233

1 INTRODUCTION

The recent decade has seen a increasing development

in ship routeing services whereby more reliable

weather conditions, sea states and surface currents are

taken into account (e.g. [3, 13, 11]). Optimum ship

routing means the “best route” for a ship based on the

marine weather forecasts including wave and surface

current conditions, ship characteristics and cargo

requirements. For most voyages this will mean the

minimum transit time that avoids significant risk to

the vessel, crew and cargo. The goal is not to avoid all

adverse weather, waves and current conditions but to

find the best balance to minimize time of transit and

fuel consumption without placing the vessel at risk to

damage or crew injury. In recent years concern

regarding CO2 emission from ships has also emerged

and should therefore preferably be taken into account

regarding ship routing and voyage optimization (e.g.

[7, 8]).

In this paper we will present preliminary results of

the investigation and development of tools for

optimizing performance of voyages at sea (TOPVOYS

project) funded by the MarTERA EraNet Co-fund

program for a 3-year period from 2018-2021. The data

is outlined in section 2, followed by the approach for

the development of the routing optimization tool in

Section 3. Section 4 describes the validation method

and the summary is provided in Section 5.

2 DATA AND APPROACH

The types of data and information products

considered necessary for provision of reliable and

optimized ship routing can be grouped into marine

weather data, model forecast fields, near real time

satellite data and in-situ measurements. Regarding the

satellite data there is a wide range of oceanic variables

that will be used to retrieve and validate the surface

currents and frontal structures as indicated in Table 1,

including sea surface temperature (SST), chlorophyll

(Chl) observations, surface geostrophic current,

significant wave height and wave length and

propagation direction.

Tools for Optimizing Performance of VOYages at Sea

J.A. Johannessen

, A. Perrin

, L. Gaultier

, S. Herlédan

, C. Pouplin

, F. Collard

, J.P.Maze

M. Dussauze

, J. Rapp

, R. Fanebust

, S. Andersen

, O. Franks

& R. Meyer

Nansen Environmental and Remote Sensing Center, Bergen, Norway

OceanDataLab, Plouzané, France

Actimar, Brest, France

CMA-CGM, Marseille, France

Grieg Star, Bergen, Norway

Nelson Mandela University, Port Elisabeth, South Africa

CSIR, Cape Town, South Africa

ABSTRACT: The aim of the TOPVOYS project supported by the MarTERA ERA-Net Cofund program within

the European Commission is to advance and implement analyses tools and decision support system for voyage

optimisation. Based on marine weather analyses and forecasts combined with near real time satellite-based

observations of wind, wave and surface current conditions as well as sea surface temperature fields the best

shipping route are examined. The proposed approach aims to identify the optimum balance between

minimisation of transit time and fuel consumption as well as reduction of emissions without placing the vessel

at risk to damage and or crew injury. As such it is compliant with the International Maritime

Organization guidelines [6] for ship routeing to keep the traffic smooth and avoid accidents, notably in the

presence of unfavorable marine meteorological conditions. The tool performances will be demonstrated both in

post-voyage analyses and real time operations for the North Atlantic Ocean crossings, voyages from Europe

through the Mediterranean Sea and the Suez Channel to the Far East (e.g. China, South Korea) and voyages

around Southern Africa.

http://www.transnav.eu

the International Journal

on Marine Navigation

and Safety of Sea Transportation

Volume 15

Number 1

March 2021

DOI: 10.12716/1001.15.01.25

234

Table 1. Key satellite sensor data (level, resolution, provider). Note that radar altimeter data (wave height) are available in

the CMEMS multi-observation data set.

__________________________________________________________________________________________________

Sensor Product Level Resolution Data Provider

__________________________________________________________________________________________________

Sentinel-3 SLSTR SST and SEVIRI Sea surface temperature/fronts L2 ~ 1 km EUMETSAT

Sentinel-3 OLCI Chl Chlorophyll/fronts L2 ~ 300 m EUMETSAT

Sentinel-3 and Jason altimeters Surface geostrophic current/fronts L3 ~ 10 km CLS/Salto Duacs

Sentinel-3 and Jason altimeters Significant wave height L3 ~ 10 km CLS/Salto Duacs

