oceanography

Oceanography

Toward an Assessment of Ocean Acidification inthe Adriatic Sea and Impacts on the Biogeochem-istry of Marine Carbonate System

A. Luchetta, C. Cantoni, G. Catalano, S. CozziInstitute of Marine Sciences, CNR, Trieste, [email protected]

Abstract

The increase of CO2 amount in the atmosphere has created great concern: itwill in all probability result in changes in temperature, precipitation and/or their sea-sonal amplitudes with consequences not only on sea level rise but also on chemicalequilibrium of the CO2 system in seawater, mainly reducing pH and carbonate ionconcentration (Ocean Acidification). The process is now well documented in fielddata from all around the world. However is not sufficiently witnessed in the Mediter-reranean Sea, due to the scarcity of good quality data. On this concern, results for theAdriatic Sea are presented: from experimental measures of pH and total alkalinity,two seasonal pictures of pH and carbonate system parameters have been drawn. Inaddition, a pH decrease of 0.063 pHT units with related chemistry changes has beeninferred in the North Adriatic Dense Water (NAdDW) over the two last decades.These results, although preliminary, merit attention as confirm that N. A. sea hasbeen affected by OA, being sensitive to the climate forcing. Potential impacts of OAare several and should be assessed, as many might even exacerbate hyopoxia/anoxiaevents, already affecting the area. OA might also affetc the food web, as the car-bonate reduction has the potential to alter the distribution and abundance of marineorganisms that use calcium carbonate to build their shells or skeletons (corals, plank-ton) and the organisms that depend on them for survival (fishes, marine mammals).

1 Introduction

Over the past 250 years the carbon diox-ide (CO2) concentration in the atmospherehas continuously raised, reaching values upto 380 ppmv, that have never been expe-rienced on Earth in the last 800,000 years[1]. It is now largely recognized that thisincrease was mainly determined by humanactivities related to the combustion of fos-sil fuels and deforestation, which, in allprobability will result in changes in tem-perature, precipitation and/or their seasonal

amplitudes with consequences on the sta-bility of our climate. Since the beginningof the industrial age, between 1800 and2000, mankind has emitted 361 Gt C tothe atmosphere [2]. The ocean has ab-sorbed approximately 155 Gt C: this makethe world ocean the largest sink of anthro-pogenic CO2, without it atmospheric car-bon dioxide levels would be approximately450 ppmv today [2, 3]. The uptake of CO2by the ocean is primarily due to physico-chemical processes. As CO2 solubilises inseawater, it behaves like a weak acid that

Oceanography

dissociates according to:

CO2(atm) CO2(aq) + H2O

H2CO3(aq)

H+(aq) + HCO3 (aq)

H+(aq) + CO23 (aq),

leading to an increase of [H+] and to adecrease of pH value (pH = -log [H+]).The overall process resulting in a reduc-tion of pH and shifts in carbonate specia-tion is referred to as Ocean Acidification(OA). Climate change and ocean acidifica-tion are both caused by the increasing ofatmospheric CO2 levels and ocean acid-ification has been recently referred to asThe other CO2 problem [4]. However, ifclimate change forecasts suffer from someuncertainties, in contrast OA is a well pre-dictable consequence of rising atmosphericCO2. On global scale, OA is now doc-umented with hydrographic surveys, timeseries data and well verified from models[5, 6, 7, 8, 9, 10].Since preindustrial times, the averageoceanic surface pH has fallen down byapproximately 0.1 units, from approxi-mately 8.21 to 8.10 [11], and is ex-pected to decrease a further 0.30.4 pHunits [9] if atmospheric CO2 concen-trations reachs 800 ppmv (the projectedend-of-century concentration according tothe Intergovernmental Panel on ClimateChange (IPCC) business-as-usual emissionscenario). Ocean acidification alters sea-water speciation and biogeochemical cy-cles of many elements and compounds, in-cluding nitrogen, phosphorus, silicon andtrace elements (iron, zinc), thus chang-ing their availability for phytoplankton [4].Acidification of ocean water occurs in tan-dem with decreases in carbonate ion (CO=3 )

