Hide

To give you the best possible experience, this site uses cookies. Continuing to use this site means you agree
to our use of cookies. If you'd like to learn more about the cookies we use please find out more

Session 3.2 abstracts

Symposium home | Session 2 | Session 3 | Session 4 | Abstracts by Author |

 

3.2 Evaluating the Impact of Observations on Ocean Forecasting Systems: Dividends from Investing in a Global Ocean Observing System

Session conveners: Gilles Larnicol and Peter Oke


The table below lists all abstracts for Session 3.2 by author. To read the full abstract click on the title-link.

The unique reference number (ref. no.) relates to the abstract submission process and must be used in any communications with the organisers.

All abstracts from session 3.2 are available for download - pdf

 Ref.NoPrimary AuthorAffiliationCountryAbstract titlePoster
S3.2-01Alonso Balmaseda, MagdalenaECMWFUnited KingdomQuantifying the impact of the Ocean Observing System in reanalysis of the global oceanPoster-pdf
S3.2-02Bayler, EricNOAA/NESDIS/STARUnited StatesPreliminary Results from Assimilating SMOS Satellite Sea-Surface Salinity Fields in an NCEP Operational Ocean Forecast SystemPoster-pdf
S3.2-03Benkiran, MounirCLSFranceImpact of the assimilation of high-frequency data in a regional model 
S3.2-04Benkiran, MounirCLSFranceImpact of the assimilation of the High-Resolution Sea surface Temperature in forecast model 
S3.2-05Carse, FionaMet OfficeUnited KingdomImpact of marine mammal temperature and salinity data on ocean model fieldsPoster-pdf
S3.2-06Fontana, ClémentUPMCFranceA data assimilation framework for the optimal deployment of the BioArgo network 
S3.2-07Fujii, YosukeMRI-JMAJapanActivities of Observation System Evaluation and Design in JMA/MRIPoster-pdf
S3.2-08Hackert, EricUNIV OF MD / ESSICUnited StatesImpact of Satellite Sea Surface Salinity Assimilation on ENSO Forecasts for the Tropical Indo-PacificPoster-pdf
S3.2-09Huang, XinmeiAustralian Bureau of MeteorologyAustraliaMonitoring real-time observational data used by operational OceanMAPSPoster-pdf
S3.2-10Krasnopolsky, VladimirEMC/NCEP/NWS/NOAAUnited StatesDownscaling Geophysical Fields Using Neural Networks 
S3.2-11Liu, LiyanNOAA/NCEP/EMC/United StatesSeasonal Cycle in SSH over an Atlantic Ocean Sector 
S3.2-12Mourre, BaptisteCentre for Maritime Research and ExperimentationItalyCooperation versus coordination of underwater glider networks: an assessment from OSSEs in the Ligurian SeaPoster-pdf
S3.2-13Nadiga, SudhirNOAA/NESDIS/STARUnited StatesExperiments with Satellite Ocean Color Fields in an NCEP Operational Ocean Forecast SystemPoster-pdf
S3.2-14Oke, PeterCSIROAustraliaHow many satellites are needed for operational oceanography?Poster-pdf
S3.2-15Remy, ElisabethMercator OceanFranceObservation impact studies at Mercator OceanOral
S3.2-16Ravichandran, MINCOISIndiaImpact of in-situ observations on global ocean analysis: Tropical Indian Ocean Cancelled
S3.2-17Wen, CaihongCPC/NCEP/NOAAUnited StatesAssessment of surface heat fluxes from the NCEP CFSR at the KEO buoy sitePoster-pdf
S3.2-18Xue, YanClimate Prediction Center/NCEPUnited StatesThe NCEP/GFDL Observing System Experiments for Tropical Pacific Observing System: Early ResultsPoster-pdf

 


ID 3.2-01

Quantifying the impact of the Ocean Observing System in reanalysis of the global ocean

M.A. Balmaseda1, K. Mogensen1

1 ECMWF, Reading, UK

2 CERFACS, Toulousse, France

Abstract

A set of observing system experiments (OSES) has been conducted using the latest ECMWF Ocean Reanalysis System 4 (ORAS4). The experiments consist on withdrawing, once at a time, observing systems, namely Argo, Altimeter and Moorings. In addition to ORAS4, other 2 reference experiments are used to evaluate the observing system: a CNTL, where only SST is used, and a ORAS4-NoBC, an equivalent to ORAS4 where no bias correction is used. Impact of the observations is quantified using assimilation statistics, impact on climate signals such as ocean heat absorption, and comparison with independent data.


