SYNERGISTIC USE OF FERRYBOX AND ENVISAT DATA WITH THE …€¦ · Remote sensing from space offers...

6
SYNERGISTIC USE OF FERRYBOX AND ENVISAT DATA WITH THE AID OF A HYDRODYNAMIC TRANSPORT MODEL Hajo Krasemann and Wilhelm Petersen GKSS-Research Centre, 21502 Geesthacht, Germany ABSTRACT An automatic measuring system called FerryBox was installed on a ferry travelling between Cuxhaven (Ger- many) and Harwich (Great Britain) enabling on-line oceanographic and biological measurements such as salinity, temperature, fluorescence, turbidity, oxygen, pH, and nutrient concentrations. The high frequent sampling of data allows to detect even short-term events. Obser- vations made along the ferry transect reveal characteris- tic phenomena such as high salinity inflow through the Channel into the southern Bight, algal bloom dynamics and related oxygen and pH changes. For the same time period data from ENVISAT (MERIS) was used to re- trieve water content concentrations of chlorophyll, total suspended matter and yellow substance. Combination of these in-situ on-line observations with remote sens- ing reveales an improved knowledge of coastal water sta- tus. Examples of the synergy between both operational measuring strategies are shown, both for large scale algal blooms in the North Sea as well as for local intense but short term blooms in the German Bight. Coherence of the data sets was improved by using hydrodynamic trans- port models in order to obtain synoptic overviews of the remotely sensed and FerryBox related parameters. The importance of future applications of these combination of methods for monitoring of coastal waters is emphasised. Key words: Ferrybox; Meris; Monitoring; Model. 1. INTRODUCTION Monitoring of a highly dynamic system such as coastal waters requires dense sampling in space and time in or- der to catch short-term events which might have a strong impact on the coastal ecosystem, such as exceptional phy- toplankton blooms or changes caused by storms. Existing observations mostly lack the spatial coverage and tempo- ral resolution required to determine the state of the marine environment and changes therein. The lack of monitoring systems that provide continuous observations of the ma- rine environment in the coastal areas and shelf seas of Eu- rope is a serious hindrance to understand these systems. Remote sensing from space offers the opportunity to get almost daily synoptic estimates of water quality param- eters over large areas. The global coverage of satellites allows for the estimation of water quality in remote and inaccessible areas. However, there are also significant limitations of satellite observations. The ability to dis- tinguish between the various constituents of the water is limited to parameters which can be correlated to optical properties. These are concentrations of suspended sedi- ment, phytoplankton, and yellow substance whose major part is coloured dissolved organic matter. Data are lim- ited by cloud coverage and the signal is a depth integral depending on the clarity of the water and varies in coastal regions between a few centimeters and 10 meters. Open ocean (Case 1 waters) reflectance is mainly affected by chlorophyll and related degradation pigments and allows good estimates of chlorophyll concentrations. In estuar- ine and coastal waters (Case-2 waters) it is more diffi- cult to discriminate between chlorophyll, suspended mat- ter, and yellow substance concentrations because of high interdependencies of these water constitutents and their specific optical properties which vary intime time and for different regions. Therefore, site specific algorithms have to be developed for these types of waters [3]. Currently, in situ monitoring is mainly carried out by manual sampling and analysis during ship cruises and autonomous measuring systems on buoys. In order to increase the temporal and spati2005al coverage the use of ships of opportunity (SoO) has been promoted by Eu- roGOOS [2]. Recently, sophisticated systems have been implemented on ferry boats which allow precise measure- ments of temperature, salinity and chlorophyll and even currents by ADCP. In the FerryBox-Project supported by the European Union eleven groups worked together in or- der to enhance the use of SoO for operational monitoring of the marine environment [6]. SoO on regular routes of- fer a cheap and reliable measuring platform to obtain con- tinuous and spatial high resolution observations of near surface water parameters. Such systems are able to mea- sure a whole set of oceanographic and biological param- eter which are also observed by satellites (temperature, chlorophyll, turbidity, yellow substances). However, the data of FerryBox systems are restricted to the transect and to a certain depth (depth of the inlet of the FerryBox) which is a limitation for stratified waters. In this paper we will show how FerryBox transect data can be combined _____________________________________________________ Proc. ‘Envisat Symposium 2007’, Montreux, Switzerland 23–27 April 2007 (ESA SP-636, July 2007)

Transcript of SYNERGISTIC USE OF FERRYBOX AND ENVISAT DATA WITH THE …€¦ · Remote sensing from space offers...