concentration and saturation state of cal-cium carbonate minerals (CaCO3), whichdirectly impact the formation and disso-lution. In the marine environment, car-bonate formation is largely a biotic pro-cess: corals, foraminifera, coccoliths, bi-valves and other marine organisms formshells and skeletons composed of a varietyof carbonate minerals.CO2 solubility in seawater is the highestat low temperatures, thus the most pro-nounced effects of OA on marine ecosys-tems are expected to affect sub polar seasand not the Mediterranean Sea, a semi-enclosed water body in a temperate cli-mate region [12]. In addition Mediter-ranean seawater is characterized, on aver-age, by higher alkalinity than open ocean,that would increase the buffering capacityof the Mediterranean waters thus limitingOA. Notwithstanding, ocean acidificationis particularly interesting to be investigatedin the Mediterranean Sea as the basin issupposed to be very sensitive to the globalclimatic change (giving a rapid response),because of the faster water renewal (shorterresidence times of water masses) comparedto the oceans and the high anthropogenicpressure concurrently with the high CO2carrying capacities of the cold surface wa-ters in the northernmost regions. Conse-quently the Mediterranean area would al-ready present significant pH drops [12], es-pecially in the regions where cold densewaters are formed. But the OA processin the Mediterranean Sea is not sufficientlywitnessed, due to the scarcity of good qual-ity pH measurements, particularly in theEastern sub-basin [12, 13].The Adriatic basin is one of the few siteswhere dense waters are formed in wintryseason. This process represents an im-portant driving engine for the circulationand ventilation of deep waters of the east-

1244

Marine research at CNR

ern Mediterranean Sea. Adriatic densewaters form either on the northern shal-low shelf of this basin, North AdriaticDeep Water (NAdDW), and by the deepSouthern Adriatic Pit, Adriatic Deep Water(ADW). Afterward, they usually sink andoutflow through the Otranto Strait sill (750m), which controls the export to Ionianand Eastern Mediterranean Seas [14]. TheAdriatic Sea has been therefore consideredthe dominant source region of dense wa-ters for the Eastern Med until the occur-rence of the Eastern Mediterranean Tran-sient [15], in the end of 1980s, whichabruptly changed the deep circulation pat-tern. At present time, the deep circula-tion scheme seems to have switched backto pre-Transient conditions.The Adriatic basin is subjected to highanthropogenic pressure, being surroundedby very industrialized regions that releaseCO2 into the atmosphere. There car-bon dioxide uptake and ocean acidificationwould be particularly effective, due to lowtemperature of surface waters, especiallyin the northernmost part which representsthe largest shelf area of the entire Mediter-ranean region [14]. According to the deepand bottom layers circulation scheme theAdriatic dense water masses are expectedto have the possibility of spreading acid-ified waters around, through the EasternMediterranean.Therefore monitoring the interannual vari-ability of pH and studying the forcings onthe carbonate system in the Adriatic Seaappears worth and new. Here are presentedthe major findings of the research activitycarried out by ISMAR Trieste in last fewyears, within a few national and europeanprojects and in collaboration with local en-vironmental agency.

2 MethodsThe determination of pH was performed bythe spectrophotometric method describedby Dickson [16], values are expressed onthe total H+ scale (pHT , [H+] in molH+/kgSW ), at 25 C (as recommendedby protocols for quality control of results),with a precision of 0.003 pH units.To our knowledge the dataset is the firstcollected with such a precision over thewhole basin. The Total Alkalinity (AT )has been measured by potentiometric titra-tion in an open cell, with precision of 3.0/kgsw, the accuracy was controlled againstcertified reference materials (CRM) sup-plied by Andrew Dickson (Scripps Institu-tion of Oceanography, San Diego, USA).Both the in situ fugacity of carbon dioxide(fCO2) and the in situ pH values were cal-culated with the CO2 calculation program(CO2SYS program) developed by Lewisand Wallace [17] by using the parameterspHT , AT , silicate, phosphate, T in situ andS for each discrete sample. However, pHdistributions in Figures 1, 2, 3, are shown atfixed temperature (25 C) because fixing thetemperature means lock the equilibriumreactions of the carbonate system, gettingrid of the temperature contribution. Thusany comparison between different watermasses and different seasons is more im-mediate.