ID 3.2-02

Preliminary Results from Assimilating SMOS Satellite Sea-Surface Salinity Fields in an NCEP Operational Ocean Forecast System

E. Bayler1, S. Nadiga2, A. Mehra3, and D. Behringer3

1 NOAA/NESDIS/STAR, NCWCP, College Park, MD

2 IMSG at NOAA/NWS/NCEP/EMC, College Park, MD, USA

3 NOAA/NWS/NCEP/EMC, College Park, MD.

Abstract

Although an ocean state variable, salinity has long been sparsely measured in time and space. Recently available satellite sea-surface salinity (SSS) observations provide important global data for assimilating into ocean forecast systems. The European Space Agency’s (ESA) Soil Moisture – Ocean Salinity (SMOS) mission, launched in 2009, began providing the first repeated global high spatial and temporal resolution coverage of SSS. Here, we present results from assimilating SMOS SSS data into National Oceanic and Atmospheric Administration’s (NOAA) operational global Modular Ocean Model version 4 (MOM4), the modeling platform of NOAA’s Global Ocean Data Assimilation System (GODAS) and the ocean component of NOAA’s seasonal-interannual Climate Forecast System (CFS). This variable resolution model provides an efficient and robust platform for analyzing ocean dynamics at greater than eddy-resolving spatial scales. Forcing this model with NOAA’s daily Climate Forecast System Reanalysis (CFSR) fluxes and assimilating SMOS Barcelona Expert Centre (SMOS-BEC) 0.25-degree gridded SSS fields (2010-2012, 3-day averages updated every 3 days) using a relaxation technique, sensitivity experiments are run with different relaxation time periods to evaluate the importance of high-frequency (daily to mesoscale) and low-frequency (seasonal) SSS variability on the ocean’s overall state. Comparison is also made with the results from a model run constrained by NOAA’s 2009 World Ocean Atlas monthly climatological SSS values.

To establish robustness, we examine the global signal-to-error ratio, unconstrained ocean model SSS values divided by the error estimates provided for the SMOS SSS data, to identify oceanic regions and frequency bands for which the SMOS SSS fields can reliably and significantly improve the oceanic state and initialization of coupled forecast models. We then apply the Empirical Orthogonal Function (EOF) technique to analyze the SSS field’s modes of variability in the tropical Pacific, a domain where oceanic dynamics are dominated by El Niño interannual variability in the cold tongue region and by high-frequency precipitation events in the western Pacific warm pool region. We expect that assimilating salinity fields at the sea surface will impact surface circulation and, through changes in baroclinic pressure gradients, the three-dimensional circulation patterns in the upper ocean. Thus, we examine changes in upper-ocean heat content, mixed-layer depths, and velocity in the top 300 m of the water column. It is shown here that assimilating surface salinity fields causes significant seasonal and interannual changes in the three-dimensional circulation patterns of mass, momentum, and heat in model results. Finally, preliminary verification studies are conducted using independent (i.e., non-assimilated) observations to show that satellite SSS data assimilation improves ocean state representation. The observations used for verification are global merged satellite SSH fields and daily vertical profiles of scalars (temperature, salinity and velocity) from multiple fixed-location buoys located in the equatorial Pacific ocean. Our results show that SMOS SSS fields help improve the simulated ocean state, thus providing better initialization of coupled seasonal and tropical cyclone forecast systems.


ID 3.2-03

Impact of the assimilation of high-frequency data in a regional model

Mounir Benkiran1, Claire Dufau1 and Yann Drillet 2

1 CLS DOS 8-10 Rue Hermes 31526 Ramonville St Agne Cedex, France

2 MERCATOR Ocean 8-10 Rue Hermes 31526 Ramonville St Agne Cedex, France

Abstract

Mercator-Ocean has developed a regional forecasting system at 1/12° resolution over the North East Atlantic (IBI: Iberia, Biscay and Irish), taking advantage of the recent developments in NEMO. The model was forced by ERA-interim products (every 3 hours) including the atmospheric pressure. In addition to atmospheric forcing, the model includes astronomical tidal forcing. This regional forecasting system uses boundary conditions from the Mercator-Ocean global reanalysis (GLORYS: GLobal Ocean ReanalYses and Simulations).The assimilation component of the Mercator Ocean system, is based on a reduced-order Kalman filter (the SEEK or Singular Extended Evolutive Kalman filter). An IAU method (Incremental Analysis Updates) is used to apply the increments in the system. The error statistics are represented in a sub-space spanned by a small number of dominant 3D error directions. A 3D-Var scheme corrects for the slowly evolving large-scale biases in temperature and salinity. The data assimilation system allows to constrain the model in a multivariate way with Sea Surface Temperature (AVHRR + Multi-satellite High resolution), together with all available satellite Sea Level Anomalies, and with in situ observations from the CORA-03 data base, including ARGO floats temperature and salinity measurements. The background SLA field accounts for the high frequency signal determined by the model and the forcing by atmospheric pressure.

