£1.2 billion for the world’s most powerful weather and climate supercomputer

Pre­dic­ting seve­re wea­ther and the impacts of cli­ma­te chan­ge will be fas­ter and more accu­ra­te than ever before.

Published 17 Febru­ary 2020
  • £1.2 bil­li­on invest­ment con­firm­ed for sta­te-of-the-art super­com­pu­ter to impro­ve seve­re wea­ther and cli­ma­te forecasting
  • the latest super­com­pu­ting tech­no­lo­gy will unleash the full poten­ti­al of wea­ther and cli­ma­te data for the UK
  • data from the super­com­pu­ter will be used to inform govern­ment poli­cy as part of lea­ding the glo­bal fight against cli­ma­te chan­ge and mee­ting net zero emis­si­on targets

 

Pre­dic­ting seve­re wea­ther and the impacts of cli­ma­te chan­ge will be fas­ter and more accu­ra­te than ever befo­re, thanks to con­fir­ma­ti­on of £1.2 bil­li­on govern­ment fun­ding to deve­lop a sta­te-of-the-art super­com­pu­ter, Busi­ness and Ener­gy Secre­ta­ry and COP26 Pre­si­dent Alok Shar­ma announ­ced today (17 Febru­ary 2020).

Data from this new super­com­pu­ter – expec­ted to be the world’s most advan­ced dedi­ca­ted to wea­ther and cli­ma­te – will be used to help more accu­ra­te­ly pre­dict storms, sel­ect the most sui­ta­ble loca­ti­ons for flood defen­ces and pre­dict chan­ges to the glo­bal climate.

The new super­com­pu­ter, to be mana­ged by the Met Office, will also be used to help ensu­re com­mu­ni­ties can be bet­ter pre­pared for wea­ther dis­rup­ti­on, inclu­ding through:

  • more sophisti­ca­ted rain­fall pre­dic­tions, hel­ping the Envi­ron­ment Agen­cy rapidly deploy mobi­le flood defences
  • bet­ter fore­cas­ting at air­ports so they can plan for poten­ti­al disruption
  • more detail­ed infor­ma­ti­on for the ener­gy sec­tor to help them miti­ga­te against poten­ti­al ener­gy black­outs and surges

With the govern­ment announ­cing its Year of Cli­ma­te Action, the news fur­ther demons­tra­tes the UK is lea­ding by exam­p­le ahead of hos­ting UN cli­ma­te con­fe­rence COP26, whe­re the world will meet to agree more ambi­tious action.

Busi­ness and Ener­gy Secre­ta­ry and COP26 Pre­si­dent Alok Shar­ma said:

Over the last 30 years, new tech­no­lo­gies have meant more accu­ra­te wea­ther fore­cas­ting, with storms being pre­dic­ted up to 5 days in advance.

Come rain or shi­ne, our signi­fi­cant invest­ment for a new super­com­pu­ter will fur­ther speed up wea­ther pre­dic­tions, hel­ping peo­p­le be more pre­pared for wea­ther dis­rup­ti­on from plan­ning tra­vel jour­neys to deploy­ing flood defences.

The new super­com­pu­ter will also streng­then the UK’s super­com­pu­ting and data tech­no­lo­gy capa­bi­li­ties, dri­ving for­ward inno­va­ti­on and gro­wing world-class skills across super­com­pu­ting, data sci­ence, machi­ne lear­ning and arti­fi­ci­al intelligence.

Pro­fes­sor Pen­ny Enders­by, Met Office Chief Exe­cu­ti­ve said:

This invest­ment will ulti­m­ate­ly pro­vi­de ear­lier more accu­ra­te war­ning of seve­re wea­ther, the infor­ma­ti­on nee­ded to build a more resi­li­ent world in a chan­ging cli­ma­te and help sup­port the tran­si­ti­on to a low car­bon eco­no­my across the UK.

It will help the UK to con­ti­nue to lead the field in wea­ther and cli­ma­te sci­ence and ser­vices, working col­la­bo­ra­tively to ensu­re that the bene­fits of our work help govern­ment, the public and indus­try make bet­ter decis­i­ons to stay safe and thrive.

