WMO Anticipated advances in Numerical Weather Prediction, and the growing technology gap in weather forecasting Ken Mylne Chair DPFS (Data-processing and Forecasting Systems) & Chair SWFDP SG WMO; WDS

du = ∂p – fv dt ∂x dv = ∂p + fu dt ∂y p = RT ρ Interpretation, Risk Analysis & Communication Knowledge 70 levels 25km 80km high The Role of the GDPFS in creating weather services – Forecast Model Observations GDPFS = Global Data Processing and Forecasting System

Core Foundation Systems of the GDPFS  Nowcasting – up to ~6h forecasting  Global NWP (numerical weather prediction)  Deterministic – single best guess forecast  Ensemble – probabilistic forecast for risk estimation  Resolution now ~16-45km  Limited area NWP  Convection-permitting – high resolution, local area  Deterministic & Ensemble  Coupled Seasonal and climate models  Numerical ocean wave and storm surge prediction

© Crown copyright Met Office 1.5km model Up to 36hr f/c 6-hourly update Future planned 2km nowcast Up to 6hr f/c ≤ hourly update 4km model Up to 120hr f/c 6-hourly update ~17km model Up to 144hr f/c 6-hourly update ~35km ensemble Up to 3day f/c 6-hourly update ~60km coupled model Up to 6 months Daily lagged ensemble ~2km ensemble Up to 36hr f/c 6-hourly update 2013 model domains

UKV and MOGREPS-UK  1.5km 70L (40km model top)  3DVAR (3 hourly)  36hr forecast  4 times per day  12-member EPS - 2.2km 4x/day 36h 2013 NWP Models in Suite Global and MOGREPS-G  25km 70L (80km model top) (17km from PS34)  Hybrid 4DVAR – 60km  66hr forecast twice/day  144hr forecast twice/day  12-member EPS - 33km 4x/day 72hr  24-member EPS - 60km 2x/day 15d  Ocean coupled up to 6 months © Crown copyright Met Office Euro4  4km 70L (40km model top)  Global downscaler  66hr forecast twice/day  144hr forecast twice/day

Specialized Activities of the GDPFS  Forecasting hydrometeorological hazards  Seasonal to sub-seasonal Climate prediction  Tropical cyclone forecasting  Volcanic ash advisory services  Response to environmental emergencies  Nuclear & non-nuclear incidents  Marine pollution incidents  Sand and dust storm forecasting  Non-real-time verification and testing

GDPFS Centres  Network of Global, Regional and Specialised Centres See paper for full list of centres

Status of Global Models  Deterministic global models now running at c km grid-lengths  Ensembles at c km  Global coverage  Up to 15 days range (EPS)  Approaching resolutions of recent regional models  Capable of capturing much severe weather  Ensembles provide the probability part of “risk”

Global Ocean Wave Ensemble products  Probability of significant wave height over 4m  Wave EPS Meteogram

Met Office 15-day Global ensemble forecast evolution Gusts > 60kt – damaging strong winds in Europe From 8 days to 2 days ahead  Low probability early warning  Consistent guide to most likely location from at least 5 days ahead  Higher resolution models even better

Tropical Cyclone Products  Storm tracks and strike probabilities  Experimental TC intensity forecasts

Global Hazard Map Summary map to track features through days of forecast Daily map to overlay different hazards and vulnerability layers

Biggest Weakness – (Tropical) Convection  Useful guidance, especially when calibrated  SWFDP feedback is positive even in the Tropics, but…

Convection-Permitting Models Rapidly replacing the old regional models © Crown copyright Met Office Frames at 10min intervals With grid-lengths of 4km or less modern regional models partially resolve convection and give far better forecasts for convective severe weather … which dominates Tropical severe weather

Convection-permitting Ensembles  Probabilities for localised heavy rain  But …computationally very expensive  small domains Prob of Torrential Rain >16mm/h Not affordable for Tropics yet, but…

UK Floods 24 th November 2012 Radar MOGREPS-UK 2.2km Chance of Heavy Rain in the Hour UKV 1.5km

Meteo-France 2.5km Ensemble Radar rainfall Deterministic >20mm (Dark) Prob(20mm)>10% (Light)

Lake Victoria Model  4km deterministic models are becoming affordable over a limited number of additional domains  …with funding Met Office 4km UM  Run at Global Centre  Products supplied to Regional Centre and NMHSs

East Africa WRF model Tanzania Meteorological Agency  5km model run by Tanzania  Boundary conditions provided by DWD (German Met Service)

Global Remote Sensing  Observation systems provide basis for nowcasting Satellite Imagery ATD Lightning Detection

Rapid refresh models for nowcasting  Models are starting to be used with assimilation of radar data for nowcasting  Expected to replace extrapolation nowcasting over next few years in advanced centres

National Met Centres and the GDPFS  Each WMO Member (country) is required to have a National Meteorological Centre  A key NMC Function is preparation of National Severe Weather Warnings for protection of lives and property, as required by the Member.  To achieve this, NMCs should:  Be staffed and equipped to participate in the WWW (World Weather Watch)  be linked to the WIS (WMO Information System)  have capacity to exploit GDPFS products  Some NMCs have difficulty in this…

NMCs and the technology gap  Typical Advanced NMC  Co-located with Global/ Regional GDPFS centre  NWP Post-processing to produce automated routine forecasts  Forecasters freed to focus on High Impact Weather  Established procedures:  warnings  PWS communication for civil protection action  Typical Developing/LDC NMC  Little or no NWP capability  Experimental  Remote from NMC  Low resolution – similar to today’s global  No ingestion in products  Forecasters occupied with routine production  Limited or no procedures for warning dissemination and action

Why not provide NMHSs with Regional Models?  As noted above, a number of NMHSs have (or have had) regional NWP systems generously supplied by donors, mainly experimental  This is NOT generally an effective method of capacity building in an NMHS with limited resources:  Very complex to support effectively (see next slide)  Distracts scarce resources from core function of issuing warnings and communicating with civil protection decision-makers

Technical Requirements for operational NWP  Running effective regional NWP requires:  Large (expensive and power-hungry) supercomputer  Convection-permitting resolution (~4km or better) required to add benefit to today’s global models  Complex, reliable and high-bandwidth telecoms  import boundary conditions and observations  Advanced data assimilation  Hi-resolution observations  Large team of expert scientists and computer scientists for implementation, maintenance and upgrades  Expensive to recruit and difficult to retain once experienced  Best Concentrated in few Global/Regional Centres  NMCs then focus resources on Local Interpretation, Forecasting and Service Delivery

 Global NWP centres to provide available NWP/EPS and sat-based products, including in the form of probabilities, cut to the project window frame;  Regional centres to interpret information received from global centres, prepare daily guidance products (out to day-5) for NMCs, run limited-area model to refine products, maintain RSMC Web site, liaise with the participating NMCs;  NMCs to issue alerts, advisories, severe weather warnings; to liaise with user communities, and to contribute feedback and evaluation of the project;  NMCs have access to all products, and maintained responsibility and authority over national warnings and services. 26 Global Centres User communities, including Disaster Management authorities NMCsRSMC Pretoria SWFDP Cascading Forecasting Process – efficient delivery of GDPFS

Technical Requirements for SWFDP Benefit  Cascading of SWFDP products and guidance uses:  Simple web pages  Products as Images including animation  Requirements to benefit:  Web-browser  Low band-width connection  Forecaster and PWS training  Highly efficient and cost- effective Capacity Building

Thank you for your attention DISCUSSION