In this study we show the impact of the assimilation of altimetry data when the atmospheric correction is not applied. The method to assimilate altimetry data containing atmospheric pressure and wind effect is described and the impacts on the analysis ocean fields are quantified


ID 3.2-04

Impact of the assimilation of the High-Resolution Sea surface Temperature in forecast model

Mounir Benkiran1 and Yann Drillet 2

1 CLS DOS 8-10 Rue Hermes 31526 Ramonville St Agne Cedex, France

2 MERCATOR Ocean 8-10 Rue Hermes 31526 Ramonville St Agne Cedex, France

Abstract

Mercator-Ocean has developed a regional forecasting system at 1/12° resolution over the North East Atlantic (IBI: Iberia, Biscay and Irish), taking advantage of the recent developments in NEMO. The model was forced by ERA-interim products (every 3 hours) including the atmospheric pressure. In addition to atmospheric forcing, the model includes astronomical tidal forcing. This regional system uses boundary conditions from the Mercator-Ocean global reanalysis (GLORYS: GLobal Ocean ReanalYses and Simulations). The assimilation component of the Mercator Ocean system, is based on a reduced-order Kalman filter (the SEEK or Singular Extended Evolutive Kalman filter). An IAU method (Incremental Analysis Updates) is used to apply the increments in the system. The error statistics are represented in a sub-space by a small number of dominant 3D error directions. A 3D-Var scheme corrects the temperature and salinity large-scale biases. The data assimilation system allows to constrain the model in a multivariate way with Sea Surface Temperature (AVHRR + Multi-satellite High resolution), together with all available satellite Sea Level Anomalies, and with in situ observations from the CORA-03 data base, including ARGO floats temperature and salinity measurements. The background SLA field accounts for the high frequency signal determined by the model and the forcing by atmospheric pressure.

In this study we show the contribution of the assimilation of surface temperature at high spatial resolution (2km). We use the daily L3S SST fields produced by Météo-France at 2km of horizontal resolution over 2009. We demonstrate the system's ability to assimilate these data and how they improve the analysis.


ID 3.2-05

Impact of marine mammal temperature and salinity data on ocean model fields

Fiona Carse1, Matt J Martin1, Alistair A Sellar1

1 Met Office, Exeter, UK

Abstract

Temperature and salinity profiles are being obtained from instrumented marine mammals in near real-time (figure 1). The mammals, mostly elephant seals, are sampling in high latitude regions where there are very few other in-situ observations. The study focuses on the Southern Ocean, with the possibility of extending to the Arctic.

Mammal observations are compared to model background values using a two year hindcast of the UK Met Office’s Global FOAM v12 system. This version of Global FOAM uses the NEMO ORCA025L75 ocean model coupled to the CICE sea-ice model. Data assimilation is performed using the NEMOVAR 3D-Var system with profile observations from EN3 as well as altimeter, SST and sea-ice data. A data-withholding run has been carried out to assess the impact of the mammal data, also using Global FOAM v12. Results from both runs will be presented, with examples drawn from interesting regions such as fronts of the Antarctic Circumpolar Current and near the sea-ice edge.

S3.2-05-Carse

Figure 1: Marine mammal profile observations in 2011 (left); instrumented southern elephant seal, Mirounga leonina (right).


ID 3.2-06

A data assimilation framework for the optimal deployment of the BioArgo network

C. Fontana1, F. D'Ortenzio1, H. Claustre1, P. Brasseur2, A. Mangin3

1 Laboratoire d'Océanographie de Villefranche, UMR 7093, Villefranche-sur-mer, France

2 Laboratoire de Glaciologie et Géophysique de l’Environnement, UMR 5183, Grenoble, France

3 ACRI, Sophia-Antipolis, France

Abstract

During the last decades, several thousand of autonomous profiling floats were deployed through the oceans, measuring every 10 days parameters such as sea salinity or temperature from 2000 meter depth up to sea surface. A new generation of biogeochemical sensors dedicated to oceanography appeared recently allowing to imagine a global network of drifting autonomous platforms that will measure routinely parameters such as chlorophyll, nitrate or oxygen concentration. This network is sometime referred as the BioArgo network.

The variability of oceanic biogeochemical properties can be highly seasonal and spatially localized. As an example, the North Atlantic phytoplankton spring bloom has a major impact on the earth carbon balance while it happens only on a few month-period every year at latitudes higher than 45°N. It is thus necessary to define an efficient deployment strategy for the BioArgo network allowing the characterization of a given process through optimal spatio-temporal sampling. The optimality of a network is a subjective notion that needs to be clearly defined through relevant metrics. Such metrics have to be accurately defined and adapted to the considered variables and system specificity.

