Page 1
Developments in the UK Met Office
Shelf Seas Forecasting System
Robert King
Enda O’Dea, Jennifer Graham,
Marina Tonani, Niall McConnell, Peter Sykes, Christine Pequignet
James While, Matt Martin, David Ford
Huw Lewis, JuanMa Castillo, Alex Arnold.
COSS-TT, Madrid, Sept. 2018
www.metoffice.gov.uk © Crown Copyright 2017, Met Office
Page 2
1. Recent history – developments to 7km resolution system (AMM7).
2. Upcoming changes – new 1.5km resolution system (AMM15).
3. Future developments – plans for coupling, DA, ensembles.
Contents
Overview - assimilation - resolution – coupling – future - summary
Page 3
North West Shelf System - AMM7v9
UK Met Office is the CMEMS MFC responsible for the NWS
providing
• Daily analyses and 7-day forecasts of physical variables
• BGC forecasts (coupled with ERSEM)
• 30-year regional reanalysis
Ocean model uses NEMOv3.6 in the AMM7 configuration
• 7km resolution, 51 vertical sigma-levels, non-linear free surface
• Tidal, wind- and pressure-forced model
• Forced by Met Office NWP fluxes (hourly winds, 3hrly pressure
• LBCs from Met Office 1/12⁰ North Atlantic Model and Baltic MFC
• Climatological rivers
Overview - assimilation - resolution – coupling – future - summary
Page 4
North West Shelf System - AMM7v9
Data assimilation scheme uses NEMOVAR
• 3DVar FGAT with 24-hour window
• Multi-scale background errors, multivariate balance
• Assimilate SST, T/S profiles, SLA
• L2/L3 SST - METOP, AVHRR, SEVIRI, VIIRS
• In situ SST - ships, drifters, mooring
• T/S profiles - XBTs, CTDs, Argo, marine mammals, gliders
• SLA from Jason-3, AltiKa, Cryosat, Sentinal 3A
• Bias correction scheme for SST and SLA
• Increments are applied using IAU method over 24hr window
• BGC – unbalanced Chl assimilation in reanalysis, will be
assimilated operationally soon.
Overview - assimilation - resolution – coupling – future - summary
Page 5
Recent Updates: AMM7v9 (oper since Apr 2017)
Example coverage of satellite altimeter observations
for 1-month from Sentinal-3, Jason-3, Cryosat and
AltiKa (top) and those assimilated (bottom). Example monthly coverage of T/S
Temperature
profiles
Salinity
profiles
Example daily coverage of SST obs
Overview - assimilation - resolution – coupling – future - summary
Page 6
Left: Mean surface currents for
January 2015 from AMM7v8 and
v9 and from observation-derived
current products.
Right: Depth profile of
temperature (left) and salinity
(right) innovation statistics from
AMM7 v8 (grey) and v9 (black)
averaged over full domain and 2-
year period.
King et al., 2018, Ocean Modelling
New system with assimilation of SST, SLA, and T/S profiles
Large reduction (>25%) in T/S RMS errors compared to v8 system.
Assimilation of SLA also introduces eddy-scale features to our analyses
which compares well with altimeter-derived surface current estimates.
Temperature Salinity
OLD
NEW
OLD
NEW
--- mean error
— RMSE
Recent History: AMM7v9 (oper since Apr 2017)
Overview - assimilation - resolution – coupling – future - summary
Page 7
Upcoming changes: AMM15
AMM15
AMM7
Overview - assimilation - resolution – coupling – future - summary
Recently upgrading the ocean model component to the 1.5km
resolution AMM15.
Higher resolution, slightly smaller domain.
New, higher resolution EMODnet bathymetry and coastline.
Surface forcing now used ECMWF bulk fluxes.
No change in vertical resolution.
Allows for fine detail of flow speed and variability, and can now begin
to resolve mesoscale eddies and internal tides.
Off-shelf biases are reduced, but on-shelf biases remain relatively
unchanged.
Graham et al., 2018, GMD
Page 8
Upcoming changes: AMM15
Recently upgrading the ocean model component to the 1.5km
resolution AMM15.
Higher resolution, slightly smaller domain.
New, higher resolution EMODnet bathymetry and coastline.
Surface forcing now used ECMWF bulk fluxes.
No change in vertical resolution.