3 Results and discussion

3.1 pHT spatial and seasonalvariability in the AdriaticSea

In Figures 1 and 2 are reported the distri-butions of potential density (t), apparentoxygen utilization (AOU) and pHT gath-

1245

Oceanography

Figure 1: Potential density (t), apparent oxygen utilization (AOU), pHT at 25 C distri-butions along the Adriatic Sea in February 2008. In the smaller frames a more detailedview of the North Adriatic shelf area is given.

ered at meso scale in the Adriatic basin.They were measured during two surveys,conducted in February and October 2008,within the frame of VECTOR (national)and SESAME (EU FP-6) projects.In February 2008 the North Adriatic shelfwas involved in a dense water forma-tion event, being shallow and exposed tocold dry winds (Bora) as reported alsoin the past [18]. The water column waswell mixed, ventilated down to the bottom(mean AOU = -4.1 M), cool (10.35 C)and dense (t > 29.3 kgm3). The T/Sproperties were in agreement with those ofNAdDW, that are among the densest of theMediterranean Sea [19].The whole water column was rich of DIN

(1.00-7.00 M) and SiO2 (1.205.33 M),even at surface, thus suggesting that intenseprimary production had not yet started.The pHT values were homogeneously dis-tributed in the water column, ranging be-tween 7.917 and 7.973 pHT units, witha mean value of 7.946 in the core ofdense water mass (NAdDW). Biologicalprocesses such as primary production andremineralization of organic matter, couldcontribute to the final pH of seawater byconsuming or adding CO2. Such a homo-geneous distribution of all parameters overan extended area mirrored the winter timeconditions encountered. The pHT valueswere driven by the high CO2 solubility incold seawater while intense biological pro-

1246


Figure 2: Potential density (t), apparent oxygen utilization (AOU), pHT at 25 C distri-butions along the Adriatic Sea in October 2008. In the smaller frames a more detailedview of the North Adriatic shelf area is shown.

cesses had not yet started (as indicated byAOU and nutrients values), that is typicalof wintry season.fCO2 values calculated at fixed tempera-ture (25C) were high (avg 583 atm) thusindicating that such cold and ventilated wa-ters had adsorbed high CO2 amount, thatdetermined the decrease of pHT value ob-serveds. At the same time fCO2 values,calculated at the in situ temperature, rangedfrom 222.4 to 334.6 atm from surfaceto the bottom over the whole area. Theyresulted much lower than the equilibriumvalue with atmospheric CO2 (398 atm,mean value on measurements conducted onboard). This clearly indicated the occur-rence of under saturated conditions underwhich the northern Adriatic shelf region

was a potential sink for atmospheric CO2.Generally, NAdDW water mass flowssouthward and accumulates at the bottomof the Meso and Southern Adriatic pits(250 and 1250 m, respectively) [14] asevidenced in Figure 1 by density, higherthan 29.3 and 29.2 respectively, at the bot-tom. The dense waters of Meso Adri-atic pit exhibited pHT values lower (6.0 M , DIN > 5.0 M).In the southern part of the section, a veryclear event of deep convection was ob-

1247

Oceanography

served on February 2008 in the deepestcentral stations, as pointed out by the ho-mogeneous water column down to 600 m(t around 29.15-29.16 kgm3 down to600 m, in Figure 1). pHT had an aver-age value of 7.947 0.003 pHT units,AOU was positive (15-35 M) and nutri-ents were homogeneously distributed downto 600m. Such values suggest that the deepconvection allowed mixing between sur-face and older (AOU > 0) CO2 enricheddeep waters: for such reason surface AOUwere positive in contrast to what occurringalong the whole basin.These results are in agreement to whatreported in literature for the area, whichis known to be dominated by a quasi-permanent cyclonic circulation intensify-ing in autumn and creating the conditionsfor the production of dense and oxygenateddeep waters during winter deep convectionevents [14]. However fCO2 in situ values,around 370-380 atm, were still slightlylower than the equilibrium value with at-mospheric CO2 (398 atm). At the begin-ning of October 08 the situation appearedcompletely changed as the whole Adriaticbasin was characterized by the thermal anddensity stratifications typical of late sum-mer conditions.In the northern shallow shelf region sea wa-ter temperature varied between 13.0 C to19.4 C from surface to bottom with an av-erage value of 17.6, 7.3 degrees higher thanFebruary. The upper water column wascharacterized by higher pHT values (be-tween 7.960 and 8.050 pHT units, Figure2) than in February, due to primary pro-duction process witnessed by the releaseof oxygen (negative AOU values) and thedepletion of all nutrients. In contrast, thelayer below pycnocline, was strongly de-pleted of oxygen (AOU mean value = 102.3M), more acidic (7.784 pHT units) and

