deepwater current profile data sources for riser ... · riser engineering offshore brazil gus jeans...

DEEPWATER CURRENT PROFILE DATA SOURCES FOR RISER ENGINEERING OFFSHORE BRAZIL

Gus Jeans Fugro GEOS, Wallingford, UK

Marc Prevosto Ifremer, Brest, France

Liam Harrington-Missin

Fugro GEOS, Wallingford, UK Christophe Maisondieu

Ifremer, Brest, France

Christelle Herry Actimar, Brest, France

José Antonio M. Lima Petrobras, Rio de Janeiro, Brazil

ABSTRACT A variety of current profile data sources are compared for a

deepwater site offshore Brazil. These data were gathered for consideration as part of the Worldwide Approximations of Current Profiles (WACUP) Joint Industry Project, described separately in OMAE2012-83348. The primary source of data for current profile characterisation is site specific full water column measurement. Sufficiently high vertical and temporal resolutions are required to capture the dominant oceanographic processes. Such in-situ data are generally expensive and time consuming to collect, so there is an increasing tendency for numerical model current data to be considered for engineering applications. In addition to being relatively inexpensive and quick to obtain, model data are also typically of much longer duration. This potentially allows inter-annual variability and rare extreme events to be captured. However, the accuracy and reliability of numerical model data remains questionable, or unproven, in many deepwater development regions. This paper explores the suitability of such models to represent a deepwater site offshore Brazil, in relation to the key oceanographic processes revealed within the in-situ data.

1 INTRODUCTION The objectives and methodology of the Worldwide

Approximations of Current Profiles (WACUP) Joint Industry Project are described in a separate OMAE 2012 paper [1]. The key objective is quantitative assessment of different current profile reduction techniques for efficient and reliable riser design.

The project considers current profile characterisation at

four different offshore development sites, one of which is the Jubarte deepwater development offshore Brazil. The present paper explores the various data sources considered for representing the current profile at the Jubarte location.

It has become established best practice to derive current

profile data for engineering applications from site specific full water column measurement. This provides the most reliable quantification of the dominant oceanographic processes that impact the site in question. Sufficiently high vertical and temporal resolutions are required to capture the dominant oceanographic processes.

However, in-situ measured current profile data have some

disadvantages compared to the various numerical model sources that have become available in recent years. It is

Proceedings of the ASME 2012 31st International Conference on Ocean, Offshore and Arctic Engineering OMAE2012

July 1-6, 2012, Rio de Janeiro, Brazil

OMAE2012-83400

1 Copyright © 2012 by ASME

relatively expensive and time consuming to collect in-situ data, compared to model data which are generally inexpensive and quick to obtain. In-situ data sets are typically of much shorter duration, with measurement periods greater than one year being rare. Model data therefore have the potential to provide superior coverage of inter-annual variability and rare extreme events.

Despite the apparently clear potential advantages of model

data, it is very important to recognise the limitations of these data sources. The underlying physics and model resolution will usually allow a subset of all potential oceanographic processes to the represented. Even when a model can be expected to represent a certain process, the accuracy and reliability need to be validated using representative in-situ data before the model output can be safely used for a given engineering application.

Currents are the result of interactions between many

different processes such as general circulation, tides, inertial currents, and turbulence. The identification and quantification of the corresponding signals, occurring at different time scales, can make it possible to effectively complement in-situ data with results from numerical models.

2 IN-SITU CURRENT MEASUREMENT

The Jubarte Field is located in 1400m, Block BC-60 of the Campos Basin, 70km offshore the state of Espírito Santo, Brazil (Figure 1). The Jubarte “gold standard” dataset was created from data collected during a single measurement campaign carried out between October 2004 and September 2005.

The measurement campaign consisted of a single upward

looking Acoustic Doppler Profiler (ADP) and four single point Vector Averaging Current Meters (VACM) throughout the water column. In total, 12 measurement bins provide near-complete water column coverage. However, vertical resolution below 100 m (below the ADP) was coarse with measurement levels only at 350 m, 800 m, 1250 m and 1400 m. Vertical coverage is illustrated in Figure 2.