Page 1: SYNERGISTIC USE OF FERRYBOX AND ENVISAT DATA WITH THE …€¦ · Remote sensing from space offers the opportunity to get almost daily synoptic estimates of water quality param-eters

SYNERGISTIC USE OF FERRYBOX AND ENVISAT DATA WITH THE AID OF AHYDRODYNAMIC TRANSPORT MODEL

Hajo Krasemann and Wilhelm Petersen

GKSS-Research Centre, 21502 Geesthacht, Germany

ABSTRACT

An automatic measuring system called FerryBox wasinstalled on a ferry travelling between Cuxhaven (Ger-many) and Harwich (Great Britain) enabling on-lineoceanographic and biological measurements such assalinity, temperature, fluorescence, turbidity, oxygen, pH,and nutrient concentrations. The high frequent samplingof data allows to detect even short-term events. Obser-vations made along the ferry transect reveal characteris-tic phenomena such as high salinity inflow through theChannel into the southern Bight, algal bloom dynamicsand related oxygen and pH changes. For the same timeperiod data from ENVISAT (MERIS) was used to re-trieve water content concentrations of chlorophyll, totalsuspended matter and yellow substance. Combinationof these in-situ on-line observations with remote sens-ing reveales an improved knowledge of coastal water sta-tus. Examples of the synergy between both operationalmeasuring strategies are shown, both for large scale algalblooms in the North Sea as well as for local intense butshort term blooms in the German Bight. Coherence ofthe data sets was improved by using hydrodynamic trans-port models in order to obtain synoptic overviews of theremotely sensed and FerryBox related parameters. Theimportance of future applications of these combination ofmethods for monitoring of coastal waters is emphasised.

Key words: Ferrybox; Meris; Monitoring; Model.

1. INTRODUCTION

Monitoring of a highly dynamic system such as coastalwaters requires dense sampling in space and time in or-der to catch short-term events which might have a strongimpact on the coastal ecosystem, such as exceptional phy-toplankton blooms or changes caused by storms. Existingobservations mostly lack the spatial coverage and tempo-ral resolution required to determine the state of the marineenvironment and changes therein. The lack of monitoringsystems that provide continuous observations of the ma-rine environment in the coastal areas and shelf seas of Eu-rope is a serious hindrance to understand these systems.

Remote sensing from space offers the opportunity to getalmost daily synoptic estimates of water quality param-eters over large areas. The global coverage of satellitesallows for the estimation of water quality in remote andinaccessible areas. However, there are also significantlimitations of satellite observations. The ability to dis-tinguish between the various constituents of the water islimited to parameters which can be correlated to opticalproperties. These are concentrations of suspended sedi-ment, phytoplankton, and yellow substance whose majorpart is coloured dissolved organic matter. Data are lim-ited by cloud coverage and the signal is a depth integraldepending on the clarity of the water and varies in coastalregions between a few centimeters and 10 meters. Openocean (Case 1 waters) reflectance is mainly affected bychlorophyll and related degradation pigments and allowsgood estimates of chlorophyll concentrations. In estuar-ine and coastal waters (Case-2 waters) it is more diffi-cult to discriminate between chlorophyll, suspended mat-ter, and yellow substance concentrations because of highinterdependencies of these water constitutents and theirspecific optical properties which vary intime time and fordifferent regions. Therefore, site specific algorithms haveto be developed for these types of waters [3].

