Developments to the Met Office Global and Regional ... - UCAR
Developments to the Met Office Global and Regional ... - UCAR
Developments to the Met Office Global and Regional ... - UCAR
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
<strong>Developments</strong> <strong>to</strong> <strong>the</strong> <strong>Met</strong> <strong>Office</strong><br />
<strong>Global</strong> <strong>and</strong> <strong>Regional</strong> Ensemble<br />
Prediction System (MOGREPS)<br />
Richard Swinbank, Warren Tennant, Sarah Beare, Ken Mylne, Neill Bowler,<br />
<strong>and</strong> o<strong>the</strong>rs<br />
© Crown copyright <strong>Met</strong> <strong>Office</strong><br />
GIFS-TIGGE/NCAR/NOAA workshop<br />
27 th June 2012, Boulder, CO, USA
© Crown copyright <strong>Met</strong> <strong>Office</strong><br />
Contents<br />
• Operational MOGREPS set-up <strong>and</strong><br />
2012 developments<br />
• MOGREPS-UK<br />
• Coupled ensemble forecasting<br />
• S<strong>to</strong>chastic physics<br />
• SST & soil moisture perturbations
Operational setup <strong>and</strong> 2012<br />
upgrades<br />
© Crown copyright <strong>Met</strong> <strong>Office</strong>
© Crown copyright <strong>Met</strong> <strong>Office</strong><br />
MOGREPS Model domains<br />
18km grid<br />
Up <strong>to</strong> 54hr<br />
6-hourly update<br />
2.2km grid (new)<br />
Up <strong>to</strong> 36hr<br />
6-hourly update<br />
60km grid<br />
Up <strong>to</strong> 72hr<br />
6-hourly update
© Crown copyright <strong>Met</strong> <strong>Office</strong><br />
Overview of 2012 Upgrades<br />
• 4 cycles per day, 12 forecast members per cycle (control + 11<br />
perturbed members)<br />
• ETKF uses a full set of 22 perturbed members<br />
• 24-member products by lagging last 2 cycles<br />
• <strong>Global</strong> 3-day ensemble MOGREPS-G ~32 km (N400)<br />
• NB. MOGREPS-15 (used for TIGGE) stays at ~60km (N216)<br />
• UK convective-scale ensemble MOGREPS-UK 2.2 km<br />
• Direct nesting in <strong>Global</strong><br />
• Increase ETKF members from 22 <strong>to</strong> 44<br />
• Retire MOGREPS-R (early in 2013)
© Crown copyright <strong>Met</strong> <strong>Office</strong><br />
Increased<br />
resolution<br />
<strong>Global</strong>: 60km70L → 32km70L<br />
(planned for Aug 2012)<br />
Decreased RMSE in control<br />
member <strong>and</strong> ensemble mean<br />
Compensating change in<br />
spread:<br />
increase in 10m wind spread<br />
(more eddies resolved)<br />
Decrease in 500Z spread<br />
Improved RPS scores<br />
Increased cost <strong>and</strong> runtime
MOGREPS-UK<br />
Convective-scale ensemble
© Crown copyright <strong>Met</strong> <strong>Office</strong><br />
MOGREPS-UK<br />
2.2km UKV model<br />
•First operational UK<br />
ensemble 2012<br />
•12 members, 2.2km<br />
•Now running technical<br />
trial with 36h forecasts<br />
Products will use Neighbourhood<br />
Processing <strong>to</strong> account for spatial<br />
uncertainty not covered by ensemble<br />
spread<br />
Probability of more than 100 mm of<br />
rain in 18 hours within 10 miles of any<br />
location
Olympic Showcase<br />
• Forecasts from MOGREPS-UK will be displayed on <strong>the</strong><br />
<strong>Met</strong> <strong>Office</strong> website (under <strong>the</strong> “Invent” tab) for <strong>the</strong> London<br />
2012 Olympics
© Crown copyright <strong>Met</strong> <strong>Office</strong><br />
Scotl<strong>and</strong> 8 December 2011<br />
“The conditions are exactly as predicted when <strong>the</strong> <strong>Met</strong> <strong>Office</strong><br />
issued its red warning” Nicola Sturgeon, Deputy First Minister<br />
• Mon 5 Dec – Yellow warning issued –<br />
meetings with Scottish Govt, Transport<br />
Scotl<strong>and</strong> <strong>and</strong> Resilience community begin<br />
• Wed 7 Dec – Red warning issued<br />
• Thu 8 Dec – Gusts in excess of 100kts<br />
Valid from 10:00 am on Thu 08 Dec 2011<br />
until 21:00 on Thu 08 Dec 2011
© Crown copyright <strong>Met</strong> <strong>Office</strong><br />
MOGREPS-UK Case Study:<br />
Gales in Scotl<strong>and</strong> on 08 Dec 2011
Coupled ensemble forecasting,<br />