We wel­co­me this plan­ned invest­ment from UK government.

Chair of the Sci­ence Review Group Pro­fes­sor Ted She­p­herd said:

The agree­ment to upgrade the Met Office high per­for­mance com­pu­ter is wel­co­me news. The impro­ved pro­ces­sing power will deli­ver a step-chan­ge in wea­ther fore­cas­ting and cli­ma­te model­ling capa­bi­li­ty for the UK, such as the fur­ther deve­lo­p­ment of the Earth Sys­tems Model, which invol­ves col­la­bo­ra­ti­on with the many UKRI-NERC fun­ded rese­arch centres.

Impro­ved dai­ly to sea­so­nal fore­casts and lon­ger-term cli­ma­te pro­jec­tions will equip socie­ty with a grea­ter abili­ty to proac­tively pro­tect its­elf against the adver­se impacts of cli­ma­te change.

The Met Office is at the fore­front of super­com­pu­ting, using its cur­rent tech­no­lo­gy to dri­ve advan­ces in envi­ron­men­tal forecasting.

As a result, detail­ed wea­ther pre­dic­tions for the UK now take place every hour ins­tead of every 3 hours, pro­vi­ding cru­cial and time­ly updates when extre­me wea­ther is approaching.

The bene­fit of this has been felt recent­ly: major storms Cia­ra and Den­nis, and the ‘Beast from the East’ in 2018, were fore­cast 5 days in advan­ce, enab­ling local coun­cils and emer­gen­cy ser­vices to prepa­re and ins­ti­ga­te resi­li­ence plans. Simi­lar­ly, the Envi­ron­ment Agen­cy has used the Met Office’s latest UK cli­ma­te pro­jec­tions set out poten­ti­al future floo­ding sce­na­ri­os and how fun­ding can be best allocated.

UK supercomputer breakthroughs

Today, the govern­ment also announ­ced £30 mil­li­on invest­ment for advan­ced super­com­pu­ting ser­vices, pro­vi­ding rese­ar­chers with access to the latest tech­no­lo­gy and expert soft­ware engi­neers. It will also help them speed up sci­en­ti­fic breakth­roughs like deve­lo­ping ‘food fin­ger­prin­ting’ to detect che­mi­cal con­ta­mi­nants in food and impro­ving drug design.

The fun­ding will sup­port 7 High Per­for­mance Com­pu­ting (HPC) ser­vices run by uni­ver­si­ties from across the UK, inclu­ding Queen’s Uni­ver­si­ty Bel­fast, the Uni­ver­si­ty of Edin­burgh, and Dur­ham Uni­ver­si­ty. The ser­vices will pro­vi­de rese­ar­chers with inva­luable access to powerful sys­tems to sup­port ground-brea­king work in are­as from Arti­fi­ci­al Intel­li­gence, ener­gy sto­rage and sup­p­ly, and the­ra­peu­tic drug design, as well as boos­ting the skills of UK scientists.

UK govern­ment Minis­ter for Scot­land Dou­glas Ross said:

The UK govern­ment invest­ment in Edinburgh’s super­com­pu­ters helps keep our capi­tal at the fore­front of cut­ting edge technology.

The Uni­ver­si­ty of Edin­burgh faci­li­ty will bene­fit sci­en­tists from across the UK as they are given the oppor­tu­ni­ty to use this new tech­no­lo­gy. This addi­tio­nal fun­ding builds on the work of the Edin­burgh and South East Scot­land City Regi­on Deal which is crea­ting world-lea­ding hubs for AI research.

The UK govern­ment is com­mit­ted to com­bat­ting the impact of cli­ma­te chan­ge on top of crea­ting thou­sands of high-ear­ning jobs and ensu­ring busi­nesses and public ser­vices in the UK are the first to bene­fit from the latest innovations.

Notes to editors

The govern­ment invest­ment will replace Met Office super­com­pu­ting capa­bi­li­ties over a 10-year peri­od from 2022 to 2032.

The cur­rent Met Office Cray super­com­pu­ters reach their end of life in late 2022.