Currently, in situ monitoring is mainly carried out bymanual sampling and analysis during ship cruises andautonomous measuring systems on buoys. In order toincrease the temporal and spati2005al coverage the useof ships of opportunity (SoO) has been promoted by Eu-roGOOS [2]. Recently, sophisticated systems have beenimplemented on ferry boats which allow precise measure-ments of temperature, salinity and chlorophyll and evencurrents by ADCP. In the FerryBox-Project supported bythe European Union eleven groups worked together in or-der to enhance the use of SoO for operational monitoringof the marine environment [6]. SoO on regular routes of-fer a cheap and reliable measuring platform to obtain con-tinuous and spatial high resolution observations of nearsurface water parameters. Such systems are able to mea-sure a whole set of oceanographic and biological param-eter which are also observed by satellites (temperature,chlorophyll, turbidity, yellow substances). However, thedata of FerryBox systems are restricted to the transectand to a certain depth (depth of the inlet of the FerryBox)which is a limitation for stratified waters. In this paper wewill show how FerryBox transect data can be combined

_____________________________________________________

Proc. ‘Envisat Symposium 2007’, Montreux, Switzerland 23–27 April 2007 (ESA SP-636, July 2007)

Page 2: SYNERGISTIC USE OF FERRYBOX AND ENVISAT DATA WITH THE …€¦ · Remote sensing from space offers the opportunity to get almost daily synoptic estimates of water quality param-eters

with large scale satellite data to improve the quality ofmonitoring of coastal waters. We focused on two aspects:the measuring accuracy as well as the spatial and tempo-ral resolution of FerryBox and remotes sensing data. Aspecific problem is posed by the temporal difference be-tween the two types of observations which is shown to besolved by using water transport models.

2. MATERIALS AND METHODS

The ”German FerryBox” system consists of an automatedflow-through system with different sensors (temperatureand salinity (FSI, USA), in-vivo fluorescence as a proxyfor chlorophyll-a (SCUFA-II, Turner Design, USA), algalgroup analyser (AOA, bbe-moldaenke, Germany), tur-bidity (Turner Design, USA and Endress Hauser, Ger-many), oxygen and pH (Endress Hauser, Germany) andautomatic wet chemical nutrient analysers (APP, ME-Grisard, Germany) (dissolved inorganic nitrate, ammo-nia, o-phosphate and silicate) and an additional UV-nitrate detector (ProPS, TRIOS, Germany) for optical de-tection of nitrate. A schematically diagram is shown inFig. 1 :

Figure 1. Schematical view of the FerryBox flow-throughsystem.

In this contribution for chlorophyll-a fluorescence datawere used. At certain times water automatically sampledby the FerryBox were analysed by HPCL according tothe procedure described by Wiltshire et al. [10] in or-der to validate the measured chlorophyll values. It turnedout that the ratio between in-vivo chlorophyll-a fluores-cence and chlorophyll-a concentration analysed by HPLCshows a seasonal behavior as the chlorophyll-a fluores-cence depends on the species as well as on the physiolog-ical status of the algal. In order to make the data com-parable for the chlorophyll-a values the calibration of themanufacturer was used. Thus, these values may differfrom time to time from measured chlorophyll-a concen-trations in the lab by the HPLC method.

The water intake is in front of the ships cooling systemat a fixed depth (5 m). The water is pumped continu-ously along the sensors. The water temperature measuredby the FerryBox is influences by heating up in the tube

to the FerryBox. Differences between intake and Ferry-Box were maximally up to 0.4◦C based on measurementswith a sensor at the intake and the sensor in the FerryBox.Data including time and position (latitude, longitude) areautomatically transferred to shore and the system can beremotely operated via mobile phone connection (GSM).Spatial resolution of recording is about 100 m with theexception of the data of the chemical nutrient analyserswhich have a spatial resolution of 6-9 km depending ontime needed for the analysis. On shore data enters a SQLdatabase (Oracle) with access from a public web page(http://ferrydata.gkss.de). Water samples for subsequentanalysis in the laboratory can be taken by an automaticwater sampler at predefined positions. The samples arekept refrigerated (4◦C) and in the dark. A more detaileddescription of the system is reported in Petersen et al. [5].

The German FerryBox system was installed on the ferry’Duchess of Scandinavia’ (DFDS Seaways, Copenhagen,Denmark) on the route between Cuxhaven (Germany)and Harwich (Great Britain) since 2002. The ferry trav-els along the southern part of the North Sea coveringthe coastal zones of Germany and the Netherlands andcrosses the inflow of the English Channel into the NorthSea. The map of the route is shown in Fig. 2. The ferryis scheduled to travel daily between the two ports, mainlyovernight.