Sentinel-2 spectral imager Wave length - direction/glitter L2 ~ 1 km ODL

Sentinel-1 A/B SAR Wave length - direction L2 ~ 1 km Scihub/ESA

Sentinel-1 SAR Doppler shift Radial surface current L3 ~ 2 km Scihub/ESA

CMEMS-Multi-Obs (Global) All above from Sentinel-3 L3/L4 ~ 10 km CMEMS

__________________________________________________________________________________________________

Table 2. In-situ sensor data and providers

__________________________________________________________________________________________________

Sensor Key products/resolution Coverage Data providers

__________________________________________________________________________________________________

HF radars Surface current/ order km surface EMODNET PHYSICS

Loch (ship-based) Surface current/ tens of meters surface CMA CGM (Watch Report)

Argo Surface current/ ~100 m surface CMEMS, Coriolis

Surface drifting buoys Current/~100m 15m depth CMEMS, Coriolis

__________________________________________________________________________________________________

Table 3. Complementary model-based surface current fields. *The GlobCurrent fields is an interpolated regular global

surface current product derived from satellite data. Geostrophic balance and Ekman current estimation applied.

__________________________________________________________________________________________________

Product Coverage Resolution Model Provider

__________________________________________________________________________________________________

CMEMS-GLOBAL global ~ 8 km NEMO CMEMS

RTOFS global ~ 8 km HYCOM NOAA

GOFS global ~ 8 km HYCOM NRL

MED-CMEMS Mediterranean Sea ~4 km NEMO CMEMS

IBI Iberian Peninsula & Bay of Biscay ~2 km NEMO CMEMS

GlobCurrent* global ~ 25 km Geo/Ekman CMEMS

Wave Model global ~ 10 km MFWAM MeteoFrance

__________________________________________________________________________________________________

Importantly, these satellite data can often be

complemented and collocated with in-situ data

allowing comparison of the surface current and

frontal structures derived from the satellite data to the

Argo floats, surface drifter data, HF-radars and on-

board estimates of surface currents as shown in Table

Finally, the satellite and in-situ based observation

data are combined and extended with surface current

and wave field forecast products offering global and

regional coverages at spatial resolutions ranging from

25 km to 2km as shown in Table 3.

A major innovation in this project is the systematic

use of satellite observations of the marine

environment in near real time to generate information

products tailored to ship locations and their planned

course for the next 24 hours. Presently, the joint EU-

ESA Copernicus program

(https://marine.copernicus.eu) ensures routine access

to the sea surface current, significant wave height,

wave spectra and sea surface temperature derived

from the Sentinel satellite missions (see Table 1).

These variables, in turn, allows the identification and

location of meandering surface current frontal

boundaries and eddies, evidence of wave-current

interactions and presence of crossing seas.

Satellite data regularly collected over time is also

highly useful to establish climatology that function as

reference conditions for assessing the magnitude of

the departure of the near real time product from the

climatology mean. This is illustrated in Figure 1

displaying the 4-year mean of significant wave height,

significant wave height gradient and surface

geostrophic current vorticity (estimated from the

gradient in meridional minus zonal current). Not

surprisingly the roughest sea state conditions are

found in the Southern Ocean with a mean significant

wave height between 4 and 5 m. In comparison, the

mean significant wave height in the North Atlantic

and North Pacific respectively ranges between 3-4 m

and 2-3m.

On the other hand, when looking at the mean of

the significant wave height gradient and the surface

geostrophic current vorticity the pictures largely

change towards the manifestation of the boundaries of

the basin-scale surface current system such as the Gulf

Stream, the Kuroshio Current and the greater Agulhas

Current, known to reach surface current speeds of 1-2

m/s. These intense current regimes are recognized

with strong mesoscale and sub-mesoscale variabilities

that have large influence on the sea state, in particular

due to the change in wave heights invoked by wave

refraction from the spatially varying surface current

[9]. As noticed in Figure 1, the two fields show a

significant degree of collocated expressions of distinct

anomalies in both the significant wave height gradient

and surface geostrophic current vorticity. This is a key

indicator of strong wave-current interaction, notably

caused by:

− refraction of the longer waves (> 200 m) as they

propagate across the surface current boundaries

and feel the significant change in surface

geostrophic current and associated vorticity field;

− steepening of the waves and in particular the

shorter wind waves (< 50 m) as they propagate

against the strong surface currents.

Wave refractions by the surface current are

observed in both Sentinel-1 Synthetic Aperture Radar

(SAR) images and Sentinel-2 multispectral images

(under cloud free conditions) revealing both the

incident wavelength and direction and their changes

when propagating across the surface current

boundaries. Moreover, complementary collocated

235

observations from Sentinel-3 deliver measurements of

the surface current and sea surface temperature

(under cloud free conditions).

