Status and future of the operational oceanography in Denmark

Vasily Korabel, Jens Murawski, Kristine Skovgaad Madsen, Jun

She, Vibeke Huess, Jacob Woge Nielsen, Till Rassmusen, Jacob Hoyer,

and the rest of the R&D team at DMI

OceanPredict19 , Halifax, 6-th May 2019

Denmark’s place in the oceanographic world

Denmark is a leading oceanographic center providing operational products for BS/North Sea• Storm surges• Wave forecasts• Ice forecasts for Greenland• Oil spill

DMI (Denmark) 2015-2016:WAM: local implementation to handle unresolved islandsonline-coupled bio model ERGOM HBM: code optimization for performance improvements (DMI)HBM+ERGOM: implemented improved river run-off and nutrients data (DMI)HBM: simple ice dynamic model + fast ice module implemented (DMI) SMHI (Sweden) 2017:Reanalysis system: NEMO + SCOBI + Ensemble 3dVarMSI (Estonia) 2015-2017CalVal system and dissemination unitBSH (Germany) 2017:ERGOM: improved chl calculations + light routine (Kd) FMI (Finland) 2017:WAM: local implementation to handle partly covered ice grids HBM: tuning ice-thermodynamic & heat exchange routines for improved sea ice product (DMI)

Copernicus Marine Environmental Monitoring Service (CMEMS): Consortium Service developments with DMI contribution

Existing ocean forecasting systems in Baltic Sea

ProductionUnit

Basin system Local system

Models Resolution Model Resolution Area

BAL MFC HBM 1.85km

BSH (GE) HBM 1.85km HBM 0.93km Transition waters

DMI (DK) HBM 5.55km HBM 0.40km Limfjoden

HBM 0.93km Transition waters

FMI (FI) OAAS (2D) 3.7km HELMI(sea ice)

1.85km Northern Baltic Sea

HBM 5.55km

MSI (EE) HBM 0.93km Gulf of Finland, Gulf of Riga

SMHI (SE) NEMO-Nordic

1.85km NEMO 0.36km Lake Vänern

NEMO 0.06km Brofjorden

UL (LV) HBM 102m -1.85km

Eastern Baltic Sea, lakes

Forecasting: Baltic-North Sea cooperation based on: Multiple forecastsReal-time data exchange (obs. & forecast) (Golbeck etal

2015)

≈1nm

≈0.5nm (1’x1.66’) ≈185m (6’’x10’’)

≈0.5nm (30’’x50’’)

2-way nesting:• to model Danish

territorial watersin high resolution

• to resolve narrowstraits

Dynamical nesting:Mass and momentum are exchanged on time step level during the update of the hydrodynamic equations.

Current operational CMEMS Model Setup

Fig. 1. Impact of model resolution on the Baltic inflow - bottom salinity difference (a) before and after the Major Inflow Event 2014/15 HBM0.9; (b) same as in (a) but for HBM1.8 (c) between HBM0.9 and HBM1.8 after the major inflow event (31-th May 2015).

HBM0.9 HBM0.9 –HBM1.8

May 2015minus Nov. 2014 May 2015

Bottom Salinity HBM1.8

HBM1.8

May 2015minus Nov. 2014

Bottom Salinity HBM0.9 Bottom Salinity HBM0.9-1.8

Inter-basin water exchange

1. Waves generate additional

drift currents of locally up

to 1 cm/s in the Gulf of

Finland (2 cm/s during

storms).

2. Wave induced oil drift

increases onshore oil

advection:

