© Crown copyright Met Office IGST XIII UK national report Matt Martin Silver Spring, USA, 2 - 4 June 2008

© Crown copyright Met Office

IGST XIIIUK national report

Matt Martin

Silver Spring, USA, 2 - 4 June 2008


Contents

• Update on NCOF

• Progress with NEMO ocean model framework

• Data assimilation progress and plans

• Ecosystems

• OSTIA

• Shelf-seas

• Data delivery


Update on NCOF

• NCOF was launched in March 2005

• Partners are the Met Office and the NERC marine research institutes

• POL, NOCS, ESSC, PML

• CEFAS are Observers, with a view to joining

• The Consortium Agreement is nearing completion

• The scope of NCOF is being broadened to include a wider range of timescales and to more explicitly include observational programmes.

• NCOF Workshops

• Annual gatherings

• NCOF website: http://www.ncof.co.uk


Progress with NEMOOverview

• NEMO is the Nucleus for European Modelling of the Ocean

• All Met Office ocean modelling activities will use NEMO, including shelf-seas, FOAM, seasonal forecasting and climate prediction.

• NEMO Consortium Agreement has been signed

• NEMO Systems Team

• Core of team at CNRS (Paris), but consortium agreement commits resources from NOCS (Southampton) and Met Office (Exeter)

• NEMO Users

• Includes the open ocean, shelf seas, seasonal forecasting and climate prediction communities

• Intention to use NEMO as the primary open ocean modelling tool across all communities


• FOAM model component is being transitioned to use NEMO.

• For the My Ocean project, responsibility for global and North Atlantic coverage is led by Mercator with FOAM providing backup.

• Model configurations are shared with different choices for surface forcing, parameter settings and assimilation.

• Currently using LIM2 ice model. Plan to switch to CICE.

• The initial system has been run in hindcast mode and is being run in the trial operational suite.

• First ORCA025 hindcast run from 1st April 2005 – April 2007:

• Initial conditions produced by running model only for 3 months (starting from operational FOAM 1˚ restart file on 1st Oct 2004) then with assimilation for 3 months.

• Forced by 6-hourly Met Office NWP surface fluxes.

• With LIM2 ice model.

Progress with NEMO:FOAM


Global ¼˚ (ORCA025)

grid, bathymetry and river outflow climatology provided by Mercator

N. Atlantic 1/12˚

Indian Ocean 1/12˚

Med. 1/12˚

All configurations have 50 levels with 1m resolution near surface.

Progress with NEMO:FOAM configurations


• Initial results of verification of hindcast of ¼˚ NEMO.

• Comparison of model background with observations (before they’ve been assimilated).

• Averaged errors for Jan 2006 (after 10 months).

Temperature Salinity

Temperature overall RMS error 0.77K


Progress with NEMO:Comparison of trial operational NEMO with OSTIA SST

FOAM-NEMO SST on 14th May 2008 OSTIA SST analysis on 14th May 2008


• The existing operational FOAM system and the new global FOAM-NEMO system will be used as part of the GODAE intercomparison project.

• The new FOAM-NEMO system will be used to run data withholding experiments for the GODAE OSE project.

• Future plans are to fully implement the global and regional FOAM-NEMO configurations operationally in 2008. These will take over from the existing FOAM system at that stage.

• A number of hindcasts with the models will be run and assessed.

• Future model developments will aim to improve the results of the hindcasts.

• Transition the system to the new supercomputer in 2009.

Progress with NEMO:FOAM model plans


1960 1970 1980 1990 2000

year

Dra

ke p

assa

ge tr

ansp

ort (

Sv)

Drake Passage transport• High resolution hindcasts (with ecosytem models included) eg global 1/4 degree

• NEMO model has now completed 1958-2001, then started again from the 2001 state but with fluxes from 1958 to allow the ocean to properly spin-up

• Proposal to compare NOCS hindcast runs (without data assimilation) with similar runs undertaken by ESSC which do include assimilation - to assess the effect of the assimilation.

Progress with NEMO:Recent work at NOCS


Global Ocean Modelling with Adaptive Unstructured Grid Methods

Aim: produce a global circulation model which fully utilises three-dimensional adaptive mesh technology – based on ICOM

(FInite-eLement Adaptive grid Modelling of Ecosystems and Nutrient Transport)

Aim: Embed an ecosystem model into a newly developed, non-hydrostatic, finite element, adaptive grid, physical ocean model

Progress with NEMO: Long-term modelling

plans at NOCS


ESSC: Model/Synthesis details

• Completed Synthesis experiments (Hydrography only assimilation)

• Global 1° resolution (1958-2004) 46 years

• Global 1/4° resolution (1987-2004) 18 years

• NEMO modelling framework

• Collaboration with DRAKKAR consortium who have run the global 1/4° resolution model only for 46 years.