The aim of this study is to design an Observing System Experiment (OSE) allowing to quantify the expected impact of a simulated BioArgo data set on the trajectory of a large scale coupled physical-biogeochemical model. We focus our work on the computation of a Degree of Freedom of the Signal (Rabier, 2006), inherited from the Bayesian data assimilation framework. The DFS is an evaluation of the « informative content » contained in a given data set and its estimation does not imply the effective correction of the model state through a data assimilation experiment. Nevertheless, its computation is tightly linked to the data assimilation method set up (e.g. error covariance matrix, observational error, localization procedure). We will thus discuss in this contribution how this OSE could be tuned to keep its relevant aspect with regards to oceanic biogeochemical specificities. In particular the combination between space-born (ocean color) and in situ lagrangian (BioArgo floats) observing systems into a common data assimilation system will be adressed.

Finally, an OSE does not imply feed-back from the modeling system, and solely quantify the potential impact of a given data set through data assimilation procedure. Also, this work will pave the way to the realization of more complex Observing System Simulation Experiments (OSSEs) testing previously defined optimality criteria in realistic conditions of modeling of the large scale ocean biogeochemistry.

Reference :

Rabier, F., Importance of data: A meteorological perspective. In :Ocean Weather Forecasting. Springer Netherlands, 2006. p. 343-360.


ID 3.2-07

Activities of Observation System Evaluation and Design in JMA/MRI

Y. Fujii, K. Ogawa, T. Toyoda, N. Usui, T. Kuragano, M. Kamachi

Japan Meteorological Agency/ Meteorological Research Institute, Tsukuba, Japan

Abstract

Several activities for evaluating/designing ocean observation systems have been continued in JMA/MRI. For example, we evaluated the impact of Argo floats data on the data assimilation results through an Observing System Experiment (OSE) using MOVE/MRI.COM, the ocean data assimilation system developed in JMA/MRI. In this OSE, we performed 5 assimilation runs using 80%, 60%, 40%, 20%, 0% of available Argo float profiles, and the accuracies of these assimilation runs are evaluated using the 20% of Argo float profiles which are not used in all assimilation runs, that is, we evaluate the accuracies using independent observation data. The accuracy is monotonically improved with increasing the number of assimilated Argo floats from 0% to 80%, which means that Argo float data effectively improve the data assimilation result.

We also performed an OSE for evaluating the impact of Argo floats and TAO/TRITON buoys on the ENSO forecast. First, we prepare three data assimilation runs of MOVE/MRI.COM. All available data is assimilated in one run, but Argo floats or TAO/TRITON buoys data are withheld in the other two runs. We, then, perform 13-month 11-member ensemble forecasts from each run using the JMA’s operational coupled model, JMA/MRI-CGCM. The forecast scores for 6-month lead time are improved for NINO3, NINO3.4 and NINO4 SST when Argo float data is not withheld in the assimilation run, and this impact is enhanced for 12-month lead time (Figure 1). This improvement also influences several atmospheric fields, including the sea level pressure, precipitation, and the divergence in the upper troposphere.

In the presentation, we also introduce the study for identifying the water-mass pathway using an adjoint technique which offers effective information for observation system design.

S3.2-07-Fujii

Figure 1: Improvements of the Anomaly Correlation Coefficient (ACC) scores by assimilating Argo Floats (Floats) and TAO/TRITON buoys for SST averaged in the box areas in 1-7 and 8-13 Lead Time (LT) forecasts.


ID 3.2-08

Impact of Satellite Sea Surface Salinity Assimilation on ENSO Forecasts for the Tropical Indo-Pacific

Eric C. Hackert1, Antonio J. Busalacchi1

1 ESSIC University of Maryland, College Park, MD, USA

Abstract

In this presentation, we assess the impact of satellite sea surface salinity (SSS) observations on seasonal variability of tropical Indo-Pacific Ocean dynamics as well as on dynamical ENSO forecasts using a Hybrid Coupled Model (HCM). The HCM is composed of a primitive equation ocean model coupled with a SVD-based statistical atmospheric model for the tropical Indo Pacific region. An Ensemble Reduced Order Kalman Filter (EROKF) is used to assimilate observations to constrain dynamics and thermodynamics for initialization of the HCM. Gridded products from AVISO multi-satellite sea level, AVHRR OI SST, and point-wise subsurface temperature (Tz) and salinity (Sz) are assimilated in the control experiment. In addition to the control assimilation variables, gridded observed SSS from SMOS, Aquarius, in situ (mostly comprised of Argo data within 10 m of the surface), and combined products are assimilated into the model to efficiently isolate the impact of satellite SSS.

