© Crown copyright Met Office Met Office progress report Andy Brown WGNE, Tokyo, October 2010

© Crown copyright Met Office Outline Production systems Seasonal/climate Research issues Ocean-Atmosphere coupling Mid-level cloud Resolution and parametrization choices for convection-permitting models Scalability

© Crown copyright Met Office Met Office production systems

© Crown copyright Met Office DePreSys TIGGE ensemble GloSea4 MOGREPS-R ensemble Coupled atmos/ocean Earth System Timescale 36hrs48hrs5 days 15 days 6 months 10 years 30 years >100 years 1.5km 4km 12km 24km 40km 80km 150km 300km Complexity UKV UK4 NAE Global HadCM3 HadGEM1 HadGEM2 Global atmosphere-only Regional atmosphere-only HadGEM3-RA regional HadGEM3 Atmospheric grid length

© Crown copyright Met Office Operational NWP Models: Aug 2010 Global 25km 70L 4DVAR 60h forecast twice/day 144h forecast twice/day +24member EPS at 60km NAE 12km 70L 4DVAR 60h forecast 4 times per day +24member EPS at 18km UKV 1.5km 70L (variable resolution) 3DVAR 36h forecast 4 times per day

© Crown copyright Met Office Parallel Suite 22 : Nov 2009 NEW UK 1.5km Model NEW Seasonal Forecast Model based on HADGEM3 Global NWP Model to 70 Levels (as HADGEM3) Ensemble NWP Model to 60km (Global) and 18km (regional) Parallel Suite 23 : Mar 2010 Regional and Ensemble NWP Models to 70 Levels Global NWP Model to 25km Parallel Suite 24 : Jul 2010 NEW UK Air Quality Model with UKCA Chemistry PC2 cloud scheme to global NWP (as HADGEM3) Subgrid drainage shear parametrization to UK 1.5km model Parallel Suite 25 : Nov 2010 Global DA upgrades (N216; Covstats from EC ensemble) Seasonal forecast to L85 (troposphere as HADGEM3) Key Model Upgrades

© Crown copyright Met Office Chief Scientist Julia Slingo Director of Technology Alan Dickinson Deputy Director, Climate Science & Head Partnerships Chris Gordon Deputy Director, Foundation Science Andy Brown Deputy Director, Weather Science Brian Golding Science Management Technology and IS Observational based research Parametrizations Global model development Dynamics research Data assimilation Satellite applications Regional NWP Ocean forecasting Verification, Products, Applications Understanding climate change Monthly to decadal prediction Earth system Oceans Impacts and adaptation

© Crown copyright Met Office Ocean-atmosphere coupling

© Crown copyright Met Office Coupled NWP research: Background International Workshop on Upper Ocean-Atmosphere Interactions on Weather and Climate Timescales held at the Met Office (Dec 2009):

© Crown copyright Met Office Coupled NWP research: Motivation Potential additional predictability/skill from representing air-sea physical coupling (MJO, diurnal cycle, TCs…) Quantify cost-benefit for NWP skill Tackle and improve coupled process errors at "source – leading to improvement of systematic errors and drifts in model predictions for longer timescales Transpose-CMIP Seamless prediction: more unified experimental design and model science from NWP through seasonal to climate prediction timescales Framework for developing and applying new coupled data assimilation techniques

© Crown copyright Met Office Coupled NWP research: Status 1.Coupled configuration (currently running in early research mode): HadGEM3 N216L85 (A), ORCA 0.25L50 (O), CICE 0.25 (I) 3-hourly coupling (ocean vertical resolution permits diurnal mixed layer) Initialisation. A: Op. NWP analyses, O: FOAM analyses, I: Climatology Boundary conditions: Climatological surface/soil parameters, ozone; specified aerosol concentrations (model climatology) Atmosphere diagnostics: merger of seasonal-to-climate packages, allowing assessment of processes, variability and teleconnections, plus NWP packages allowing assessment of forecast verification and skill Ocean diagnostics: Operational FOAM packages, designed to output a comprehensive set of fields to validate the ocean surface and water column 2.Parallel atmosphere-only and ocean-only control experiments to allow the impact of coupling on the development of process-based errors to be isolated A: forced with OSTIA SSTs + persisted anomaly O: forced with 3-hourly mean fluxes from the atmosphere only model

© Crown copyright Met Office Mid-level cloud

© Crown copyright Met Office Missing midlevel cloud Williams et al Illingworth et al. ISCCP analysisCloudnet

© Crown copyright Met Office km Composite cyclones Rainrate mm/day km

© Crown copyright Met Office Composite midlatitude cyclone km SW difference

© Crown copyright Met Office CONSTRAIN: Cold air outbreak 31 Jan 2010

© Crown copyright Met Office TOA SW from UKV and global Lack of mid-level cloud common to many models (NWP and climate) Some evidence for congestus and layer cloud issues Detailed studies with observations and high resolution models showing promise UKV global CERES

© Crown copyright Met Office Resolution and parametrization choices for convection- permitting models

© Crown copyright Met Office Current operational regional models Regional North Atlantic European (NAE) 12 km MOGREPS-R regional ensemble 18 km -> 12 km UK variable (UKV) resolution model 1.5 km Future - Develop a 1.5 km downscale ensemble system Issues: Choice of 1.5 km small ensemble cf coarser resolution larger ensemble What to do with convection? Convective-scale ensembles Nigel Roberts

© Crown copyright Met Office First case Hailstorm in Ottery Dramatic thunderstorm Very localised flash floods in Otter Valley Courtesy of Ken Mylne

© Crown copyright Met Office MOGREPS output 00 UTC 30/10/08 T+18 - selected members xx w 950 hPa Nigel Roberts Different frontal structure and positioning in each MOGREPS-R (and hence UKV) member Downscaling ensemble can add value

© Crown copyright Met Office 2.2 km compared to 1.5 km Focus of convection is moved from coastal strip to over the sea Nigel Roberts Number of pixels exceeding the extreme threshold 50mm in 6 hours from all 24 members Limited testing supports choice of small 1.5km ensemble xxx Thanks to Changgui Wang

© Crown copyright Met Office Convection closure 4km CAPE limited CAPE closure 1.5km No convection scheme Both models add significant value to coarser models (surface forcing; shower advection) Convection decisions based on case-studies / verification scores

© Crown copyright Met Office 12/10/00 flood event Peak rainfall rates within 60x60km squares over 12 hours 4km gridlength forecasts mm/hour No convection scheme With convection scheme CAPE dependent Too high Too low Closer to radar mm/hour mm/hour Example of testing of existing 4km convection closure

© Crown copyright Met Office Convection closure 4km CAPE limited CAPE closure 1.5km No convection scheme Both models add significant value to coarser models (surface forcing; shower advection) Convection decisions based on case-studies / verification scores Role for idealized work to supplement case study / verification work (metrics need care)? Relevance to global mid-level cloud issues?

© Crown copyright Met Office Scalability

© Crown copyright Met Office UM Scalability Cores Speedup

© Crown copyright Met Office Global Model Scalability Problems Most problems due to global lat-long grid ADI preconditioner for Helmholtz solver Advection communications and polar re-mapping Polar filtering Constant pole requirement ENDGAME dynamics expected to help Change to polar grid structure Ultimately expect to need to move away from lat-long I/O problems being addressed by I/O servers