This dataset contained relatively few spurious records,

making the creation of the gold standard relatively simple. The only notable issue was the infilling of data at the 350 m level, at which there was a long duration gap during which a strong event was recorded. The interpolation methodology was carefully considered and agreed among the project team and participants.

Figure 1. Map of the Approximate Location of the Jubarte Field (Water Depths are in meters)


Figure 2. Depth Coverage of the Jubarte Measurement Campaign


The flow diagram in Figure 3 summarises the work involved in the creation of the Jubarte gold standard dataset. Further details were provided in a report provided to project participants [2].

The Jubarte gold standard dataset was provided in

MATLAB® Binary and Microsoft Excel format to the WACUP partners. These data, along with corresponding data from the other three locations, were subjected to numerous further analyses. The identification and quantification of different components of the flow is considered next in this paper.

Figure 3. Jubarte Data Preparation Flow Diagram

3 IN-SITU CURRENT ANALYSIS The Jubarte Field is situated in the path of the southerly

and south-westerly meandering Brazil Current (BC), which was seen to dominate the upper 100 m in this dataset. The lower water column (350 m and below) shows evidence of the weaker northerly to north-westerly flowing Intermediate Western Boundary Current (IWBC) which peaks in the 800 m layer. This baroclinic phenomenon is well documented [3, 4, 5]

Due to the coarse vertical sampling of the Jubarte dataset

below 93.5 m, the actual depth level of the BC IWBC interface can only be approximated. Additionally, the linear interpolation adopted by riser engineers to create a full water column profile will create a gradual transition from BC to IWBC. This is unlikely to be representative of the more pronounced interface expected with a feature of this type.

Without additional data (e.g. current models, conductivity,

temperature, and depth (CTD) dips etc.) there is no easy way to determine whether these interpolated profiles are representative of true conditions. If a non-linear interpolation method were to be used instead to capture the potentially more pronounced water mass interface, the resultant total load on the riser would likely increase.

Spectral analysis of the Jubarte dataset revealed no

unexpected results with a clear peak in energy at approximately 12 hrs relating to the semi-diurnal tidal period. The diurnal period (~24hrs) is less pronounced and merges into the inertial period (~33 hrs at this latitude) with no clear distinction between the two (Figures 4 and 5).

Fourier filtering was used to decompose the total observed

current into three period bands; less than 4 hrs (short period), between 4 hrs and 48 hrs (Tidal and Inertial period) and greater than 48 hrs (long period). Figures 6 to 8 present vertical profiles of statistics calculated at each measurement depth for each flow component. In all cases the long period flow is shown to dominate.

Jubarte QC’d Dataset

Remove all Bins after 19 September 2005

Infill temporal gaps of 28 hours or less using linear interpolation

on the velocity components

Infill missing velocity at 350 m (v350m) using the relationship between velocity at 93.5 m

(v93.5m) and velocity at 800 m (v800m):

mmm vvv 8005.93350 *64.0*36.0 +=

Gold Standard Jubarte Dataset


Figure 4. Spectral Analysis of the Jubarte Gold Standard Dataset – Velocity North (Vn)


Figure 5. Spectral Analysis of the Jubarte Gold Standard Dataset – Velocity East (Ve)


Figure 6. Jubarte Fourier Decomposed Statistical Profile Plot - Mean


Figure 7. Jubarte Fourier Decomposed Statistical Profile Plot – 95th Percentile


Figure 8. Jubarte Fourier Decomposed Statistical Profile Plot – Maximum


4 NUMERICAL MODEL SELECTION A comprehensive review of available numerical model

current data was undertaken to support data selection and gathering as part of the WACUP project [6]. The primary model selected for comparison with in-situ data in all four regions was the widely used dataset freely available from the HYCOM Consortium (http://www.hycom.org/dataserver/).

A number of selection factors were considered, including

global coverage, spatial and temporal resolution, model physics, data assimilation and data acquisition costs. The selection of this model for the purposes of the project does not imply that this is the best available in any given region. However this data source has become very widely used, so a detailed evaluation at the Jubarte site is worthwhile.

The HYbrid Coordinate Ocean Model or HYCOM is a 3D

global model with 1/12° spatial resolution and 32 levels on the vertical. The model bathymetry is derived from a digital terrain dataset (AVO-NRL DBDB2).