Figure 2. Map of the Southern North Sea and transect ofthe ferry (Duchess of Scandinavias route).

On its route the Ferry crosses several water masses thatare separated by permanently existing fronts with onlylimited water exchanges. Near Cuxhaven the watermasses are largely influenced by the outflow of the Elberiver. Despite the low salinities off Cuxhaven the turbid-ity is not very high since the turbidity zone of the Elbeestuary is located more upstream in the inner estuary.Along the Dutch Wadden Sea lower salinities are foundthat originate from the fresh water inputs of rivers and theIJsselmeer. In the further course of the route the influenceof the riverine input decreases and the water masses areinfluenced by water masses with Atlantic water proper-ties which reach the track of the ferry through the EnglishChannel. Approaching the English Coast a slight de-crease of salinity is observed and the turbidity increasesstrongly reaching even higher values then in the Elbe es-tuary due to long term erosion along the English coast.Due to the strong tidal energy the water column is moreor less vertically mixed along the transect throughout theyear [7]. This justifies the application of a FerryBox sys-tem although only one water horizon is monitored withthe applied system.

Page 3: SYNERGISTIC USE OF FERRYBOX AND ENVISAT DATA WITH THE …€¦ · Remote sensing from space offers the opportunity to get almost daily synoptic estimates of water quality param-eters

Remote sensing and in-situ measurements are in generalnot carried out at the same time. The movement of watermasses in between can be calculated by application ofnumerical models. We used the operational model system[1] of the Federal Maritime and Hydrographic Agency(BSH) to compute the drift and dispersion of substances(Lagrangian and Eulerian dispersion models).

In order to compare the remote sensing data and the mea-surements of the FerryBox the drift of the water massessampled by the FerryBox was estimated by applying theLagrangian dispersion model. The drift was calculatedwithin the time lag between each FerryBox observationand the satellite overpass, thus achieving observations ona ship route synoptic with the remote sensed data.

The remote sensed data were taken by the MEdium Reso-lution Imaging Spectrometer Instrument (MERIS) on theENVISAT satellite launched in 2002. MERIS measuresthe solar radiation reflected by the earth, at a maximalground spatial resolution of 300 m, in 15 spectral bands,from the visible to the near infra-red. We used partly300 m-resolution-data (FR, Full resolution) and due tohigher availability 1200 m-resolution-data (RR reducedresolution). MERIS allows global coverage of the earthin 3 days. For medium latitudes on two out of three daysMERIS-data are available.

We used Level-1b data, which are among others cali-brated radiances at top of the atmosphere. For Level-2-data, the Meris-level-2-products were used which yieldedthe geophysical data products on total suspended matter,yellow substance and chlorophyll-a concentration. Theseproducts are intended for blue open ocean water (Case-1),the product called algal-1 relying mainly on reflectanceratios. For a wider range of concentrations that can befound in coastal waters and which are characterised byhigher turbidity and higher yellow substance absorptionsthe products tsm (total suspended matter), ys (yellow sub-stance) and algal-2 were applied. These case-2 productsare determined by an algorithm taking into account all9 bands of MERIS in the visible(except for fluorscence)and then invert the forward model by the use of a multiplenon-linear regression method [8], shortly referred to as aneural-net.

3. RESULTS AND DISCUSSION

3.1. Comparison of the algal dynamics in the South-ern North Sea in spring 2004 and 2005

The observations of single transects with the FerryBoxcan be pooled in a contour plot in order to get an overviewof the temporal and spatial variability of a parameter.Fig. 3 shows algal dynamics along the route between Har-wich and Cuxhaven in spring 2004, represented by thevalues of in-vivo chlorophyll-a fluorescence.

In 2004 a weak spring bloom started in late March offthe Dutch coast (4.5 - 6.2◦E), followed by a first stronger

Figure 3. In vivo fluorescence of chlorophyll-a along thetransect of the ferry between March and June 2004.