medium-range <strong>and</strong> beyond<br />
© Crown copyright <strong>Met</strong> <strong>Office</strong>
Medium-range <strong>and</strong><br />
Seamless Forecasting<br />
• Currently <strong>the</strong> medium-range ensemble, MOGREPS-15 is<br />
essentially <strong>the</strong> same as MOGREPS-G but run <strong>to</strong> 15 days using<br />
UK member-state computer time at ECMWF.<br />
• We are proposing <strong>to</strong> bring <strong>to</strong>ge<strong>the</strong>r ensemble prediction<br />
systems on <strong>the</strong> medium-range <strong>and</strong> monthly <strong>to</strong> seasonal<br />
timescales, including.<br />
• Initial condition perturbations from ETKF <strong>and</strong>, in longer term,<br />
Ensemble Data Assimilation System;<br />
• Coupled model <strong>to</strong> better represent ocean-atmosphere interactions.<br />
• 15-day forecast case studies show similar performance from<br />
coupled <strong>and</strong> uncoupled models for week 1, with some<br />
improvements at week 2.<br />
• A more comprehensive comparison of MOGREPS-15 <strong>and</strong><br />
GloSea5 (seasonal) ensembles planned, once GloSea is<br />
upgraded <strong>to</strong> N216 <strong>and</strong> MOGREPS-15 <strong>to</strong> L85.
Suite 1<br />
20 members<br />
15 days<br />
Suite 2<br />
2 members<br />
2 months<br />
Suite 3<br />
2 members<br />
7 months<br />
Suite 4<br />
X members<br />
Hindcast<br />
Medium-range<br />
products<br />
Schematic of possible<br />
coupled medium-range/<br />
monthly/ seasonal EPS<br />
Monthly<br />
products<br />
Seasonal<br />
products
S<strong>to</strong>chastic Physics <strong>and</strong><br />
SST <strong>and</strong> Soil Moisture Perturbations<br />
© Crown copyright <strong>Met</strong> <strong>Office</strong>
© Crown copyright <strong>Met</strong> <strong>Office</strong><br />
S<strong>to</strong>chastic Physics<br />
• SKEB:<br />
• Modulate a s<strong>to</strong>chastic pattern that is generated with<br />
specified spatial <strong>and</strong> temporal characteristics<br />
• Includes effects from numerical scheme smoothing, KE<br />
from convection, diffusion of large-scale flow<br />
• Produces rotational <strong>and</strong> divergent wind components at<br />
each time-step<br />
• R<strong>and</strong>om Parameters:<br />
• ~15 parameter values varied by AR1 process in set range<br />
• Each ensemble member equally likely<br />
• Perturbed Physics Tendencies:<br />
• Under test, but difficulty with model stability
• SST:<br />
© Crown copyright <strong>Met</strong> <strong>Office</strong><br />
SST & Soil Moisture<br />
Perturbations<br />
• Derive statistics of SST fields, e.g.<br />
day-<strong>to</strong>-day �t<br />
• Calculate a power-law for r<strong>and</strong>om<br />
forcing pattern<br />
• Perturbation added <strong>to</strong> SST field<br />
(which remains constant during <strong>the</strong><br />
forecast)<br />
• Soil Moisture:<br />
• Pass soil moisture of each<br />
member through forecast cycles<br />
• Re-centre perts <strong>and</strong> check values<br />
SST pert example<br />
Soil moisture pert example
© Crown copyright <strong>Met</strong> <strong>Office</strong><br />
SST Perturbations:<br />
Verification scores
© Crown copyright <strong>Met</strong> <strong>Office</strong><br />
Impact of Soil Moisture Perturbations<br />
T2m Tropics
© Crown copyright <strong>Met</strong> <strong>Office</strong><br />
Conclusions<br />
• Main MOGREPS changes in 2012 are:<br />
• 6-hr ETKF cycling with lagging<br />
• resolution upgrade of global system <strong>to</strong> 32km, <strong>and</strong><br />
• introduction of a convective-permitting suite<br />
• SST <strong>and</strong> SMC perturbations provide useful<br />
increase in spread of near-surface variables<br />
• Ensembles are no longer just a nice-<strong>to</strong>-have<br />
extra – <strong>the</strong>y are an integral part of <strong>the</strong><br />
operational NWP<br />
• Ensembles help us turn wea<strong>the</strong>r forecasts in<strong>to</strong><br />
risk management <strong>to</strong>ols
Any Questions?<br />
© Crown copyright <strong>Met</strong> <strong>Office</strong>