In earlier work we have demonstrated that assimilation of gridded fields of SSS, derived from in situ SSS observations, has led to significantly improved coupled forecasts for lead times greater than 6 months ([Hackert et al., 2011]). We found that the positive impact of SSS assimilation is brought about by surface freshening in the western Pacific that led to increased barrier layer thickness (BLT) and shallower mixed layer depths. Thus, the net effect of assimilating SSS is to increase stability and reduce mixing, which concentrates the wind impact of ENSO coupling. Specifically, the main benefit of SSS assimilation comes from improvement to the Spring Predictability Barrier (SPB) period. SSS assimilation increases NINO3 SST anomaly correlation for 6-12 month forecasts by 0.2-0.5 and reduces RMS error by 0.3-0.6oC for forecasts initiated between December and March, a period key to long-lead ENSO forecasts. Satellite salinity assimilation in the Indo-Pacific region is therefore expected to lead to a better estimation of the dynamical features that are associated with El Niño.

In this study coupled experiments are initiated from EROKF initial conditions with and without SSS assimilation and run for 12 months for each month covering the SMOS and Aquarius periods to test the impact of satellite SSS assimilation. The results are validated and inter-compared. Although there is not enough satellite SSS to rigorously statistically validate the coupled experiments, we show preliminary results highlighting the positive impact of satellite SSS on ENSO forecasts.


ID 3.2-09

Monitoring real-time observational data used by operational OceanMAPS

X.Huang1, L. Majewski1, H. Beggs2, G. Brassington2, P. Sakov2, M. Entel1, C. Spillman 2

1 Australian Bureau of Meteorology, Melbourne, Australia

2 CAWCR/Bureau of Meteorology, Australia

Abstract

The Ocean Model Analysis and Prediction System (OceanMAPS) has been operational at the Australian Bureau of Meteorology since August 2007. OceanMAPS has resulted from the development of the BLUElink project, which is a collaboration between the Australian Bureau of Meteorology, the Commonwealth Scientific and Industrial Research Organisation (CSIRO), and the Royal Australian Navy. OceanMAPS is a global ocean forecasting system with high horizontal resolution of 0.1o in the Australasian region and coarse resolution elsewhere. There is high vertical resolution over the top 200m to resolve the mixed layer. OceanMAPS provides analyses and forecasts of ocean temperature, salinity and currents.

OceanMAPS consists of a sequential procedure:

  • Collection of observations from both in situ and remotely sensed sources.
  • Preparation of surface forcing fields from the Numerical Weather Prediction (NWP) output.
  • BLUElink Ocean Data Assimilation System (BODAS) which provides the best estimate of the initial ocean state.
  • Ocean Forecast Australia Model (OFAM) which provides a 7 day ocean forecast.

The observational data used in the OceanMAPS system come from remotely sensed altimetry and Sea Surface Temperature (SST), as well as in-situ profiles. Current OceanMAPS retrieves altimetry data from the Radar Altimeter Database System (RADS), which was developed by the Delft Institute of Earth-Oriented Space Research and the NOAA laboratory for Satellite Altimetry. It uses satellite altimeter data from Jason1-1, Jason-2 and CryoSat-2. These data are processed with quality control in the Australian Bureau of Meteorology before they are used for ocean data assimilation.

The U.S. Naval Oceanographic Office’s (NAVOCEANO) AVHRR MCSST (Advanced Very High Resolution Radiometer Multi-Channel Sea Surface Temperature) real-time global area coverage Sea Surface Temperature (NAVOCEANO SST) has been used in BODAS since 11 February 2011. NAVOCEANO SST daily retrievals are from NOAA18 and 19 and ESA Metop-A. WindSat SST has also been added to OceanMAPS from 14 November 2012.

The in-situ profile data is collected under the ship-of-opportunity program (SOOP), the Argo project and moored buoys. These messages from around the globe are decoded and stored in the local real-time data base (rtdb). More detailed Argo data is also obtained by ftp from two Global Data Assembly Centers (GDAC) located in the USA and in France.

Monitoring results of the real-time observational data used by OceanMAPS are presented and discussed.


ID 3.2-10

Downscaling Geophysical Fields Using Neural Networks

V. Krasnopolsky1,*, C. Lozano1, H. Kim2

1 Environmental Modeling Center, NCEP/NWS/NOAA, College Park, USA

2 IMSG at Environmental Modeling Center, NCEP/NWS/NOAA, College Park, USA

* vladimir.krasnopolsky@noaa.gov

Abstract

Traditional downscaling of gridded fields from a coarse to a fine resolution grids are usually based upon a linear interpolation algorithms intended, among other things, to preserved most of the scales contained in the coarse grid in the downscaled fields defined on the target grid with fine resolution. In this presentation we explore the use of non-linear mappings based upon Neural Networks (NN) for downscaling. Time collocated, but not necessarily spatially collocated sets of fields defined in the coarse grid and in the fine grid are used as a training set for NN based algorithms intended to reconstruct, partially, the intermediate scales contained in the fine resolution fields. NN algorithms with different degree of non-locality are investigated. Examples are drawn from data-derived analyses and model-derived fields.


