz, SIS

. AGCM2012INLAND

. MOM4p1

Topaz, SIS

. AGCM2013INLAND,Chem

. MOM5

Topaz, SIS

AGCM2013Chem

. INLAND_THMB

. MOM5

Topaz, SIS

2010 2012 2013 2014 BESM-OA1.0 BESM-OA2.3

BESM-Ibis2.5

BESM-Inland2.7 BESM-Mozart2.9 BESM 3.0

T062L28 atmos 1.0 L50 ocean

BESM Version Releases Plan

T126L64 atmos ¼ L64 oceans

The greatest Challenge:‘Peopleware’ Model Component Present* In 5 years* In 10 years*

Atmosphere 15 + 10 30 + 15 45 + 30

Land 20 + 25 30 + 50 50 + 70

Chemistry 2 + 2 10 + 20 25 + 30

Ocean 8 + 5 20 + 40 30 + 80

TOTAL 45 + 42 90 + 125 150 + 210

* Researchers + Students/Collaborators

• Long term research programs: FAPESP Research Program on Global Climate Change; Rede CLIMA; INCT for Climate Change • 10 Doctoral programs supporting capacity building in Earth System Modeling • Annual “International Summer Schools” will engage some 40 doctoral students/post-docs from S. America, S. Africa and India fellows per year.

Challenges Ahead

• Building a trully interactive science-policy making-private sectors network that is capable to understand and use the products of the BESM for decision making.

• Bringing the whole of the scientific community, professors & students, in Brazil and “Southern Hemisphere” countries to cooperate for that end.