The first pha­se of the new super­com­pu­ter will increase the Met Office com­pu­ting capa­ci­ty by 6‑fold alo­ne. The Met Office will look to deli­ver at least a fur­ther 3 times increase in super­com­pu­ting capa­ci­ty for years 6–10.

£1.2 bil­li­on refers to the total expec­ted invest­ment from govern­ment. The expec­ted con­trac­tu­al value for the super­com­pu­ting capa­bi­li­ty is £854 mil­li­on. Other cos­ts include invest­ment in the Obser­va­tions Net­work, exploi­ting the capa­bi­li­ties of the super­com­pu­ter and the pro­gram­me office costs.

2020 marks the 30th anni­ver­sa­ry of the estab­lish­ment of the Met Office Had­ley Cent­re for Cli­ma­te Sci­ence, working at the fore­front of cli­ma­te sci­ence and pio­nee­ring research.

Met Office Supercomputer Case studies

Case stu­dy 1: The Met Office sup­port­ed the respon­se to the Toddbrook Reser­voir inci­dent and the pro­tec­tion of resi­dents of Wha­ley Bridge. It work­ed direct­ly with the emer­gen­cy ser­vices and orga­ni­sa­ti­ons invol­ved in the emer­gen­cy respon­se advi­sing on short-term and long-term wea­ther fore­cast con­di­ti­ons. This was cri­ti­cal as fur­ther rain­fall would impact the water level in the Dam’s reservoir.

Case stu­dy 2: In a glo­bal first, the govern­ment brought tog­e­ther the Met Office, the Uni­ver­si­ty of West Vir­gi­nia, the Uni­ver­si­ty of Mary­land and NASA to help save lives by accu­ra­te­ly pre­dic­ting cho­le­ra hot­spots. The Met Office pro­vi­ded gui­dance on fore­cas­ted rain­fall in the coun­try for 14 days in advan­ce, to help UNICEF and Oxfam tar­get their on the ground efforts to pre­vent lar­ge out­breaks of cholera.

Case stu­dy 3: Afri­ca is one of the world’s most vul­nerable regi­ons to cli­ma­te chan­ge with mil­li­ons of peo­p­le rely­ing on rain­fall for agri­cul­tu­re. As part of the UK aid fun­ded IMPALA pro­gram­me, the Met Office led the deve­lo­p­ment of the first pan-Afri­can con­vec­tion-per­mit­ting model CP4-Afri­ca. This ground-brea­king rese­arch deve­lo­ped high-reso­lu­ti­on cli­ma­te pro­jec­tions that pro­vi­ded a glim­pse into future wea­ther and cli­ma­te extre­mes across Afri­ca, which were more seve­re than pre­vious­ly thought. The infor­ma­ti­on will help decis­i­on-makers redu­ce cli­ma­te-rela­ted risks.

Case Stu­dy 4: UK Cli­ma­te Pro­jec­tions (UKCP) pro­vi­des the most detail­ed pic­tu­re yet of future cli­ma­te in the UK. As part of this, UKCP Local (2.2km) pro­vi­des the most rea­li­stic set of pro­jec­tions of future cli­ma­te extre­mes like soaring tem­pe­ra­tures during the sum­mer and hour­ly sum­mer rain­fall for local are­as in the coming deca­des. This enhan­ced detail could help inform future risk assess­ments and local decis­i­on-making on the future of cli­ma­te chan­ge. The UKCP tool­kit will be used to inform the next Cli­ma­te Chan­ge Risk Assess­ment, due in 2022, and is being wide­ly used by local aut­ho­ri­ties, indus­try and aca­de­mia to help inform plans to mana­ge future cli­ma­te risks and enhan­ce resilience.