The model takes into account: • Forcing from Navy Operational Global Atmospheric

Prediction System (NOGAPS) • Rivers runoff from climatology values

It assimilates: • Satellite altimeter observations (Jason GFO Envisat) • Satellite and in-situ Sea Surface Temperature • In-situ temperature and salinity profiles (XBTs,

ARGO floats, moored buoys) Daily values of current are available from 2003 to 2010.

They have been extracted at the closest grid point to the measurement location, less than 2km away. HYCOM currents are available at 21 levels from the surface to 1200m depth

5 NUMERICAL MODEL VALIDATION The freely available global HYCOM data source provides

only daily snapshots of model current velocities. To facilitate meaningful comparison, daily vector mean values were derived from the measured currents at each depth.

Following the analysis described in Section 3, in which we

found that the long period component of the flow dominates at this location, it is reasonable to expect that the key processes might be represented by the daily values available from the model.

Model current speeds are quantitatively compared with measurements in Figures 9 to 13. The closest level in the HYCOM model was selected for each measurement depth considered, with no vertical interpolation.

These results demonstrate that currents are underestimated

by the model throughout the water column. At each depth, the maximum observed current is nearly twice as high as the maximum HYCOM model results.

The scatter plots show a poor correlation between

simultaneous measured and model currents. However Q-Q plots suggest a near linear underestimation in the statistical distributions. This could be corrected via model calibration to meet some engineering applications.

Further details of the model validation were provided in a

report provided to project participants [7]. This included an examination of current direction, which was much better represented in the model than current speed. This supports for the hypothesis that the key large scale oceanographic processes are captured by the model, but require calibration. Validation also considered term seasonal trends, which were represented to some extent by HYCOM. However, real seasonal trends cannot be reliably inferred from a single year of measured data [8].

So far we have considered how well the model represents

the temporal variability and statistics of currents at each separate measurement level. However the vertical coherence and structure of individual current profiles is critical for riser engineering applications. The ability of the model to produce profiles with realistic profile shapes was also examined in the WACUP project [9].

Altogether, the validation studies show that model profiles

require careful validation and calibration before being used to derive criteria for riser design applications.

6 SPATIAL VARIABILITY OF CURRENTS Despite the above limitations of model data, one clear

advantage over site specific measurement was highlighted [10]. The large scale synoptic view provided by the model provides a very valuable tool to assess and quantify spatial variability across a wide region. The cost of collecting in-situ data at so many locations would be considerable.

The model is believed to capture key spatial trends in the

large scale circulation in the region. The Brazil Current (BC) is clearly seen as a band of relatively strong speed, concentrated over regions of steeper seabed bathymetry (Figure 14). The intensity of the surface current appears stronger in regions to the south and west of the Jubarte location.


Figure 9. Measured and Model Daily Current Speed Time Series

Figure 10. Scatter and Q-Q plots of Model (20m depth) and Measured (23.5m depth) Daily Current Speed


Figure 11. Scatter and Q-Q plots of Model (100m depth) and Measured (93.5m depth) Daily Current Speed

Figure 12. Scatter and Q-Q plots of Model (400m depth) and Measured (350m depth) Daily Current Speed

Figure 13. Scatter and Q-Q plots of Model (800m depth) and Measured 800m depth) Daily Current Speed


An effective method of investigating spatial variability was developed, by computing the ratio between model current speeds at the site in question and speeds at all other grid points. These were derived using linear regression, with results presented in Figure 15. Values over unity indicate higher speeds and vice-versa. Spatial variations in the absolute difference in current direction are also shown.

It is important to note that the linear regression ratio is most significant when the correlation coefficient between the grid points was high. Regions where the correlation was higher than 0.8 and 0.5 are therefore encircled by dashed lines on the figures. Further investigation into spatial regression was undertaken, but details are beyond the scope of this brief paper.

Figure 14. Model Surface Current Speed Maximum (Left) and 95% Percentile (Right)

Figure 15. Model Surface Current Speed Ratio (Left) and Direction Difference (Right)


7 CONCLUSIONS Analysis of in-situ data reveals that the long period

component of the flow dominates at the Jubarte location. This supports the use of freely available model data to assess key characteristics of the current regime.

