Back to Gary Brassington’s HOMEPAGE

 


 


 

 

GODAE Inter-comparisons in the south eastern Indian Ocean

Gary B. Brassington, Peter R. Oke, James Cummings, Fabrice Hernandez and Matthew Martin

 

Introduction ● Quantitative Assessment ●  Qualitative Assessment

 

Introduction

The goals of this study are to:

-         to understand when, where and why different systems perform well, or poorly, so that all systems can be improved in terms of their modelling and data assimilation techniques.

-         determine whether some ocean processes are consistently well/poorly reproduced by all systems and to understand why.

 

Data

Comparisons are made between model products provided under the GODAE Inter-comparison activity. The important elements of the four operational ocean forecast systems considered here are summarised in Table 1. Of these systems, Bluelink and HYCOM are eddy-resolving, and FOAM and Mercator are eddy permitting. Bluelink, FOAM and Mercator all use z-level models, while HYCOM is characterised by a hybrid, adaptive vertical grid. Both FOAM and Mercator use the same model code and grid. All systems use different NWP flux products. The NWP fluxes for Bluelink, FOAM and HYCOM represent diurnal variability, while Mercator uses daily averaged fluxes.

 

Table 1: Model characteristics.

 

Bluelink

UKMet / FOAM

NRL / HYCOM

Mercator Ocean

Model code

MOM4

NEMO

HYCOM

NEMO

Horizontal grid

1/10o

1/4o

1/12o

1/4o

Vertical grid

47 levels

50 levels

32 hybrid

50 levels

NWP fluxes

ECMWF or GASP 3-hr

UKMO 6-hr

NOGAPS 3-hr

ECMWF 1-d

Forecast range

7-d

5-d

7-d

7-d (14-d)

Frequency

twice weekly

daily

daily

daily (weekly)

Hindcast

11-d

1-d

5-d

14-d

 

All systems, except Bluelink, produce daily forecasts of at least 5-days. Each system uses a hindcast period to spin the model up prior to a forecast. FOAM uses a short hindcast period of 1-day, while all other systems use a hindcast period of at least 5-days.

 

Comparisons presented here are based on:

Bluelink ReANalysis – delayed mode quality controlled along-track altimeter data; NRT AMSR-E data, NRT Argo data (no XBT data), NRT ECMWF surface fluxes, 7-d update cycle.

OceanMAPS – 3-4-day forecasts using operational system (fields will soon be added).

UKMetNowcasts from the operational system … correct?

HYCOM – 5-day behind real-time analyses … correct?

Mercator – 14-day behind real-time analyses … correct?

 

Quantitative Assessment

Comparisons are presented for four regions in the south eastern Indian Ocean:

-         Tropical east Indian Ocean (90-100E, 22S-20N)

-         Indonesian Throughflow (100-120E, 8S-2N)

-         Timor Sea (100-120E, 22S-8S)

-         South east Indian Ocean (90-120E, 40S-22S).

-         South China Sea (100-120E, 2N-20N).

These regions are chosen as a subset of the Indian Ocean GODAE Inter-comparison region to isolate the different dynamical regimes of the South-eastern Indian Ocean circulation.

Click here for image (~128K)

 

Surface Drifting Buoys

Comparison of drifter-derived and modelled near-surface zonal (left) and meridional (right) velocities.

Model velocities are near the surface (Bluelink = 15 m; HYCOM = 10 m; Mercator = 10 m; UKMet = 1 m).

The statistics in parentheses in each panel show the RMS difference, correlation and linear regression coefficient with observations.

Results are based on comparisons with 58 drifters, including 1816 individual observations for drifters in the Indian Ocean. Data points where the observed speed is less than 10 cm/s are excluded from these calculations to eliminate errors in data processing (all comparisons degrade when these data points are included).

Click here for image (~180K)

 

Taylor diagram (see Taylor 2001; JGR-Atmos) showing comparisons between drifter-derived and modelled near-surface zonal (top) and meridional (bottom) velocities.

Click here for image (~500K)

Click here for image (~500K)

 

SST

North-east Indian Ocean: Time series of (a) anomaly correlation and (b) RMS difference between modelled SST and the Regional Australian Muliti-Sensor SST Analysis (RAMSSA) fields.

The statistics in the legend show the time-mean RMS difference and anomaly correlation.

We use the RAMSSA 1/12 degree resolution SST fields produced operationally at the BoM as described by Beggs (2007; BMRC Research Report; ~3.7M). RAMSSA is a GHRSST L4 product that produces a foundation SST estimate by combining AVHRR, AMSR-E, AATSR, and in situ buoy data using OI.

Taylor diagram (see Taylor 2001; JGR-Atmos) showing comparisons between AMSRE and modelled SST.

Click here for image (~280K)

Indonesian Throughflow: as above

Taylor Diagram for comparison with AMSRE SST

Click here for image (~280K)

Timor Sea: as above

Taylor Diagram for comparison with AMSRE SST

Click here for image (~280K)

South-east Indian Ocean: as above

Taylor Diagram for comparison with AMSRE SST

Click here for image (~280K)

South China Sea: as above

Taylor Diagram for comparison with AMSRE SST

Click here for image (~280K)

 

SLA

North-east Indian Ocean: Time series of (a) anomaly correlation and (b) RMS difference between modelled SLA and along-track SLA (from delayed-mode Jason, Envisat and GFO).

The statistics in the legend show the time-mean RMS difference and anomaly correlation.

Model sea-surface height is converted to SLA by removing the mean sea-level (mean dynamic topography) fields provided by each GODAE partner.

I spatial average for the region of interest has also been removed from each model field to eliminate any mean bias error.

 

Taylor diagram (see Taylor 2001; JGR-Atmos) showing comparisons between along-track and modelled SLA.

Click here for image (~270K)

Indonesian Throughflow: as above

Taylor Diagram

Click here for image (~270K)

Timor Sea: as above

Taylor Diagram

Click here for image (~270K)

South-east Indian Ocean: as above

Taylor Diagram

Click here for image (~270K)

South China Sea: as above

Taylor Diagram

Click here for image (~270K)

 

 

Mean and variance intercomparison

SST

A comparison of monthly mean and standard deviations for Feb/Mar/Apr for the east Indian Ocean. The BLUElink panel represents the OMAPS-an.

The values obtained from GODAE systems are compared with available GHRSST systems including RAMSSA (Beggs et al., 2007), NAVO and ODYSSEA..

Click here for image (~2.7M)

Click here for image (~3.2M)

SSS

A comparison of monthly mean and standard deviations for Feb/Mar/Apr for the east Indian Ocean. The BLUElink panel represents the OMAPS-an.

Click here for image (~1.7M)

Click here for image (~2.0M)

 

 

 



  Bureau Home   ||   BMRC Home  ||  Search  ||  Contact BMRC Webmaster


© Copyright Commonwealth of Australia 2009, Bureau of Meteorology (ABN 92 637 533 532)
Please note the Copyright Notice and Disclaimer statements relating to the use of the information on this site and our site Privacy and Accessibility statements. Users of these web pages are deemed to have read and accepted the conditions described in the Copyright, Disclaimer, and Privacy statements. Please also note the Acknowledgement notice relating to the use of information on this site. No unsolicited commercial email.