High Performance Computing case studies

The Materials and Molecular Modelling Hub

Led by: UCL

Part­ners: Queen Mary Uni­ver­si­ty of Lon­don, Queen’s Uni­ver­si­ty Bel­fast, Bru­nel Uni­ver­si­ty, Impe­ri­al Col­lege Lon­don, King’s Col­lege Lon­don, Uni­ver­si­ties of Cam­bridge, Lin­coln, Kent, Rea­ding, Sout­hamp­ton and York

EPSRC sup­port: £4.5 million

Mate­ri­als are at the heart of almost every modern tech­no­lo­gy, inclu­ding ener­gy gene­ra­ti­on, sto­rage and sup­p­ly, trans­por­ta­ti­on, elec­tro­nic devices, defence and secu­ri­ty, health­ca­re, and the envi­ron­ment. It is mate­ri­als that place prac­ti­cal limits on effi­ci­en­cy, relia­bi­li­ty and cost. The MMM Hub pro­vi­des high per­for­mance com­pu­ting capa­ci­ty for rese­ar­chers to car­ry out ground-brea­king rese­arch on the pro­per­ties of new and exis­ting mate­ri­als, and this fun­ding will build on the hub’s capability.

The­se include under­stan­ding and pre­ven­ting sur­face degra­da­ti­on, such as cor­ro­si­on and wear, on a ran­ge of dif­fe­rent mate­ri­als; rese­ar­ching how chan­ges to the recy­cling of metals can redu­ce the envi­ron­men­tal dama­ge cau­sed by metal extra­c­tion; and deve­lo­ping the next gene­ra­ti­on of mate­ri­als for solar ener­gy generation.

GW4 Tier‑2 HPC Centre for Advanced Architectures (Isambard 2)

Led by: The GW4 Alli­ance of the Uni­ver­si­ties of Bath, Bris­tol, Car­diff and Exe­ter, and hos­ted by the Met Office

EPSRC sup­port: £4.1 million

The Isam­bard 2 ser­vice will use the very latest tech­no­lo­gy from the UK-based Arm Hol­dings to pro­vi­de sci­en­tists with a world-class High Per­for­mance Com­pu­ting ser­vice. The same tech­no­lo­gy is expec­ted to be used in some of the first super­com­pu­ters capa­ble of a bil­li­on bil­li­on cal­cu­la­ti­ons per second, cal­led Exas­ca­le super­com­pu­ters. This tech­no­lo­gi­cal step chan­ge will requi­re sci­en­tists to adapt their codes in order to run as quick­ly and effi­ci­ent­ly as pos­si­ble and thus acce­le­ra­te sci­en­ti­fic dis­co­very. Isam­bard 2 will suc­ceed the cur­rent Isam­bard sys­tem and will help enable Bri­tish rese­ar­chers to prepa­re their codes for the wide­spread use of Exas­ca­le systems.

Isam­bard has alre­a­dy been used to inves­ti­ga­te poten­ti­al drugs to tre­at osteo­po­ro­sis and simu­la­te Parkinson’s dise­a­se at the mole­cu­lar level. Isam­bard 2 will enable rese­ar­chers to expand this fur­ther with the poten­ti­al for sci­en­ti­fic breakthroughs.

Kelvin‑2

Led by: Queen’s Uni­ver­si­ty Bel­fast and Uls­ter University

EPSRC sup­port: £2.1 million

The Kelvin‑2 ser­vice will pro­vi­de access to an enhan­ced com­pu­ting faci­li­ty focu­sed on arti­fi­ci­al intel­li­gence-based rese­arch. The pro­ject will initi­al­ly focus on acce­le­ra­ting rese­arch in 6 spe­cia­list are­as which are eco­no­mic­al­ly and soci­al­ly important to the UK. This includes neu­ro­tech­no­lo­gy and com­pu­ta­tio­nal neu­ro­sci­ence, inclu­ding work on brain-com­pu­ter inter­faces and hete­ro­ge­neous cata­ly­sis, such as model­ling che­mi­cal pro­ces­ses which con­tri­bu­te to the pro­duc­tion of items used in ever­y­day life.

The are­as also include inno­va­ti­ve drug deli­very for impro­ving drug-based the­ra­pies and for use in dia­gno­stics, as well as a focus on pre­cis­i­on medi­ci­ne whe­re auto­ma­ted tools will be crea­ted to ana­ly­se data and iden­ti­fy indi­ca­tors for health con­di­ti­ons. The­re will also be a focus on food fin­ger­prin­ting, inclu­ding tech­ni­ques for detec­ting che­mi­cal con­ta­mi­nants in food; and hydro­gen defla­gra­ti­on to assist with deve­lo­ping acci­dent pre­ven­ti­on and miti­ga­ti­on for hydro­gen tanks.