Figure 4. In vivo fluorescence of chlorophyll-a along thetransect of the ferry from March to June 2005.

bloom later in early April between 3.8◦E and 5.8◦E. Thebloom spreaded into the German Bight between the endof April and early May, most likely by advective trans-port with movement of the water masses along the Dutchand German coast. In May, the first strong bloom oc-curred in the water masses that reach the Ferry transectfrom the English Channel (2.3 - 4.2◦E). It drifted towardsthe Dutch coast and disappeared in mid-June. After thesespring blooms in the Channel and along the Dutch andGerman coast the chlorophyll concentrations remainedlow and weak blooms with high spatial patchiness wereobserved between 5◦E and 7.5◦E at the end of June. Lateron, there was no visible algal activity with exception ofa short-term event happening in the German Bight at thebeginning of August which will be discussed in detail inthe following section.

In 2005 the development of the spring bloom observedalong the transect by FerryBox is shown in Fig. 4. Sim-ilar to the bloom dynamics in 2004, weak chlorophyllconcentrations were found in March and the first bloomstarted at the Dutch coast between 4.5◦E and 6◦E at theend of March 2005. Whilst in 2004 the bloom spread outto the German coast in 2005 the bloom remained on afixed location and strongly increased at mid of April.

The bloom in the water mass reaching the ferry transectfrom the English Channel between 3.2◦E and 4.2◦E ini-tiated earlier (mid of April) than in 2004 and lasted un-til June with a comparable drift to the Dutch coast as in2004. Both blooms disappeared end of May and at thebeginning of June 2005 respectively. A second bloom

Page 4: SYNERGISTIC USE OF FERRYBOX AND ENVISAT DATA WITH THE …€¦ · Remote sensing from space offers the opportunity to get almost daily synoptic estimates of water quality param-eters

appeared later in water masses that reached the ferry tran-sect from the English Channel between 3.2◦E and 4.2◦E.It started in mid of April and continued until June with adrift to the Dutch coast. Despite of slightly different tim-ing these observations are very similar to the measure-ments in 2004. Both blooms disappeared at the end ofMay and at the beginning of June 2005, respectively.

In 2005 again an intensive bloom was observed in theGerman Bight off the Elbe estuary starting in May anddisappearing mid of June 2005. This bloom is charac-terised by considerably higher chlorophyll concentrationsthan in 2004 (about 1.5 times higher).

3.2. Spring bloom observed by remote sensing in2005

In the North Sea the availability of suitable remote sens-ing images is often restricted by cloud coverage. In 2005a considerable set of more or less cloud free images couldbe obtained during spring time. Unfortunately most partof the scenes water content concentrations are flagged asunreliable. As this flag is known to be restrictive in thisproduct version (IPF 4.x) (pers.com. Roland Doerffer)we modified for this analysis the minimal quality crite-rion. The reliability for each part is discussed below indetail. Thus, the spatial distribution of the blooms at thelocations observed by the FerryBox in 2005 can be ex-tracted from remote sensing images and compared to thedata of the FerryBox. Fig. 5 shows the distribution ofderived chlorophyll-a concentrations by MERIS on 11thand 21st of April as well as on 15th and 28th of May.In addition to each image the comparison of the MERISchlorophyll-a concentrations along the track of the ferrywith the FerryBox chlorophyll-a fluorescence are shown.The FerryBox positions were drift corrected for the timethe satellite passed the North Sea. For all comparisonsFerryBox data from the particular cruise starting fromHarwich on the day before the image was taken wereused. Thus the time difference between the satellite im-age and the FerryBox observation is at maximum 19hours in the western part but reduces to 2 hours in theeastern part.

On 11th of April a strong bloom along the Belgian andDutch coast and near shore of the German coast can bedistracted from the satellite images. The comparison ofthe extracted chlorophyll values with the chlorophyll-afluorescence of the ferry sampled on the cruise from Har-wich to Cuxhaven on 10th to 11th of April shows goodagreement and even the shoulders of the curves fit wellfrom the Channel to the Dutch-German border (6◦E).Along the English coast very high suspended matter con-centration of 25 to 60 mg/l can be derived from remotesensing. The high signal from suspended matter maycause an overestimated chlorophyll result west of 2.2◦E.From 6.5◦E and eastwards the agreement with the Ferry-Box data degrades. For this part of the image the low-quality flag is raised and remote sensed data should beinterpreted with care expecting strong deviations. In ad-

Figure 5. Chlorophyll-a concentration derived fromENVISAT-MERIS for the North Sea and comparison withextracted chlorophyll-a fluorescence from the FerryBox.Black lines show the tracks of the ferry. a) 11th of April,b) 21st of April c) 15 of May, d) 28th of May.

dition around 5◦E are some smaller clouds which alsocause a flagging of the data.