Figure 1. 4-year mean (2013–2016) for significant wave

height (top), significant wave height gradient (middle) and

surface geostrophic current vorticity (lower) computed

using the constellation of 4 satellite altimeters and projected

on a 0.5° * 0.5° grid. The color bars mark the value in the

given units.

An example of multi-sensor satellite-based

observations of the spatial variability in the significant

wave height, the wave propagation direction and the

surface geostrophic current blended with swell

propagation simulations is shown in Figure 2 with

focus on the core of the Agulhas Current.

Figure 2. Daily surface geostrophic currents on 2016.02.28

(velocity as black arrows). The magenta lines map the swell

propagation rays. The magenta circles give the Sentinel-1

wave-mode images location the same day. Two Jason-2

altimeter tracks are shown, whose significant wave height

values are normalized to fit the current scale [12].

Evidence of altimeter-based observations of strong

wave-current interactions are clearly depicted in

which increased significant wave height are collocated

with areas of intense surface geostrophic currents, in

particular for the opposing currents such as seen near

the retroflection region of the Agulhas Current

centered around 39°S and 18°E. The complementary

simulated wave-current refraction are highly in

consistence with these observations and reveals how

the refractions lead to changes in wave propagation

and the set-up of crossing seas. The importance of the

wave-current interaction can also be explored from

the model simulations as highlighted by the

comparison of the significant wave height field with

and without the presence of the surface current (see

Figure 3). As noticed the significant wave height is

enhanced by around 50% in the core of the Agulhas

Current. Hence, the convergence (growth) of wave

energy (higher sea states) and directional spreading

(dangerous crossing seas) can be located both from the

observations and the model simulations leading to a

more reliable assessment of potential navigational risk

areas. The assessment may also yield more confidence

in model predicted sea state and location of unwanted

extreme waves.

Figure 3. Significant wave height from WW3 sea state

model for Sept 9, 2015 at 0h UTC. (left) without surface

current; (right) relative variations of significant wave height

( Hs) when considering surface current from the CMEMS

operational Mercator model (from [9] pers. com).

3 DEVELOPMENT OF THE ROUTING

OPTIMIZATION TOOL

An innovative tool aiming at providing value-added

surface current products currently tailored to the

Mediterranean Sea, North Indian Ocean, East Asia

seas, North Atlantic, South Atlantic and seas around

Southern Africa is under development and testing.

The products are provided both from available

forecasts and from observations. A series of post-

processing routines have been developed in order to

help the ocean forecaster build the optimized surface

current forecast. Different proxies have been defined

to qualify the surface current forecast performances at

each in-situ measurement location. Moreover, the

comparison of models with satellite derived sea

surface temperature and surface geostrophic current is

used to assess the ability of the ocean models to locate

the mesoscale structures (e.g. eddies, meanders

fronts). This analysis provides comparison scores

ranging from 1 (poor) to 5 (excellent) and is tailored to

both current direction and current magnitude as

indicated below.

These scores are established automatically through

comparison with direct current estimation made on

the bridge (e.g. Watch Reports), with surface drifters

and visually by comparison with SST field. The closer

236

the score is to 5, the better is the agreement between

the forecast product and the observations. Hence,

according to the scores within a subregion, the routing

software (ACTIROUTE) proposes an optimized and

qualified surface current field used by external

software to select the most preferred route. Otherwise,

the most direct route is followed. In order to facilitate

the decision support for the operator which makes the

analysis of the performance of the various forecast

products, the current maps from the different

products can be overlaid and displayed in the same

SEAScope visualization portal. The operator can then

quickly verify if the products and observations are

coherent. In the same way, the scores obtained by

comparing the forecast products to the observations

are saved in a format readable by SEAScope.

4 VALIDATION

In the TOPVOYS study, new diagnostics have been

implemented to validate surface currents using tracer

observations such as the sea surface temperature and

the Chlorophyll. The position of the dynamical

current structures can then be assessed when a

satellite-based SST or Chlorophyll map is available.

The frontal structures are extracted from the tracer

image and compared with the Lagrangian Coherent

Structures derived from the surface currents. To

extract the frontal structures, an algorithm consistent

with [1] and [2] is implemented to locate the position

of the fronts as schematically illustrated in Figure 4.

For each moving window, a histogram is used to

detect different population. The points that are

separating the two population are considered as the

representation of the surface front. For each point, a

probability of having a front is then estimated by

counting the number of times it has been detected on

the moving window. A contour-following processing

on the probability for the presence of the front is then

performed to reconstruct the frontal structures.