The selected model significantly underestimates the

magnitude of currents at the location in question, typically by a factor of about two. There was a poor correlation between simultaneous measured and model currents, but the statistical distributions show a clear trend, allowing calibration for some engineering applications. Current directions from the model proved much more reliable than current speeds.

The primary source of data for characterisation of current

profiles in riser design remains full water column, site specific, in-situ measurement. Models do play an increasingly useful supplementary role. However they must by probably validated and calibrated before use in any engineering application.

Models are superior to measurements for providing a large

scale regional interpretation of key oceanographic processes. They can also be used to cost effectively quantify spatial variability in the current regime.

ACKNOWLEDGMENTS WACUP participants are acknowledged for their

contributions towards the project and permission to publish this paper. Particular thanks are given to Petrobras for approval to release details of their measured data.

Key participant contributions are also acknowledged from

Valerie Quiniou-Ramus and Claire Channelliere of Total, Einar Nygaard and Børge Kvingedal of Statoil, Colin Grant and Richard Gibson of BP.

The remainder of the WACUP project team are also

acknowledged, including George Forristall, Raymond Nerzic, Cyril Frelin, Gavin Harte, Kavanagh Kieran, Mark Calverley and Andrew Watson.

REFERENCES [1] Prevosto, M., Forristall, G.Z., Jeans, G., Herry, C.,

Harte, G., Harrington-Missin, L. and Dooley, P. (2012). Worldwide Approximations of Current Profiles - the WACUP project. Proceedings of the 31st International Conference on Ocean, Offshore and Arctic Engineering, Rio de Janeiro, Brazil, 10-15 June 2012. OMAE2012-83348

[2] Harrington-Missin, L., Watson, A. and Jeans, G.,

(2011). Worldwide Approximations of Current Profiles. Fugro GEOS Report C50605/5881/R0. June 2011. Confidential to WACUP Participants.

[3] Evans, D.L, Signorini, S.S. and Miranda, L.B. (1983).

A note on the transport of the Brazil Current. Journal of Physical Oceanography, 13, 1732-1738.

[4] Silveira, I. C. A, Calado, L., Castro, B. M., Cirano, M.,

Lima, J. A. M., and Mascarenhas, A. d. S. (2004). On the baroclinic structure of the Brazil Current-Intermediate Western Boundary Current System at 22°-23°S. Geophysical Research Letters, 31, 2004.

[5] Silveira, I. C. A., Lima, J. A. M., Schmidt, A. C. K.,

Ceccopieri, W., Sartori, A., Franscisco, C. P. F., and Fontes, R. F. C. (2008). Is the meander growth in the Brazil Current system off Southeast Brazil due to baroclinic instability? Dynamics of Atmospheres and Oceans, 45, 187-207

[6] Muller, H., and Herry, C., (2010). State of the Art of

the Available 3D Ocean Model Outputs. Actimar Report 09.52 V1.2. September 2010. Confidential to WACUP Participants.

[7] Casitas, S., Herry, C., (2010). Comparisons Between

HYCOM Model & Jubarte Measurements. Actimar Report 10.27 V1.1. September 2010. Confidential to WACUP Participants.

[8] Harrington-Missin, L., Jeans, G. and Calverley, M.,

(2009). Synergistic Use of Satellite and In-Situ Current Data to Improve the Characterisation of Seasonal Trends. Proceedings of the 28th International Conference on Offshore Mechanics and Arctic Engineering, Honolulu, Hawaii, 31 May - 5 June 2009. OMAE2009-79172.

[9] Maisondieu, C., (2011). Are the Model Current Profiles

Representative of the Actual In-Situ Local Profiles. Ifremer Report R11HO008. June 2011. Confidential to WACUP Participants.

[10] Herry, C., (2011). Spatial Variability in HYCOM

Model at Jubarte & Girassol. Actimar Report 10.52 V2.0. March 2011. Confidential to WACUP Participants.


deepwater current profile data sources for riser ... · riser engineering offshore brazil gus jeans...

Documents