enriched in CO2 and nutrients (DIN upto 22.9 M; fCO2>600 atm) due to theremineralization of newly produced POCand labile DOC, all is characteristic of theautumn season.The bottom water mass in the Meso Adri-atic pit in October was about 0.1 Ccolder than in February, less saline (0.1psu) and more oxigenated (DO 20 mol/Lhigher), thus indicating at least the par-tial renewal of this water mass with theNAdDW formed during previous winter onthe northern shelf. Also in October pHTmean value was low (7.861 pHT units,in Figure 2) and mainly determined byremineralization processes as evidenced byhigh AOU values (> 60 M) and nutrientmaxima (DIN up to 7.0 M, SiO2 up to 9.0M).The southern part of the section wasalso characterized by late summer ther-mal and density stratifications (13.178 < T< 21.082 C), with the upper layer domi-nated by primary production processes (0< AOU < 15) and characterized by higherpHT (7.950 < pH < 8.0). In the layers be-low the pycnocline, remineralization pro-cesses prevailed as suggested by oxygenand nutrient values (AOU > 40 M; DIN> 3.0 M; SiO2 > 5.0 M). Seasonal pri-mary production provides POC (in differ-ent amounts) to the euphotic layers, whichis available to the microbial community forremineralization during sinking to the bot-tom. This phenomenology was probably atthe basis of the pHT shift observed in thebottom water mass of the Southern Adri-atic pit between February (pHT = 7.937)and October (pHT = 7.898).This preliminary comparison between pHdistributions and oceanographic conditionsmet in February and October 2008 pointedout the high spatial and seasonal variabil-ity of pHT in the Adriatic sea. The fi-

1248


nal values were determined by the com-bined effects of circulation patterns, tem-perature driven CO2 solubility and biologi-cal processes (primary production, increas-ing pHT values, and respiration of organicmatter, lowering them).However the most significant conclusionwe can infer is the fact that the lower pHTvalues of surface waters in winter were de-termined by the high dissolved CO2 con-centrations (due to atmospheric CO2 forc-ing). In contrast the drawdown of dissolvedCO2 by photosynthetic planktonic organ-isms led to higher pHT values in late sum-mer.

3.2 Long term variability ofpHT in the dense waters ofthe northern Adriatic Sea

The first comparison between two setsof data related to the dense cold waters(Northern Adriatic Dense Water) formed,respectively, in winters 1982-1983 and2007-2008 has been recently published[20]. Values of pH on the NBS scale fromthe old dataset were converted to the newtotal hydrogen ion concentration scaleadopted for the new dataset and expressedin +/kgsw, as recommended by the inter-national scientific community. Some re-sults at 25 C are summarized in Table 1,they show the decrease of both pHT aver-age value (-0.063 pH units) and carbonateion concentration (-19.6H+/kgsw), whichnamed OA. In contrast, the total alkalinity,dissolved inorganic carbon and CO2 fugac-ity exhibit net increases over the same pe-riod.