Figure 4. Algorithm for extraction of fronts from a tracer

image

Next, in order to compare the frontal structure

with the different available velocity products it is

necessary to compute a proxy for each velocity

product using the Finite-Size Lyapunov Exponent

(FSLE) to reveal the possible position of the tracer

gradient in the velocity field (e.g. [5, 10]). From the

FSLE image a contour following algorithm is then

applied to retrieve the corresponding frontal

structures followed by a comparison to the

corresponding fronts derived in the tracer image as

shown in Figure 5. This yield estimates of the distance

between the fronts as well as the differences in

curvature and direction and enables the selection of

the best velocity for each points along the front.

Figure 5. Validation of the velocity field (k) using the tracer

image (i)

An example of the practical use of this validation

tool is shown in Figure 6. This is based on

reconstruction of a surface current field from the

satellite-based SST frontal maps (derived from

SEVIRI) followed by an interpolation onto the grid of

the surface current products derived from

GlobCurrent. The two surface current maps are then

compared and assessed for consistency and accuracy.

Only points containing frontal information are used

for this validation. The comparison demonstrates that

the mesoscale anticyclonic eddy is satisfactorily

positioned in both fields although there are slight

differences in both surface current magnitudes and

directions.

In the following a more comprehensive test case is

presented for the region extending from the Gulf of

Aden to the East of the Socotra Island. This area (see

Figure 7) encompasses various hydrodynamic

features (frontal areas, mesoscale eddies, meandering

currents) that in some cases may require a potential

change in routing. The challenge is thus to precisely

locate the position of these features. Hence, the

comparison shall preferably enable the assessment of

the different products in terms of quality and

reliability in order to select the best route.

237

Figure 6. Comparison of independent surface current fields.

The white arrows represent the surface current field derived

from the SST frontal map and re-interpolated on the grid to

validate against the black arrows independently derived

from the GlobCurrent products.

Figure 7. Schematic hydrodynamic features in the mouth of

the Gulf of Aden (extracted from [4] revealing presence of

anticyclonic (A) and cyclonic (C) eddies and meandering

surface currents.

The results of the comparison between the FSLE

retrieved from different observation-based and

model-based surface current velocity fields and the

SST fronts derived from the SEVIRI product in the

North-West of the Arabian Sea in February 2021 are

shown in Figure 8. Inside the Gulf of Aden (yellow

square), the fronts detected from the SEVIRI SST

display the edges of two rings (consistent with

structures in Figure 7). The distances between these

SST-based fronts and the FSLE-based fronts are

smaller for the HYCOM product (~20 km: blue/green

color) than for Mercator or the Total Current derived

from observations (~80-100 km: brown color). This

implies that HYCOM model should be considered for

the routing in this area. In contrast, at the mouth of

the Gulf of Aden (red rectangle), the distance between

the FSLE computed from the Mercator model field

and the fronts detected from SEVIRI is on average

smaller than the other two products.

The routing (following the ACTIROUTE software)

will update its optimization procedure with these new

observations (SST) by considering the metrics

retrieved from the comparison between FSLE-based

and SST-based frontal locations, orientation,

structures and curvatures. In particular, it will

penalize the local products with larger separation

distance between the observed-based and model-

based fronts. This work is still under progress.

Figure 8. Comparison between the FSLE-based fronts (white

lines) and the observed SST-based fronts (colour). The

colour represents the distance from SST to FSLE based

fronts from dark blue (0 km) to dark red (80 km). Each line

represents the result of one velocity field: FSLE based fronts

from the observed total surface current (top), from the

HYCOM model (middle) and from the MERCATOR model

(bottom) The background grey-scale image is from the

SEVIRI L3 SST product (missing data due to cloud cover).

5 SUMMARY

In this paper we have used near real time satellite data

and in-situ data of the surface current and sea surface

temperature fields for assessment and optimization of

the surface current field for ship routing. It has been

demonstrated that synoptic maps of surface frontal

structures provide highly important products and

information on meandering patterns and motions

which are proxy for the surface currents dynamics,

and as such allowing assessment and validation on

the quality of the delivered surface current products

with emphasis on the upper 10 to 20 m. Moreover,

regular use of wave ray-tracing model with different

surface currents will be run for simulations of rapidly

changing and possibly occurrences of extreme waves

invoked by wave-current interaction.