On 21st of April only low to moderate chlorophyll wasfound in the algal-2 signal from MERIS along the ferrywhereas the FerryBox detected enhanced levels up to 22µg/l chlorophyll-a off the Dutch coast. Almost the wholescene is flagged as unreliable but reasonable gradientsfrom coast to outer regions and a weak bloom in theGerman Bight are visible. It confirms the importance ofquantified error indicators in the standard product fromESA.

In May a wide spread bloom was observed in the South-ern Bight between longitude 3◦E and 4◦E and a strongbloom with chlorophyll-a concentrations up to 25µg/l isobserved along the Dutch coast and in near shore Belgianwaters starting from outflow of the Rhine estuary by theFerryBox. In the German Bight a bloom was detectedwhich was only touched by the ferry track at its south-ern edge. In the satellite image of 15th of May the broadpeak at 5◦E seems to lack the eastern shoulder. The qual-ity flag was raised for the whole scene. We look hereinto the data anyhow but the disagreement at the easternshoulder is very likely the effect of sun-glint hamperingMERIS. It occurs when the sensor is directed more to-wards the sun and higher winds causing waves to moreoften reflect direct sun-light towards the sensors. In thisscene the glint effect is maximal between longitudes 5◦Eand 7◦E.

Two weeks later on 28th of May the bloom in the South-ern Bight region still persists with a small drift to thenorth. At this time the bloom along the Dutch coastshows a high patchiness indicating the break down of thisbloom as also can be seen in the contour plot of the Fer-ryBox data (Fig. 4). The pattern of the bloom is char-acteristic for the algal species Phaeocystis forming re-current blooms in the North Sea when it approaches its

Page 5: SYNERGISTIC USE OF FERRYBOX AND ENVISAT DATA WITH THE …€¦ · Remote sensing from space offers the opportunity to get almost daily synoptic estimates of water quality param-eters

Figure 6. Chlorophyll-a concentrations derived fromMERIS compared with chlorophyll-a fluorescence fromthe FerryBox. Green are data from the ferry, red aredata from MERIS along ferry track. The lower figureshows MERIS data extracted along a drift-model cor-rected track. The grey shaded areas are unreliable dueto high sun glint.

culmination (pers. com. Rudiger Rottgers) and was alsoobserved by remote sensing in 2003 [4]. This patchinessof chlorophyll-a fluorescence as measured along the trackby the FerryBox only partly corresponds with the patternsof chlorophyll-a derived from remote sensing. The sceneof this day is strongly influenced by sun-glint with themaximum between 6◦E and 8◦E. The low-quality flag israised for almost the whole image. At the end of May thebloom in the German Bight already disappeared.

In general the comparisons showed a good agreement ofmain features between remote sensed data and FerryBoxmeasurements. Some drawbacks are visible. A modelingwhere each water parcel measured by the ferry is trans-ported by tidal and wind induced currents to the time ofthe satellite overpass clearly improves the agreement ofthe shape of the curves in most cases as can be seen inFig. 6

The difference in observed chlorophyll concentrationsbetween FerryBox and remote sensed measuring is ex-pectable. The algorithms used to produce the remotesensed standard coastal data are not yet tuned to their op-timum. The methods to determine chlorophyll are verydistinctive. In the FerryBox system a measurement offluorescence is used to derive chlorophyll concentrations.This is highly influenced by physiological status of the al-gae. Remote sensing of chlorophyll is always a determi-nation of absorption. The specific absorption is depend-ing on species as well and may easily vary by a factorof two, for instance between the species Chatonella andthe diatoms. On top of these uncertainties is a profoundvariation between measurements of chlorophyll in labo-ratory by means of HPLC in the order of 20% standarddeviation possible, as round robin tests with several lab-oratories showed [9].