Allows for fine detail of flow speed and variability, and can now begin
to resolve mesoscale eddies and internal tides.
Off-shelf biases are reduced, but on-shelf biases remain relatively
unchanged.
Graham et al., 2018, GMD
Overview - assimilation - resolution – coupling – future - summary
Page 9
Upcoming changes: AMM15
Recently upgrading the ocean model component to the 1.5km resolution AMM15.
Reduced biases in deep water regions going from AMM7 to AMM15, but on-shelf biases are mostly
unchanged as vertical processes dominate in shallow water.
AMM7
AMM15
Overview - assimilation - resolution – coupling – future - summary
Page 10
AMM7
AMM7
AMM15
AMM15
SST (°C)
1st Feb 2016
Upcoming changes: AMM15 with assimilation
AMM15 assimilation
follows same approach
as for AMM7
Assimilation of satellite
and in situ SST data,
satellite altimeter SLA
data (in the deep part of
the domain), and in situ
temperature/salinity
profiles has been
implemented in AMM15
Retains high resolution
model information.
Overview - assimilation - resolution – coupling – future - summary
Page 11
AMM15+DA vs AMM7+DA
SST
AMM15+DA
AMM7+DAAMM15+DA
AMM7+DA
S profilesT profiles
SLA mean bias
AMM7
AMM15
Overview - assimilation - resolution – coupling – future - summary
Page 12
AMM7 AMM15
AMM15+DA vs AMM7+DA
Overview - assimilation - resolution – coupling – future - summary
Comparison to independent glider observations (Orkney-Faroes).
Fresh salinity bias and subsurface cool bias in model reduced (AMM7 -> AMM15)
Likely to be due to a combination of assimilation and model improvements.
Obs-model
T profiles
Obs-model
S profiles
Page 13
OCEAN
ATMOSPHERE WAVES
Bottom stress
SEDIMENTS/BIOGEOCHEM
Surface stress
Wind
LAND SURFACE
Freshwater, Nutrients, Temperature
Surfacefluxes
Radiation, Temp, Precip, Evap
Towards coupled prediction?
Page 14
Monthly-mean impact of wave forcing on SST – difference between ocean-wave and ocean-only
April 2014 July 2014 Oct 2014 Feb 2015
Sensitivity of SST to wave coupling (without assimilation)
Inclusion of wave processes tend to cool simulated sea surface temperature
by order 0.5 K relative to ocean-only configurations during spring, summer
and autumn periods.
Overview - assimilation - resolution – coupling – future - summary
Page 15
Impact of coupling ocean to atmosphere and/or wave model
(without assimilation)
Improved flux forcing provided through a fully coupled ocean-wave-
atmosphere system can lead to systematic improvement of ocean sea
surface temperature simulations for the North West Shelf region.
ocean-only ocean-wave atm-ocean atm-ocean-wave
Overview - assimilation - resolution – coupling – future - summary
Monthly-mean impact on SST – difference to OSTIA SST analysis
Page 16
2018/19:
• Test changes to vertical mixing and light attenuation schemes.
• Test new high resolution altimeter observations.
• Estimate new error co-variances for AMM15.
• Implement coupled ocean-wave system operationally.
• Continue testing Chl PFT assimilation.
• Assimilate optical properties and test multivariate BGC
assimilation
2019-2021:
• Upgrading DA to work with NEMO4.0 and wetting-and-drying.
• Prepare for SWOT – swath altimeter.
• Investigate assimilation of altimeter SLA data in shallower waters.
• Continue testing fully coupled UKEP system
• Investigate impact of physical DA on BGC
• Ensemble forecasts and ensemble DA.
Future Developments
Overview - assimilation - resolution – coupling – future - summary
Summertime surface and bottom temperature
sensitivity to vertical mixing and light
attenuation scheme.
Page 17
Summary
Significant impact of assimilation of SLA and profiles into operational AMM7
- limited by spatial and temporal sampling of sub-surface obs
Upgrading to a higher resolution model (AMM15)
- lack of observations hinders validation
Existing assimilation capability has been implemented in AMM15
- operational later this year
From next year will couple AMM15 to wave model.
- again hindered by lack of (ocean and surface flux) observations
Lots of developments planned over coming few years.
Overview - assimilation - resolution – coupling – future - summary