J. Murawski, J. Woge Nielsen, (2013) “Applications of an Oil Drift

and Fate Model for Fairway Design”, Chapter 11 of the book

“Preventive Methods for Coastal Protection – Towards the

Use of Ocean Dynamics for Pollution Control” (Editors

Tarmo Soomere, Ewald Quak), Springer 2013

T. Soomere, K. Döös, A. Lehmann, H.E. Markus Meier, J.

Murawski, K. Myrberg, E. Stanev (2014), “The Potential of

Current-and Wind-Driven Transport for Environmental

Management of the Baltic Sea”, 2014, AMBIO

Xi Lu, T. Soomere, E. Stanev, J. Murawski (2011), “Identification

of the environmentally safe fairway in the south-western

Baltic Sea and Kattegat using the Lagrangian trajectories”,

2012, Ocean Dynamics

landing probability20%

10%

-10%

-20%

0%

Coupled ocean-wave modelling

DMI, Research Evaluation , Copenhagen, 28-th Aug 2018

Shear Stress at

the sea bed

Fine Sediment

concentration in

the sea bed

SPM concentration

at the surface

Re-Suspension

Erosion

Sedimentation

Transport, sinking and upwards mixing of SPM

Processes:

•Sinking versus mixing

•Advection

•Sedimentation

•Re-suspension

•Erosion

•Consumption

•Bioturbation

3 SPM fractions:

wsink(f1)=0.0001 m/s

wsink(f2)=0.00002 m/s

wsink(f3)=0.001 m/s

EOOS, YEOS,

BALTADAPT:→ CLAIM

Improved shear stress

formulation, dependent

on the wave spectra.

Phenomenological model that describes theformation, consolidation and disintegration of fast ice, and the advance and retreat of its front.

Coastal Fastice Model

Ice growth

Damaging,Deformation

Sea ice thickness

Hirlam SKA

Requirement: high quality weather forcing for the entire domain

Storm surge prediction servicedevelopment, examples

Geographical representation: bathymetry, model tuning: bed friction, CD

100 years event inthe southern BS

Silent Storm SurgeJanuary 2017

November 2011

Multi-model ensemble and satellite observations. The Bodil storm surge example

DMI operational storm surge model overlaid with Cryosat satellite observations

BOOS ensemble of Baltic Sea models + observations

Future of the operational oceanagraphy in Denmark

Future activities focus on 3 issues: > filling observation gaps> improve forecast > modelling and provide better data and product service.

Ongoing : Development of a new operational NEMO-ERGOM-WAM-PDAF coupled system

i) being able to fully exploit benefits from future observationsii) generate meaningful products in finer scales e.g., sub-mesoscale and in estuary-coast-sea

continuum iii) efficient parallel computing and model grid structureiv) provide high quality forecasts as forcing to NWP and coastal climate modelsv) resolving correctly inter-basin and inter-sub-basin water exchange,vi) resolving synoptic variability and predictability in marine ecosystems, e.g., for algae bloom, vii) being able to address critical and relevant issues in coastal applications, e.g., marine spatial

planning, maritime safety, marine pollution protection, disaster prevention, offshore wind energy, climate change adaptation and mitigation

HBM v4

New pre-operational Model Setup

Nemo-Nordic

● Horizontal resolution of 1 nm

● 56 z* vertical layers

● GLS turbulence scheme

● OBCs from FOAM system

(T,S,U,V,SSH including tides)

● ATM forcing: mix of ECMWF 9km

and Arome 2.5km

● EHYPE hydrology

● 4 two-way nested subdomains with

horizontal resolution of 1-0.5 nm

● 21-122 z vertical layers

● k-e turbulence scheme

● Tides forced at open boundaries

● Climatological T/S OBCs

● ATM forcing: HIRLAM/HARMONY

2.3 km

● EHYPE hydrology

Water level validation

Intercomparison of HBM-NEMO water level

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

0.2

Aar

hu

s

Bal

len

Deg

erb

y

Ecke

rnfo

erd

e

Fors

mar

k

Tors

ham

nen

Gre

na

Hel

sin

ki

Hes

nae

s

Ho

rnb

aek

Kalix

Kem

i

Kie

lLT

Kla

iped

a

Kolk

a

Kose

row

Kuiv

astu

Lan

dso

rtN

orr

a

Mar

vike

n

Pald

iski

Rat

an

Ro

dvi

g

Ro

nn

e

Sch

leim

uen

de

Sim

rish

amn

Skan

or

Spik

arn

a

Talli

nn

Trav

emu

end

e

Vik

en

War

nem

uen

de

RM

SD (

m)

RMSD of water level: 2011-2015

HBM-NoDA

NEMO-NoDA

NEMO-DA

Summary on water level comparison

• Validation shows lower quality of NEMO-Nordic water levelproducts at all 62 Baltic Sea stations in comparing with CMEMS HBM V4

• MME results show that NEMO’s water level forecast at SE and FI coast are in reasonable quality but has significantly lowquality at Jutland E. coast, Belt Sea and German coast.