4. CONCLUSION AND OUTLOOK

FerryBox systems proved to be a cost effective tool formonitoring coastal and shelf seas with an extensive setof water quality parameters (oceanographic as well asbiological relevant data e.g. chlorophyll, oxygen, pHand nutrients). The data availability is nearly indepen-dent from weather conditions exceptional extreme stormevents and a full coverage in time and persistent trackscan be achieved as in contrast to surveys by research ves-sels.

The spatial restriction to the transect can be overcomefor certain parameters such as chlorophyll-a and sus-pended matter by combining FerryBox measurementswith remote sensing. Although the detection methodsfor chlorophyll-a are different (FerryBox: in-vivo fluores-cence, remote sensing data: derived from absorption andscattering from the surface layer) the comparisons of thechlorophyll data from FerryBox and MERIS show thatthe results of both methods in most cases are comparablewithin the limits of the used methods. This is especiallyremarkable for such difficult regions like the coastal zoneof the North Sea with high contents of other constituentsinfluencing the measured reflectance of remote sensing.However, at some specific situations some strong differ-ences between both methods still occur. Thus, the algo-rithms have to be improved. It can be expected that withthe next generation of site specific algorithms for deriv-ing chlorophyll-a concentration from reflectance remotesensing data are even more reliable. Since the FerryBoxalso allows automated sampling of water samples alongthe track these samples can be used together with the con-tinuous in-vivo chlorophyll-fluorescence measurementsto get a dense data set of ground truth data for calibrationand validation of remote sensing. This will provide a con-siderable improvement in the accuracy of the calculatedchlorophyll data. Thus, the spatially covered informationcan be provided with competent accuracy even for Case2 waters.

The results obtained in this investigation show that theassimilation of FerryBox data along a transect with satel-lite observations increase the information value comparedwith the use of either of the two individual informationsources. By using FerryBox systems on more lines cross-ing the area of interest on different tracks the informationdensity about the water quality could be significantly im-proved. In addition, by including the information aboutmany other water quality parameters measured by theFerryBox (e.g. nutrients etc.) a much deeper insight inthe processes controlling the water quality of coastal wa-ters can be obtained.

The value of such a combination of ship-of-opportunitydata with remote sensing can even be improved furtherby including numerical model data: The assimilation ofthese data with numerical models can be used to simu-late the circulation of the water masses in order to fill thegaps between remote sensing and transect data in timeand space. For instance, the disadvantage of the restricted

Page 6: SYNERGISTIC USE OF FERRYBOX AND ENVISAT DATA WITH THE …€¦ · Remote sensing from space offers the opportunity to get almost daily synoptic estimates of water quality param-eters

availability of satellite data in areas such as the North Seadue to cloud coverage can be overcome by applying a nu-merical model with data assimilation which fills the gapsbetween the satellite passes. This investigation showedalready how important the hydrodynamic models in suchstrongly tidal influenced areas as the North Sea are fora comparison between different measurements with dif-ferent spatial resolution at slightly different times. Moreresearch has to be carried out in this field, e.g., to includeprocesses such as algal growth and degradation whichmay occur in the time span between the different mea-surements.

Since both methods used in this investigation are re-stricted to the water surface conventional monitoringmethods by buoys and research vessels are still necessaryat strategic locations. This will be necessary, for instance,to get information about the change of water constituentswithin the depth profile.

One of the most important aspects of the relatively newFerryBox technology is its role in not only improving re-mote sensing products but as well fostering the under-standing of marine processes: The lack of reliable tempo-ral and spatial high resolved biological data hampered inthe past the assimilation into numerical models. With thepotentials of the FerryBox systems this situation could beimproved in future and severe progress will be possiblein marine ecosystem modeling resulting in an operationalintegrated monitoring system by data assimilation of in-situ data as well as remote sensing observations.

ACKNOWLEDGMENT

We thank the ferry company DFDS Seaways (Copen-hagen, Denmark) for the possibility to operate the Fer-ryBox onboard of the Duchess of Scandinavia and grate-fully acknowledge the support by the ship crew. Theauthors also thank Michail Petschatnikov and MartinaGehrung for operating the FerryBox, Jan Bodewadtfor developing the software for data management andStephan Dick, Hartmut Komo and Dieter Schrader fromFederal Maritime and Hydrographic Agency (BSH) forperforming the drift model calculations.