After an analysis of the different forc-ing (total alkalinity and dissolved carbondioxide) impacting on water masses dur-ing the two seasons and between the two

winters, the net increase of dissolved CO2resulted to be the driving factor of theobserved inter-decadal acidification [20].This important result confirms that theAdriatic Sea is sensitive to atmospheric gassolubilisation (as CO2) and indicates thatOA has been affecting the Adriatic ma-rine waters for the last 25 years. It alsoindicates the need for a careful checking,in the coming decades, of the acidifica-tion rates as the impact on water quality,marine ecosystems and fishery resourcescould be not negligible. Although a deter-mination of acidification rates is not pos-sible on the base of only 2 specific yearsbecause interannual variations must alsobe considered, we inferred an approximateacidification rate over this time span. Itcorresponded to 0.0025 pH units/year, inagreement with acidification rates calcu-lated in other oceanic regions from timeseries: at ESTOC station in the openocean, Atlantic, Canary islands 0.0017 0.0004 pH units/year [21] 0.0012 0.0004pH units/year, in the open Atlantic ocean,Bermuda Islands, [22].

3.3 Time series in the Gulf ofTrieste (northern AdriaticSea)

The acquisition of time series, at leastby a few key sites such as those wheredense water formation occurs in winter, isa promising strategy to monitor ocean acid-ification rates and impacts in the Mediter-ranean sea.On this concern, ISMAR Trieste has re-cently started the collection of pHT andother biogeochemical parameters time se-ries in the Gulf of Trieste (very shallow, thenorthernmost of the Mediterranean Sea),which is representative of a coastal envi-

1249

Oceanography

Param. Units N samples

Average Std Dev Median

1983 pHT mol/kgsw 33 8.010 .046 8.005 Tot. Alk mol/kgsw 33 2584.5 10.9 2584.8 DIC mol/kgsw 33 2256.3 32.1 2260.9 fCO2 atm 33 485.1 63.6 489.2 H2CO3 mol/kgsw 33 13.6 1.8 13.7 HCO3- mol/kgsw 33 2003.8 49.0 2011.1 CO3= mol/kgsw 33 239.0 19.5 236.1 Revelle 33 9.773 0.469 9.809 Ca 33 5.60 0.46 5.54 Ar 33 3.71 0.30 3.66 2008 pHT mol/kgsw 56 7.946 0.012 7.947 Tot. Alk mol/kgsw 61 2658.9 18.1 2658.1 DIC mol/kgsw 54 2366.6 21.6 2370.3 fCO2 atm 54 593.4 22.3 593.2 H2CO3 mol/kgsw 54 16.6 0.6 16.6 HCO3- mol/kgsw 54 2130.6 23.9 2137.4 CO3= mol/kgsw 54 219.4 4.9 219.5 Revelle 54 10.491 0.168 10.493 Ca 54 5.14 0.11 5.14 Ar 54 3.40 0.07 3.40

Table 1: Values of the carbonate system parameters (at 25 C) in the Northern AdriaticDense Water mass formed in winter 1983 and 2008.

ronment (Figure 3).Since January 2008, pHT , AT and themayor biogeochemical and physical pa-rameters were acquired on monthly basison the whole water column at the coastalsite PALOMA (centre of the Gulf, 25mdeep, close to the mast PALOMA - Ad-vanced Oceanic Laboratory PlatforM forthe Adriatic sea, 45 37 N, 13 34 E). Firstresults evidenced a complex time evolutionof pHT , mainly driven by the combined ef-fect of strong changes in both temperatureand production/ remineralisation processes

(Figure 3). During winter pHT values weregenerally low (7.868-7.958, avg 7.920) andhomogeneous owing to the increased CO2solubility driven by the low water temper-ature (down to 8.0C) and by the absenceof intense production processes. Duringspring and summer pHT was highly vari-able and mainly driven by the biologicalprocesses: the highest values (up to 8.120,June 2008) were reached in the upper layerduring high production events (AOU= -34M) and the lowest values (down to 7.648,August 2008) in the bottom layer during

1250


Figure 3: Map of the sampling site PALOMA in the Gulf of Trieste; distribution ofmonthly values of pHT (a) and AOU (b) at PALOMA station from January 2008 to July2009.

biomass remineralisation (AOU= 142 M).During Jan-March 2008 the oceanographicproperties (average t = 29.35 kgm3, T =8.84 C) and pHT values (7.907 0.028)of site PALOMA, indicating dense waterformation, fit well to the general NorthAdriatic Sea conditions over the same pe-riod. In summer, small scale biologicalprocesses prevailed in determining pHTvalues both in PALOMA site and in theNorth Adriatic, depicting a more complexsituation.From such preliminary data, this site lo-cated in the centre of the Gulf of Triesteresults to be a good indicator not onlyof coastal dynamics/ processes but also ofsub-basin wide (North Adriatic Sea) pro-cesses and dynamics.