JADE: Joint Academic Data Science Endeavour – 2 (JADE 2)

Led by: Uni­ver­si­ty of Oxford

EPSRC sup­port: £5.5 million

Part­ners: Uni­ver­si­ties of Bath, Bris­tol, Cam­bridge, Exe­ter, Lan­cas­ter, Leeds, Lough­bo­rough, Shef­field, Sout­hamp­ton, Surrey, Sus­sex, War­wick and York, Queen Mary Uni­ver­si­ty of Lon­don, King’s Col­lege Lon­don, Impe­ri­al Col­lege Lon­don, UCL, New­cast­le Uni­ver­si­ty, The Alan Turing Insti­tu­te, Hartree Cent­re. The JADE 2 ser­vice, hos­ted at STFC’s Hartree Cent­re, will be a uni­que natio­nal resour­ce pro­vi­ding a sta­te-of-the-art GPU (Gra­phics Pro­ces­sing Unit) com­pu­ting faci­li­ty for rese­arch into Arti­fi­ci­al Intel­li­gence (AI)/Machine Lear­ning and Mole­cu­lar Dyna­mics, a com­pu­ter simu­la­ti­on method for ana­ly­sing the phy­si­cal move­ments of par­tic­les that make up molecules.

The AI tech­ni­ques deve­lo­ped using the ser­vice will have impact in a ran­ge of sec­tors inclu­ding finan­cial ser­vices, manu­fac­tu­ring, ener­gy and health­ca­re. The Mole­cu­lar Dyna­mics rese­arch con­duc­ted on JADE 2 will advan­ce under­stan­ding of the struc­tu­re and func­tion of lar­ge bio­lo­gi­cal mole­cu­les, many of which are tar­gets for the­ra­peu­tic agents for a lar­ge varie­ty of health con­di­ti­ons. The ser­vice will pro­vi­de a valuable com­pu­ting resour­ce to the new UKRI Arti­fi­ci­al Intel­li­gence Cen­tres for Doc­to­ral trai­ning, and thus pro­vi­de a cri­ti­cal com­pu­ta­tio­nal capa­ci­ty nee­ded to deve­lop the next gene­ra­ti­on of experts in AI.

Cirrus Phase II: Preparing for Heterogeneity at Exascale

Led by: The Edin­burgh Par­al­lel Com­pu­ting Cent­re at the Uni­ver­si­ty of Edinburgh

EPSRC sup­port: £3.5 million

Cir­rus Pha­se II will expand the capa­bi­li­ties of the Cir­rus ser­vice by adding spe­cia­li­zed GPUs to the cur­rent sys­tem. GPUs are com­mon­ly used as graphics/video cards in mobi­le pho­nes, per­so­nal com­pu­ters and games con­so­les. Howe­ver, spe­cia­li­zed GPUs can be also be used in super­com­pu­ters as acce­le­ra­tors enab­ling them to run nume­ri­cal cal­cu­la­ti­ons more quick­ly. The tech­no­lo­gy used in Cir­rus is expec­ted to be used in some of the first Exas­ca­le super­com­pu­ters and will allow sci­en­tists to test and adapt their code for model­ling and simu­la­ti­on to be rea­dy to advan­ce dis­co­very and inno­va­ti­on as soon as Exas­ca­le sys­tems beco­me available. It will help to ensu­re the UK has a sup­p­ly of indi­vi­du­als trai­ned with the­se spe­cia­li­sed skills and could lead to far more rapid and detail­ed dis­co­veries in new are­as and the pro­jects Cir­rus has alre­a­dy sup­port­ed, such as model­ling pro­te­in shape for bet­ter drug design and simu­la­ting tidal flows to opti­mi­se tur­bi­ne instal­la­ti­ons and their effects on sea beds. The new GPUs will also pro­vi­de a high-per­for­mance plat­form for AI trai­ning and rese­arch, a cri­ti­cal and rapidly gro­wing area.