ID 3.2-11

Seasonal Cycle in SSH over an Atlantic Ocean Sector

Liyan.Liu1, Carlos Lozano2,,Dan Iredell1

1 I.M. Systems Group, College Park, MD 20740, USA

2 Environmental Modeling Center NCEP/NWS/NOAA,

College Park, MD 20740, USA

Abstract

Daily high spatial resolution SSH analysis in an Atlantic Ocean Sector (24S-72N) for the year 2011 is employed to study the seasonal and short time scale variability. The analysis use satellite altimeter data including Jason-1, Jason-2 and Envisat observations. A two-dimensional variational analysis (2DVAR) is performed in order to obtain the daily analysis for SSH field. The associated SSH yearly time series are decomposed into a linear annual trend, seasonal cycle and high frequency residuals. Features of the seasonal cycle including annual, semiannual, triannual and even higher order components will be studied in magnitude and phase pattern with eastward and westward propagation in SSH time variability. There are also differences in the spatial patterns, in general associated to regions with different oceanic regimes. The correlation time scales of the high frequency SSH residual are regionally dependent and reveal the underlying mechanics of the Atlantic Ocean.


ID 3.2-12

Cooperation versus coordination of underwater glider networks: An assessment from OSSEs in the Ligurian Sea

B. Mourre and A. Alvarez

Centre for Maritime Research and Experimentation, NATO Science and Technology Organization, La Spezia, Italy

Abstract

Underwater gliders are robotic platforms able to fly in the ocean by using small fins to convert vertical forces induced by buoyancy changes into a horizontal motion. This propelling procedure implies low energy consumption and enables long periods of operation at sea without the need of human intervention. Moreover, the two-way satellite communication established with the glider while at the sea surface enables a regular control of the platform trajectory and opens the possibility for adaptive sampling procedures.

The deployment of glider swarms provides the opportunity to collect subsurface oceanic observations with a further increased spatio-temporal resolution. In that case, the cooperation (i.e. the fact that the gliders trajectories are driven by a common objective) between the individual platforms can be expected to improve the efficiency of oceanographic sampling missions compared to a naive collective behavior.

Based on Observing System Simulation Experiments (OSSEs), this study evaluates the performance of two levels of cooperation for a fleet of three gliders driven by the objective to reduce the temperature forecast errors of a regional ocean circulation model in the Ligurian Sea (Western Mediterranean). The performance of a coordinated network flying in a triangular formation is compared to that of a cooperative but unaware network (i.e. still sharing the prediction error minimization objective, the platforms do not follow any fleet formation).

A limited-area configuration of the Regional Ocean Modeling System (ROMS) is used to simulate the oceanic circulation in the Ligurian Sea. The nature run of the OSSEs is provided by a reference simulation of the model. An ensemble Kalman filter is used to assimilate the glider observations. Four two-day cycles of glider sampling and model predictions are generated. The results indicate a better performance of the coordinated network configuration due to an enhanced capacity to capture the eddy structure which is responsible for the largest model error in the experimental domain.


ID 3.2-13

Experiments with Satellite Ocean Color Fields in an NCEP Operational Ocean Forecast System

S. Nadiga1, E. Bayler2, A. Mehra3, and D. Behringer3

1 IMSG at NOAA/NCEP/EMC, College Park, MD, USA

2 NOAA/NESDIS/STAR, NCWCP, College Park, MD

3 NOAA/NCEP/EMC, College Park, MD.

Abstract

Solar shortwave heating of the upper layers of the ocean is dependent on the wavelength of the incoming radiation and the optical properties of the water column and correlates with chlorophyll concentration, which modulates the absorption of solar insolation in the upper ocean. NASA chlorophyll data, derived from satellite ocean color measurements, provide the basis for examining the impact of ocean color datasets on global ocean model simulations. The satellite ocean color data sets used in this study are chlorophyll fields of varying temporal resolution from the Sea-viewing Wide Field-of-view Sensor (SeaWiFS), and Visible/Infrared Imaging Radiometer Suite (VIIRS) missions. National Oceanic and Atmospheric Administration’s (NOAA) operational global variable-resolution Modular Ocean Model version 4 (MOM4) provides an efficient and robust platform for analyzing the impact of satellite chlorophyll data on ocean dynamics at greater than eddy-resolving spatial scales. The model is forced with NOAA’s daily Climate Forecast System Reanalysis (CFSR) fluxes, with relaxation techniques employed to assimilate daily satellite sea-surface temperature (SST) fields and climatological monthly sea-surface salinity (SSS) fields. Through changes in density profiles, differential heating patterns cause baroclinic pressure gradients, which, in turn, impact the three-dimensional circulation patterns of the upper ocean. Thus, we examine changes in upper-ocean heat content, mixed-layer depths, and velocity in the top 300 m of the water column. Anomalous build-up of Equatorial Pacific ocean heat content is an important variable for the recharge-discharge oscillator theory for the evolution of El Nino events. Here, we show that differences in the chlorophyll data inputs cause significant changes in the ocean heat content anomalies in the tropical Pacific. Thus, it is important for seasonal predictions that we study the impact that the ocean color data sets have on the quality of the ocean forecasts. Finally, in addition to inter-comparisons of the different simulations, we conduct preliminary verification studies between model simulations and independent (i.e, non-assimilated) observations, such as satellite sea-surface height (SSH) fields, vertical profiles of prognostic variables from in situ observations, and ocean heat content and mixed-layer depths from the NCEP Global Ocean Data Assimilation System (GODAS).