4 Impacts on biogeochem-istry and marine ecosys-tem

The most significant indication we candraw from the preliminary study onNAdDW is that the observed pH de-crease has affected the biogeochemistryof the carbonate system, causing specia-tion shifts (Table 1) as the decrease ofcarbonate ion concentration (from 236.1to 219.5 mol kg1sw ) accompanied bythe increases of bicarbonate ion and car-bonic acid concentrations (from 2011.1 to2137.4 mol kg1sw , from 13.7 to 16.6mol kg1sw , respectively) that fit well towhat generally observed in other oceanicregions [11]. Also the solubility ratios ofcalcite and aragonite show a decrease (Cafrom 5.51 to 5.14, Ar from 3.66 to 3.40)during last 25 years but, being far from1.0, they indicate that NAdDW is still over-saturated with respect to calcium carbon-

1251

Oceanography

ate, as already observed for the Mediter-ranean Sea [6]. Therefore, the AdriaticSea with its adequate carbonate saturationsstate seems to ensure still quite healthygrowth conditions for calcifying organisms(from plankton to benthic molluscs, echin-oderms and corals). The increase of theRevelle factor in the same period (R= 9.773 0.469 and 10.491 0.168, respectively)suggest a decrease of the buffering capac-ity of the whole carbonate system [23]).There are several impacts of OA on the ma-rine ecosystem which have not yet been in-vestigated. For example the observed pHdecrease might have impacted the carbonfixation capacity by photosynthetic organ-isms (calcifyer and non calcifyer), actuallya few papers report the possibility of anincrease of primary productivity [4] in re-sponse to ocean acidification. An increaseof primary production might further affecthypoxia and anoxia events, already sea-sonally occurring in the northern Adriaticbasin [24], through an increase of the mi-crobial respiration of the surplus of organicmatter. On another hand, the combinedimpacts of increased stratification, due tothe global warming, and changes in theocean biology, caused by ocean acidifica-tion, could cause a further decline in dis-solved oxygen concentrations as recentlyforecast by Oeschlies [25]. If this wouldoccur in the North Adriatic it might exac-erbate hypoxia and anoxia. Another sig-nificant impact of OA in the North Adriaticcould be represented by the increase of dis-

solved organic carbon (DOC) productionas response to the increase of primary pro-duction induced by OA [26]. If the DOCincrease would be assessed true also for theAdriatic Sea, it might have major effectson mucillages phenomena which affect thebasin, as mucillages are aggregates of mu-chopolysaccharide and would be favouredby higher DOC concentrations.In the end, all these aspects could havemassive consequences on marine resources(fishery, aquaculture, tourism). Hence theNorthern Adriatic Sea offers challengesfor future research activities of high pri-ority. Thus confirming to be an interest-ing basin where assessing the impacts ofOcean Acidification with the potential formarine organisms to adapt to increasingCO2, and broader implications for marineecosystems.

5 AcknowledgementThe authors thank: dr. Massimo Celio(ARPA-FVG) for the use of CTD data ac-quired at PALOMA station and dr. Ve-drana Kovacevic (OGS-Trieste) for the useof CTD data acquired during VECTORand SESAME campaigns. They are grate-ful to the captains and crews of R/V Ura-nia (CNR) and Effevigi (ARPA-FVG). Thestudy was funded by the VECTOR projectof the Italian Ministry for University andScientific Research and by the SESAMEproject of the European Commission.

References[1] D. Luthi, M. Le Floch, B. Bereiter, T. Blunier, J.M. Barnola, et al. High-resolution

carbon dioxide concentration record 650,000-800,000 years before present. Nature,453:379382, 2008.