Northern Intensive Computing Environment (NICE)

Led by: Dur­ham Uni­ver­si­ty as part of the N8 Rese­arch Part­ner­ship (Dur­ham and New­cast­le Uni­ver­si­ties, Uni­ver­si­ties of Lan­cas­ter, Leeds, Liver­pool, Man­ches­ter, Shef­field and York)

EPSRC sup­port: £3.1 million

The new NICE ser­vice will use the same tech­no­lo­gy as that used in the cur­rent lea­ding super­com­pu­ters in the world, exten­ding the capa­bi­li­ty of acce­le­ra­ted com­pu­ting. The tech­no­lo­gy has been cho­sen with the aim of com­bi­ning expe­ri­men­tal, model­ling and machi­ne lear­ning approa­ches, brin­ging the­se com­mu­ni­ties tog­e­ther to address new challenges.

This will mean that machi­ne lear­ning can be used along­side model­ling and simu­la­ti­on to bet­ter under­stand the vast data sets now being gene­ra­ted by expe­ri­men­ta­lists through, for exam­p­le, natio­nal faci­li­ties such as Dia­mond and the Hen­ry Roy­ce Insti­tu­te and inter­na­tio­nal faci­li­ties such as the Euro­pean Syn­chro­tron Radia­ti­on Faci­li­ty. This approach will enable sci­en­tists to, for exam­p­le, advan­ce the ima­ging tech­ni­ques neces­sa­ry to pro­du­ce the next gene­ra­ti­on of X‑ray instru­ments and deve­lop future stu­dents working with deep lear­ning tech­ni­ques at the inter­face of algo­rith­ms and High Per­for­mance Computing.

Cambridge Service for Data Driven Discovery (CSD3) – A National Data Intensive Science Cloud for Converged Simulation, AI and Analytics

Led by: Uni­ver­si­ty of Cambridge

Part­ners: Car­diff Uni­ver­si­ty, King’s Col­lege Lon­don, Uni­ver­si­ties of Edin­burgh, Lei­ces­ter, Oxford and Southampton

EPSRC sup­port: £4 million

The Cam­bridge Ser­vice for Data Dri­ven Dis­co­very is a £10 mil­li­on invest­ment, crea­ting one of the most powerful aca­de­mic super­com­pu­ters in the UK spe­ci­al­ly desi­gned to com­bi­ne lar­ge sca­le data inten­si­ve simu­la­ti­on and AI sci­ence within a sin­gle com­pu­ter system.

This breakth­rough super­com­pu­ter was deve­lo­ped via a uni­que Co-design part­ner­ship bet­ween the Uni­ver­si­ty of Cam­bridge, Dell, Intel, NVIDIA, Mel­lan­ox Tech­no­lo­gies and StackHPC a lea­ding UK SME deve­lo­ping ground brea­king HPC sys­tem soft­ware. It will acce­le­ra­te rese­arch across a wide ran­ge of engi­nee­ring and phy­si­cal sci­ence the­mes, inclu­ding mate­ri­als sci­ence and com­pu­ta­tio­nal che­mis­try, health infor­ma­tics, medi­cal ima­ging and bio-simu­la­ti­on, and AI and machi­ne lear­ning. The pro­po­sal has sti­mu­la­ted signi­fi­cant indus­tri­al invest­ment from Dell, Intel and Cam­bridge to form the Cam­bridge Open Exas­ca­le Lab, pro­vi­ding a lar­ge cri­ti­cal mass of peo­p­le and tech­no­lo­gy focu­sed on dri­ving UK com­pe­ti­ti­ve­ness in the rapidly emer­ging exas­ca­le com­pu­ting landscape.

This ser­vice will also be sup­port­ed by an addi­tio­nal £3 mil­li­on invest­ment from the Sci­ence and Tech­no­lo­gy Faci­li­ties Coun­cil, Medi­cal Rese­arch Coun­cil and UK Ato­mic Ener­gy Aut­ho­ri­ty (Cul­ham Cent­re for Fusi­on Energy).