ID 3.2-14

How many satellites are needed for operational oceanography?

Peter. R. Oke1, Madeleine L. Cahill2

1 CSIRO Marine and Atmospheric Research, Hobart, Australia

2 CSIRO Marine and Atmospheric Research, Hobart, Australia

Abstract

Satellite altimetry is widely regarded as the most critical data type for motoring the mesoscale ocean circulation. To determine how many altimeters are needed to constrain an eddy-resolving ocean model, we perform a series of Observing System Experiments (OSEs), systematically adding data from one, two, and three altimeters from a data-assimilating eddy-resolving ocean model. We find that the addition of the first altimeter has the biggest impact - reducing the mismatch to observed sea-level anomaly (velocity) observations by up to 21% (13%) in high-variability regions. The addition of a second and third altimeter provides an additional, modest constraint on the model - modifying the details of the circulation, such as the size, position, and shape of eddies - all of which are important for operational applications. Noting that near-real-time data outages of altimeter data are common, we argue that at least three altimeters are needed for operational oceanography.


ID 3.2-15

Observation impact studies at Mercator Ocean

E. Rémy1, V. Turpin1, N. Ferry2, M. Benkiran3

1 Mercator Ocean, Toulouse, France

2 Météo France, Toulouse, France

3 CLS, Toulouse, France

Abstract

To estimate how much ocean analyses and forecasts are reliable and confident is an important issue for ocean forecasting centres. This is closely related to the observation network and to the different data types that are assimilated. Characterization of the sensitivity of ocean analyses to the different assimilated observations is a central activity of GODAE OceanView Observing System Evaluation Task Team and Mercator Ocean is contributing in several ways to this issue. Different approaches are developed to assess the impact of different types of observation on the analysis and will be illustrated.

Recently, GODAE OceanView OSE-TT has defined a framework to carry out in a coordinated way near real time observing system experiments (NRT OSEs) in different ocean forecasting centers. Different kind of data types (e.g. SLA, SST, Argo profiles, etc…) are with-held during a particular month in a forecasting system running in parallel to the operational one. Mercator Ocean began to set up such experiment over the first month of 2013 with the global ¼° ocean system. This should provide answers to the following questions. What is the sensitivity of forecasting systems to a particular observation type? How robust are these results with respect to the different forecasting systems?

Longer experiments, as OSEs, are necessary for a better understanding of the observation impact on constraining longer time scale ocean processes. Such experiments were carried out to assess the impact in our analysis of the actual altimeter constellation and ARGO floats. Regions of the ¼° global ocean are clearly differently responding to the changes in the observation network.

OSSE, where observations are simulated from a fully known ocean simulation, are also planned to simulate the impact of extension of future networks such as the deep ARGO floats and large swath altimeter (SWOT).

Other complementary approaches are explored. Forecast error sensitivities in the Mercator ¼° global ocean reanalysis were computed thanks to the adjoint of the ocean model. They highlight the regions or variables important to observed to reduce the forecast error. The observation influence is another tested approach, measuring how influent are the observations in the analysis.


ID 3.2-16

Impact of in-situ observations on global ocean analysis : Tropical Indian Ocean

M. Ravichandran and S. Siva reddy

Indian National Centre for Ocean Information Services, Hyderabad 500 090, India

Abstract

Impact of different in-situ ocean Observation system was evaluated using 3d var ocean assimilation system INCOIS-GODAS (Global Ocean Data Assimilation System at Indian National Centre for Ocean Information Services). We have carried out different experiments such as (i) assimilate all temperature and salinity profiles from in-situ observations (Argo, Moorings and XBT/CTD), (ii) without any assimilation, (iii) assimilation of only Argo profiles, (iv) Assimilation of only profiles from moorings, (v) assimilation of only XBT/CTD profiles, (vi) withheld only Argo profiles, (vii) withheld only profiles from moorings, (viii) withheld only XBT/CTD profiles. These experimental results are evaluated using independent moorings (temperature, salinity and currents), altimeter (sea level), TMI/AMSR-E (SST) for the Tropical Indian Ocean. The outcome of each experiment for different in-situ observations is presented.