1252


[2] C.L. Sabine, R.A. Feely, R.M. Gruber, N.and Key, and K. Lee et al. The oceanicsink for anthropogenic CO2. Science, 305:367371, 2004.

[3] C.L. Sabine and R.A. Feely. The oceanic sink for carbon dioxide. In: Greenhousegas sink, D. Ready, N. Hewitt, J. Grace and K. Smith (editors). pages 3149, 2007.

[4] S.C. Doney, V.J. Fabry, R.A. Feely, and J. A Kleypas. Ocean Acidification: TheOther CO2 Problem. Annual. Reviews Marine Science, 1:169192, 2009.

[5] K. Caldeira and M.E. Wickett. Anthropogenic carbon and ocean pH. Nature,425:365365, 2003.

[6] A. Schneider, D.W.R. Wallace, and A. Koertzinger. Alkalinity of the MediterranenaSea. Geophys. Res. Letters, 34(L15608), 2007.

[7] R.A. Feely, C.L. Sabine, K. Lee, W. Berelson, J. Kleypas, et al. Impact of anthro-pogenic CO2 on the CaCO3 system in the oceans. Science, 305:362366, 2004.

[8] R.A. Feely, J. Orr, V.J. Fabry, J.A. Kleypas, C.L. Sabine, and C. Langdon. Presentand future changes in seawater chemistry due to ocean acidifcation. In: Carbonsequestration and its role in the global carbon cycle, Brian J. McPherson and EricT. Sundquist (eds.). AGU Monograph, 2009.

[9] J.C. Orr, V.J. Fabry, O. Aumont, L. Bopp, S.C. Doney, R. A. Feely, A. Gnanade-sikan, N. Gruber, A. Ishida, F. Joos, R.M. Key, K. Linsay, E.M. Reimer, R. Matear,P. Monfray, A. Mouchet, R.G. Najjar, G.K. Plattner, K.B. Rodgers, C.L. Sabine,J.L. Sarmiento, R. Schlitzer, R.D. Slater, I.J. Totterdell, M.F. Weirig, Y. Yamanaka,and A. Yool. Anthropogenic ocean acidification over the twenty-first century andits impact on calcifying organisms. Nature, 437:681686, 2005.

[10] S. Solomon, D. Quin, and M. Mannings et al. Climate Change 2007: the physicalscience basis: contribution of working group I to the Fourth Assessment Report ofthe Intergovernamental Panel on Climate change. New York (2007). CambridgeUniv. Press, 2007.

[11] Royal Society. Ocean acidification due to increasing carbon dioxide, Policy Docu-ment 12/05. 2005.

[12] A. Yilmaz et al. Impacts of acidification on biological, chemical and physical sys-tems in the Mediterranean and Black Seas. In: CIESM Workshop Monographs N36. F. Briand editor, CIESM, Monaco. page 124, 2008.

[13] MEDAR Group. MEDATLAS/2002 database. Mediterranean and Black Seadatabase of temperature salinity and bio-chemical parameters. Climatological At-las. IFREMER Edition (4 Cd). 2002.

1253

Oceanography

[14] M. Gacic, A. Lascaratos, B.B. Manca, and A. Mantziafoul. Adriatic deep waterand interaction with the eastern Mediterranean Sea. In: Physical Ocenaography ofthe Adriatic Sea. Past, Present and Future. In: B. Cushman-Roisin, M. Gacic, M.P.Poulain and A. Artegiani (eds.), Kluwer Academic Publishers, The Netherlands.pages 111142, 2001.

[15] W. Roether, B. Manca, B. Klein, D. Bregant, D. Georgopoulos, V. Beitzel, V. Ko-vacevic, and A. Luchetta. Recent changes in the Eastern Mediterranean deep wa-ters. Science, 271:333335, 1996.

[16] A.G. Dickson, C.L. Sabine, and J.R Christian. Guide to best practices for oceanCO2 measurements. PICES Special Publication, 3:191, 2007.