• For a 50yr event at Slipshavn in 20171029, NEMO underpredicted sea level peak by ~50cm.

Objective evaluation scheme

• Estimate error statisticsin 4 zones for SST/SSH/ ice, T/S and currents

• Bias, RMSD and correlation

• Product type: • 2016/05/03-11/02

– NEMO: without DA– HBM: without DA

• Best estimate: 2011 to 2015

Zone 1

Zone 4

Zone 3Zone 2

Quantitative evaluationperformance scores

Definition of

category &

scores

Significantly

worse (SW)

Worse

(W)

Slightly

worse (SLW)

Similar

(S)

Slightly

better (SLB)

Better

(B)

Significantly

better (SB)

Category order 1 2 3 4 5 6 7

|Bias(HBM) |-

|Bias(NEMO) |

<-0.5 -0.5 – -0.2 -0.2- -0.1 ±0.1 0.1-0.2 0.2-0.5 >0.5

RMSD(HBM) –

RMSD(NEMO)

<-0.5 -0.5 – -0.2 -0.2- -0.1 ±0.1 0.1-0.2 0.2-0.5 >0.5

Score of HBM 0.1 0.25 0.4 0.5 0.6 0.75 0.9

Score of NEMO 0.9 0.75 0.6 0.5 0.4 0.25 0.1

T/S evaluation by zone

Water temperature: results show that HBM performs better in all 4 zones for bias and

Zones 1 and 4 for RMSD. NEMO is better in Zones 2 and 3 for RMSD.

Water salinity: results show that NEMO performs better in all 4 zones for bias while HBM

performs better for RMSD.

Summary on T/S validation• HBM has significant bottom salinity bias in open Baltic Sea,

NEMO has not. (better representation of slop flow)• In transition waters, HBM has significantly smaller T bias and

salinity RSMD than NEMO• There are no big differences in RMSD in T and S in open Baltic

Sea, except for – In Zone 3 deep layer, NEMO has significantly smaller T

RMSD than HBM– HBM has smaller RMSD in 15-50m depth

• Inflow: – HBM does not give inflow signal in the central Baltic

Proper; – Inflow into Bothnian Bay are not simulated in both HBM

and NEMO.

Inflow assessment

• For inflow in Danish straits and W. Baltic, HBM gives much more accurate signals in terms of salinity peak with -3% bias while NEMO-DA has a clearly overestimation by 18% for depth-mean max salinity at Arkona.

• For inflow in Baltic Proper, HBM and NEMO-DA both show inflow signals in H2 and K2 in 2011/12. HBM does not show salinity increase in J1 (C. Baltic) bottom layer in the 2014/15 event, as existed both in observations and NEMO-DA

• For inflow in GoF, both HBM and NEMO-DA are able to model a salinity increaseevent in early 2013

• For inflow in Bothnian Bay, both HBM and NEMO-DA are not able to catch a salinityevent in early 2013.

Development of assimilation scheme

• Observations: DMI SST,

L3S multisensor merged

0.02 deg daily

Method: PDAF ( LETKF)

Ensemble-based inhomogeneous /

anistropic background error covariances

based on 10-year hindcast run

+ FMI daily ice

charts

Improvements in SST: v3 REF, V4 REF and DA

Improvements in ice cover representation

Prototype of a next generation reanalysis system Testing Agrif capabilities with tides

At DMI Nemo-Nordic code was upgraded to Nemo v4.0

● Horizontal resolution of 1 nm with added Agrif 0,5 nm nest

in the transition zone

● 56 z* vertical layers

● GLS turbulence scheme

● OBCs from the global reanalysis system + tides TPXO 7.2

● ATM forcing: 11km UERRA

● EHYPE hydrology

Summary and future directions

THANK YOU !