REFERENCES

[1] Dick, S., Kleine, E., Muller-Navarra, S.H., Klein, H.,Komo, H., 2001. The Operational Circulation Modelof BSH (BSHcmod) - Model description and valida-tion. Berichte des Bundesamtes fur Seeschifffahrt undHydrographie, Nr. 29, Hamburg, Germany, 49 pp.

[2] Flemming, N.C., Vallerga, S., Pinardi, N., Behrens,H.W.A., Manzella, G., Prandle D., Stel, J.H., 2002.Operational Oceanography: implementation at theEuropean and regional seas. Proceedings SecondInternational Conference on EuroGOOS, ElsevierOceanography Series Publication series 17, Amster-dam, Netherlands.

[3] Hellweger, F.L., Schlosser, P., Lall, U., Weissel, J.K.,2004. Use of satellite imagery for water quality stud-ies in New York harbour. Estuarine, Coastal and ShelfScience 61, 437-448.

[4] Peters, S.W.M., Eleveld, M., Pasterkamp, H., vander Woerd, H., Devolder, M., Jans, S., Park, Y., Rud-dick, K., Block, T., Brockmann, C., Doerffer, R.,Krasemann, H., Rottgers, R., Schonfeld, W., Jrgensen,P.V., Tilstone,, G., Martinez-Vicente, V., Moore, G.,Srensen, K., Hkedal, J., Johnson, T.M., Lmsland, E.R.,Aas, E., 2005. Atlas of Chlorophyll-a concentrationfor the North Sea based on MERIS imagery of 2003.Vrije Universiteit, Amsterdam, Netherlands ISBN 90-5192-026-1

[5] Petersen, W., Petschatnikov, M., Schroeder, F., Col-ijn, F., 2003. Ferry-Box Systems for MonitoringCoastal Waters. in: Dahlin, H., Flemming, N.C.,Nittis. K., Petersson, S.E.: Building the EuropeanCapacity in Operational Oceanography, Proc. ThirdInternational Conference on EuroGOOS, ElsevierOceanography Series Publication series 19, Amster-dam, Netherlands, pp. 325-333.

[6] Petersen, W., Colijn, F., Dunning, J., Hydes, D.J.,Kaitala, S., Kontoyiannis, H., Lavin, A.M., Lips,I., Howarth, M.J., Ridderinkhof, H., Pfeiffer, K.,Sørensen, K., 2005. European FerryBox Project: FromOnline Oceanographic Measurements to Environmen-tal Information. In: Dahlin, H., Flemming, N.C.,Marchand, M., Petersson, S.E. (Eds.): European Op-erational Oceanography: Present and Future, Proc.Fourth International Conference on EuroGOOS, ISBN92-894-9788-2, pp. 551-560

[7] Rodhe, J., 1998. The Baltic and the North Seas: Aprocess-oriented review of the physical oceanography.In: Robinson, A.R., Brink, K.H. (Eds.), The Sea, Vol11, John Wiley and Sons, Inc, New York, pp. 699-732.

[8] Schiller, H., Doerffer, R., 1999. Neural network foremulation of an inverse model operational derivationof case II water properties from MERIS data. Interna-tional Journal of Remote Sensing, 20, 9, 1735-1746.

[9] Tilstone, G.H., Moore, G.F., Sørensen, K., Rottgers,R., Jørgensen, P.V. Martinez-Vicente, V.,. Ruddick,K.G., 2002. REVAMP - Regional Validation ofMERIS Chlorophyll products in North Sea coastalwaters. REVAMP Inter-calibration Report; Contract:EVG1 CT 2001 00049.

[10] Wiltshire, K. H., Harsdorf, S., Smidt, B., Bloecker,G., Reuter, R., Schroeder, F. 1998. The determinationof algal biomass (as chlorophyll) in suspended matterfrom the Elbe estuary and the German Bight: a com-parison of HPLC, delayed fluorescence and promptfluorescence methods. Journal of Experimental Ma-rine Biology and Ecology. 222, 113 131