The results indicate that avoiding the temperature and salinity profiles from profiling floats (Argo) for the assimilation causes significant degradation in the quality of ocean analyses in the TIO. Interestingly, removing moored buoy based measurements from assimilation does not degrade the quality of ocean analyses significantly in the tropical IO except over thermocline ridge region. Though not so significant, avoiding moored buoy observations from existing GOOS for assimilation improved the quality of ocean currents especially in the equatorial Indian Ocean. Results from the present study further indicates the importance of assimilating observations from profiling floats along with ship-based, for obtaining better ocean analyses in the tropical IO.


ID 3.2-17

Assessment of surface heat fluxes from the NCEP CFSR at the KEO buoy site

Caihong Wen1, Meghan F. Cronin2, Yan Xue1 and Arun Kumar1

1 Climate Prediction Center, NCEP, NOAA, College Park, Maryland, U.S.A

2 Pacific Marine Environmental Laboratory, NOAA, Seattle, Washington, USA

Abstract

Western boundary currents involve intense air-sea interactions, where the ocean loses heat and moisture to the atmosphere. In the North Pacific, air-sea interactions near Kuoshio Extension can affect the weather and climate locally and remotely and thus are critical to the global climate system. An important issue is whether these high air-sea fluxes are well represented in the model based Reanalysis data sets. The recently completed Climate Forecast System Reanalysis (CFSR) at NCEP, which uses a 6-hour coupled model forecast as the first guess, and high spatial and temporal resolution, has a potential to capture surface heat fluxes well. This study evaluates the extent to which surface heat fluxes at the CFSR agree with the Kuoshio Extension Observatory (KEO) buoy. Results show that the temporal variations of latent and sensible heat fluxes are highly consistent between the CFSR and the KEO time series. The bias and the root mean square (RMS) are substantially reduced in the CFSR compared with earlier NCEP reanalysis data (e.g., R1 & R2). The total surface heat flux from the ocean to the atmosphere in the CFSR is overestimated about 25 Wm-2, which was smaller than found in the previous reanalysis. Causes for biases in the surface heat fluxes are further examined by comparing the CFSR and KEO fluxes to fluxes computed by applying the Coupled Ocean-Atmosphere Response Experiment (COARE) version 3.0 bulk algorithm to CFSR state variables. Direct comparison between the CFSR state variables and observed variables at KEO will also be conducted.


ID 3.2-18

The NCEP/GFDL Observing System Experiments for Tropical Pacific Observing System:

Early Results

Yan Xue1, Arun Kumar1, David Behringer1, Gabriel Vecchi2 and Xiaosong Yang2

1 NCEP/NWS/NOAA

2 GFDL/OAR/NOAA

Abstract

The TAO/TRITON array is the cornerstone of the ENSO observing system because it systematically measures upper ocean temperature, salinity, velocity and air-sea fluxes that contribute to the dynamics of ENSO, and are essential for ENSO monitoring and prediction. One tool to assess the value of the TAO data in the presence of the Argo data is to conduct Observing System Experiments (OSEs). We conduct coordinated OSEs and hindcast experiments using the NCEP and GFDL ocean data assimilation systems and seasonal forecast models for the post-Argo period 2004-2011. The relative roles of the TAO and Argo data towards constraining the upper ocean thermal structure and ocean currents in ocean reanalysis are assessed. Hindcast experiments initialized from the OSEs are used to show if the seasonal forecast skill of the current generation seasonal forecast models are able to show the benefits of enhanced ocean observing systems.

Four OSE runs are made, in which no observations (CTL), all observations (XBT, moorings, Argo) (ALL), all except the moorings (noTAO), and all except the Argo (noArgo) data are assimilated. Hindcast experiments are initialized with oceanic conditions from the four OSEs around January 1, April 1, July 1 and October 1 during 2004-2011. For each start time, an ensemble of 6 (10) coupled forecasts with perturbed initial conditions is integrated up to 9 (12) months ahead using the seasonal forecast model at NCEP (GFDL). For the OSE runs, we examine the mean bias, standard deviation, root-mean-square error and anomaly correlation with observations. The error statistics contrasted between a pair of OSE runs are used to assess the impacts of different ocean observing systems on the ocean analyses. For the hindcast experiments, we examine the systematic bias, root-mean-square error and anomaly correlation of the tropical Pacific SST. The results from the two seasonal forecast systems are compared and the common characteristics on the impacts of different observing systems on ocean analysis and forecast skill will be summarized. This project is part of the efforts to build a multi-model capability in NOAA for assessing impacts of ocean observing systems on seasonal forecast.

rger impact on surface currents, leading to a stronger initial reaction of the currents, yet also a delay in subsequent direction changes, such as those followed by storm events. The surface currents react more quickly to the cyclic rotations of the sea breeze as these aren’t as strong and are of shorter duration, allowing the utmost layers of the surface currents to adjust to the wind, leaving the currents at larger depths unaffected.