HYCOM

Mercator

238

The ultimate goal is to advance the development of

a decision support system for optimization of ship

routing that provides a reliable traffic-light system by

which indices for the pre-selected ship routes builds

on regular near real time updates of:

− meandering fronts and eddies;

− rapidly changing currents;

− evolving wind sea and swell fields;

− likelihood of wave energy focusing caused by

wave-current interaction;

− likelihood of crossing seas;

− likelihood of dangerous waves.

The provision of these indices will be based on the

combination satellite-based and and-situ based data

sources and model fields including:

− surface current from the GlobCurrent database;

− model-based surface current (CMEMS, etc)

− sea state from wave models with and without

wave-current interaction;

− altimeter derived significant wave height data;

− Sentinel-1 SAR-based wave mode and

interferometric wide acquisitions;

− Satellite-based sea surface temperature fields;

− ECMWF wind field and wind stress

REFERENCES

1. Cayula, J.-F., Cornillon, P.: Edge Detection Algorithm for

SST Images. Journal of Atmospheric and Oceanic

Technology. 9, 1, 67–80 (1992).

https://doi.org/10.1175/1520-

0426(1992)009<0067:EDAFSI>2.0.CO;2.

2. Cayula, J.-F., Cornillon, P.: Multi-Image Edge Detection

for SST Images. Journal of Atmospheric and Oceanic

Technology. 12, 4, 821–829 (1995).

https://doi.org/10.1175/1520-

0426(1995)012<0821:MIEDFS>2.0.CO;2.

3. Chang, Yu-Chia , Ruo-Shan Tseng, Guan-Yu Chen, Peter

C Chu and Yung-Ting Shen (2013), Ship Routing

Utilizing Strong Ocean Currents, The Journal of

Navigation 2013, doi:10.1017/S0373463313000441.

4. Fratantoni, D.M., Bower, A.S., Johns, W.E., Peters, H.:

Somali Current rings in the eastern Gulf of Aden.

Journal of Geophysical Research: Oceans. 111, C9,

(2006). https://doi.org/10.1029/2005JC003338.

5. Gaultier, L., Verron, J., Brankart, J.-M., Titaud, O.,

Brasseur, P.: On the inversion of submesoscale tracer

fields to estimate the surface ocean circulation. Journal

of Marine Systems. 126, 33–42 (2013).

https://doi.org/10.1016/j.jmarsys.2012.02.014.

6. International Maritime Organization (IMO) IG927E

Ships' Routeing, 14th Edition 2019, IM927, ISBN: 978-

92-801-0049-5 (9789280100495).

7. International Maritime Organization (2010). Reduction

of GHG emissions from ships, IMO-MEPC 61/INF.22, 2

August 2010.

8. Kim, J.-G., Kim, H.-J., and Lee, P. T. W (2013).

“Optimising container ship speed and fleet size under a

carbon tax and an emission trading

scheme.” International Journal of Shipping and Transport

Logistics, vol. 5, no. 6, pp. 571–590. DOI:

10.1504/IJSTL.2013.056835.

9. Marechal, G., Ardhuin, F.: Surface currents and

significant wave height variability: a numerical

investigation of the Agulhas current region. Earth and

Space Science Open Archive. 18 (2020).

https://doi.org/10.1002/essoar.10503641.1.

10. d’Ovidio, F., Fernández, V., Hernández-García, E.,

López, C.: Mixing structures in the Mediterranean Sea

from finite-size Lyapunov exponents. Geophysical

Research Letters. 31, 17, (2004).

https://doi.org/10.1029/2004GL020328.

11. Pennino, Silvia, Salvatore Gaglione, Anna Innac,

Vincenzo Piscopo and Antonio Scamardella (2020),

Development of a New Ship Adaptive Weather Routing

Model Based on Seakeeping Analysis and Optimization,

J. Mar. Sci. Eng. 2020, 8, 270; doi:10.3390/jmse8040270.

12. Quilfen, Y., Chapron, B.: Ocean Surface Wave-Current

Signatures From Satellite Altimeter Measurements.

Geophysical Research Letters. 46, 1, 253–261 (2019).

https://doi.org/10.1029/2018GL081029.

13. Yang, Liqian , Gang Chen, Jinlou Zhao and Niels Gorm

Mal Rytter (2020), Ship Speed Optimization

Considering Ocean Currents to Enhance Environmental

Sustainability in Maritime Shipping, J. Mar. Sci. Eng.

2020, 8, 270; doi:10.3390/jmse8040270.