• Bulk forcing (DFS3) from ERA40/ECMWF operational analyses & CORE (ISCCP)

• Data assimilation scheme is based on an isothermal analysis (changes depth of isotherms and salinity on isotherms).

• Now extending runs

• 2005-2007 using ECMWF Operational forcing

• 1989-1998-present will be done using ERA-Interim (4DVar atmosphere with better fluxes)


ESSC: Biases in the 2004 mean state

ORCA025-G70Simulation only

ORCA025-R07Assimilation

ClimatologyWOA05

Average 300-1000m


Transports (Black Simul., Green Synth.)


• All FOAM configurations assimilate a range of data using an Optimal Interpolation type method. Data types include:

• Temperature and salinity profiles including Argo floats (also XBTs, CDTs, buoys,…)

• Satellite altimeter Sea Surface Height

• In situ and satellite Sea Surface Temperature

• Sea-ice concentration

Argo float distribution

Satellite SST data: 50km product (from NESDIS)

Argo Network as of April 2008

Altimeter SSH data for 10 days

Processed for 1/9º North

Atlantic model

Data assimilation: Overview of FOAM scheme


• Overall data assimilation methodology the same as current operational FOAM. Includes assimilation of T and S profile data, satellite altimeter SLA data, in situ and satellite SST data, and sea-ice concentration data.

• Improvements include:

• First-guess-at-appropriate-time (FGAT) which provides more accurate calculation of the model counterpart of the observation.

• Improved sea ice assimilation uses OSI-SAF sea ice concentration data and uses the OI scheme as do all the other data types.

Data assimilation in FOAM:Implementation in NEMO

Observed sea-ice concentration 1st Feb 2006

Background sea-ice concentration 1st Feb 2006


To be included in next operational change and next hindcast run:

• New error covariances being generated for ORCA025 using two techniques – using (o-b) and NMC method - using data from the 2 year hindcast.

• Altimeter bias correction in order to account for inconsistencies between the mean dynamic topography (MDT) and the model.

• Pressure correction, to account for model biases in the tropics.

• Scheme to include correlation scales which vary depending on the horizontal gradients of potential vorticity in the background field.

• Assimilation of GHRSST data (high-resolution SST), including satellite SST bias correction, as is used in OSTIA.

Data assimilation in FOAM:Recent developments in FOAM-NEMO


STDBIAS

• No bias correction (black) compared to bias correction (red)• After only 1 month SSH RMS error 10 cm rather than 12 cm• Minor improvements in T and S

RMS

mean

Data assimilation in FOAM: Altimeter bias correction scheme


• Assimilate various high resolution SST data sets from GHRSST (L2p data)

• Include bias correction scheme used by OSTIA

• AATSR and in situ data treated as reference data set in bias correction

1 day of satellite data – 8th April 2008

Data assimilation in FOAM: Assimilation of GHRSST data


• A review of the various assimilation schemes available for use in operational ocean forecasting has been carried out.

• It was decided to begin implementing a variational scheme, NEMOVAR. Developed at CERFACS and ECMWF.

• Initial implementation in FOAM will be 3DVar with the capability to move to 4DVar in the future.

• Developments will be made to implement anisotropic error covariance representation within NEMOVAR.

• Investigate impact of a different change of variable to, for instance, density and spiciness.

Data assimilation in FOAM: Plans


Ecosystem modellingFOAM-HadOCC overview

• FOAM-HadOCC running at 1º, 1/3º and 1/9º resolution

• Assessment of initial year-long integrations underway

• Ocean colour data assimilation scheme (developed by NOCS) implemented and testing started

Schematic of HadOCC

• Development of open ocean water clarity capability

• Hadley Centre Ocean Carbon Cycle model (HadOCC) has been coupled with the FOAM system

• NPZD ecosystem model


Daily mean RMS errors in the North Atlantic Air-sea exchange of CO2

significantly improved after assimilating ocean colour data

Phytoplankton (mmol N/m3) Zooplankton (mmol N/m3)

Control - truth Assimilation - truthAssimilation - truth

Total DIC (mmol C/m3)

Free run

BDA run

Detritus (mmol N/m3) Nutrients (mmol N/m3)

Ecosystem modelling:Identical twin experiments

Control - truth

Assimilation - truth


Global average RMS (solid lines) and mean (dashed lines) errors compared to satellite chlorophyll data.

Green: no DA Black: only physical DA Red: physical and biological DA

The scheme appears to be effective at correcting chlorophyll Validation of other biological variables against independent data (CAVASOO)

has begun

Ecosystem modelling:Real world experiments


Ecosystem modelling:Plans

• The key next steps are:

• further quantitative validation of initial FOAM-HadOCC integrations.

• further refinement of biological assimilation scheme.

• parameter tuning (required to improve performance).

• 10-year re-analysis of FOAM-HadOCC with ocean colour and physical assimilation (1º global).

• on-line coupling to NEMO.


NanoPhyto

Si

Diatoms

Fe N

MesoZoo

MicroZoo

Slowdetritus

Fast particle flux

Model of Ecosystem Dynamics, nutrientUtilisation and SequestrAtion

Chlorophyll (mg m-3), 15 May

MEDUSA in ¼º NEMO


Shelf seas forecasting at the Met Office

6km MRCS Model

1nm Irish Sea

Model

12km Atlantic

Margin Model

• NEMO will form basis for our North-west Shelf system. Boundary conditions will come from the 1/12 degree N Atlantic FOAM-NEMO model.

• Various new schemes implemented in NEMO by different European groups to enable running in the shelf-seas

• POLCOMS and NEMO run on AMM domain, same horizontal + vertical resolution.

• 1-year tides-only experiments forcing with 15 tidal components at lateral boundaries.

• Validated against tide gauge and buoy data, plus Topex data.

• Each model run with native TKE scheme and also with k-ε model (Canuto et al 2001) within GOTM framework.


Validation against satellite data (TPXO): amplitude errors (m)

POLCOMS NEMO


OSTIA: Operational SST & Sea Ice Analysis

• Daily 1/20° (~5.6km) global SST analysis.

• Analysis of the ‘foundation’ SST [pre-dawn or below the diurnal warm layer].

• Blend of data sources, using satellite (microwave & IR) and in situ data.

• Using many GHRSST data products.

• Almost all Medspiration products.

• Now running daily, operationally.

• Using a scheme based on the FOAM assimilation scheme, with persistence based background.

• Uses sea ice analysis performed by EUMETSAT OSI-SAF (met.no / DMI).

Sample analysis for 19 Apr 2007


• OSTIA improved the RMS and bias in the NWP forecasts during the trial period (August 2007).

RMS Error

Mean Error Reduced Bias at low levels

Pre

ssur

e Le

vel


• Direct links to UK Royal Navy forecasters.

• For commercial use, data is available from the Met Office’s Data and Products Distribution System (DPDS) at http://www.metoffice.gov.uk/research/ncof/foam/dpds.html

• Data available for research use from Live Access Server at http://www.nerc-essc.ac.uk/godiva/

• A new prototype data server has been developed and is undergoing testing. One new advance with this server is the ability to visualise NCOF data using Google Earth.(http://behemoth.nerc-essc.ac.uk/ncWMS/godiva2.html)

Products distribution:Overview



• A service desk will be operated at the Met Office for all European Marine Core Services:

• Data policy still to be defined but is likely to be freely available to all users and free of charge

• Users will be able to access data via the service desk using service level agreements (SLAs)

• Data will be supplied from production centres in Europe

• Present plan is for data to be available via sftp

• OpenDAP technologies will be introduced at a later date

• On-line catalogues will be available to search for products

Products distribution:Future system in My Ocean


Phase errors for M2 component (degrees)

POLCOMS-GOTM NEMO-GOTM


• Observation Operator (also in NEMOVar) allows

• Isotherm / Isopycnal depths

• Salinity / Spiceness on Isotherm / Isopycnal surfaces

• For isotherms:

• Given T, S -> Calculate Z(T), S(T) for model and observations

• Za(T) = Zb(T) + ΔZ(T) or Ta(z) = Tb(z) + ΔT(z)

• Sbal(z), such that Sa(Ta)=Sb(Ta)

• Sa(T) = Sb(T) + ΔS(T)

• For isopycnals:

• Given T, S -> Calculate Z(ρ), π(ρ) for model and observations

• Z’(ρ) = Zb(ρ) + ΔZ(ρ)

• π’(ρ) = π b(ρ) + Δπ(ρ)

• Current Syntheses all use Isotherm analyses

ESSC: Isothermal/Isopycnal Assimilation


RMS Salinity Misfits : 1988-2004

ORCA1-R70ORCA1-R07-RA_t1ORCA025-R07-RA_t1

FULL – Global

STNA – Sub-trop. N. Atl.20N-45N, 80W-8W

SPNA – Subpolar N. Atl.45N-65N, 65W-5E

MDNA – Mid N. Atl.30N-45N, 35W-20W

Documents

© Crown copyright Met Office IGST XIII UK national report Matt Martin Silver Spring, USA, 2 - 4 June 2008