[17] E. Lewis and D.W.R. Wallace. Program Developed for CO2 System Calculations.ORNL/CDIAC-105. Carbon Dioxide Information Analysis Centre, Oak Ridge Na-tional Laboratory, U.S. Department of Energy, Oak Ridge, Tennessee. 1998.

[18] P. Malanotte-Rizzoli. The northern Adriatic Sea as a prototype of convection andwater mass formation on the continental shelf. In: Deep convection and deep waterformation in the oceans. P.C. Chu and J.C. Gascard (eds.). Elsevier OceanographySeries, 57:229239, 1991.

[19] A. Artegiani, E. Paschini, A. Russo, D. Bregant, F. Raicich, and N. Pinardi. TheAdriatic Sea Circulation, Part 1: Air-sea interactions and water mass structure. J.Phys. Oceanography, 27:14921515, 1997.

[20] A. Luchetta, C. Cantoni, and G. Catalano. New observations of CO2 induced acid-ification in the Northern Adriatiic Sea, over the last quarter century. Chemistry andEcology, 26:117, 2010.

[21] J.M. et al. Santana Casiano. The interannual variability of oceanic CO2 in thenortheast Atlantic subtropical gyre at the ESTOC site. Global Biogeochem. Cycles,21(GB1015), 2007.

[22] N.R. Bates. Interannual variability of CO2 sink in the subtropical gyre of the NorthAtlantic Ocean over the last two decades. J. Geophys. Res., 112, 2007.

[23] F.J. Millero. The marine inorganic carbon cycle. Chemical Reviews, 107:308341,2007.

[24] R. Marchetti. The problems of the Emilia Romagna coastal waters: facts and inter-pretations. In: Marine Coastal Eutrophication. pages 2133, 1992.

[25] A. Oschlies, K.G. Schulz, U. Riebesell, and A. Schmittner. Simulated 21st cen-turys increase in oceanic suboxia by CO2-enhanced biotic carbon export. GlobalBiogeochemical Cycles, 22, 2008.

1254


[26] U. Riebesell, K. G. Schulz, R. G. J. Bellerby, M. Botros, P. Fritsche, Meyerhofer,et al. Enhanced biological carbon consumption in a high CO2 ocean. Nature,450:545548, 2007.

1255

Oceanography

1256

The Roads of the Sea - Can We Predict the Motionof Particles Carried by Ocean Currents?

A. Griffa1, K. Schroeder1, S. Aliani1, A. Doglioli2, A. Molcard3, V.Taillandier4, P.M. Poulan5, T. Ozgokmen6, A. Haza6)1, Institute of Marine Sciences, CNR, Pozzuolo di Lerici (SP), Italy2, University of Marseille, Marseille, France3, University of Toulon, Toulon, France4, Laboratoire dOceanographie de Villefranche-sur-mer, Villefranche-sur-mer, France5, National Institute of Oceanography and Experimental Geophysics, Trieste, Italy6, University of Miami, Rosenstiel School of Marine and Atmospheric Science, Miami,[email protected]

Abstract

Ocean currents play a fundamental role in the transport of substances and species.Being able to monitor and predict their effects is of great relevance for a number ofapplications, such as correct management of the coastal ecosystem, manage controlin case of discharge of pollutants and understanding of pathways of invasive species.While transport by ocean currents is under many aspects very complex and domi-nated by turbulent and chaotic processes, it has been shown in recent works that it isoften possible to find a hidden structure, at least for mesoscale motion, that guides themovement of the advected quantities. Barriers of motion exist in the ocean, related tothe main Lagrangian coherent structures, i.e. to structures such as gyres, jets andeddies. In this paper, we provide examples of methods to identify such barriers andapplications in the Mediterrean Sea. The limits of these methods, that are based onthe assumption that the velocity field is well known, are also discussed, and possibleremedies in terms of Lagrangian assimilation are discussed.

1 Introduction

Currents are the roads of the sea. Theytransport physical properties such as tem-perature and salinity (T,S), chemical prop-erties, pollutants, particulate and sedimentsas well as biological quantities such as phy-toplankton, zooplankton, larvae and jellyfish. Being able to understand and pre-dict transport by ocean currents is thereforecrucial for a number of applications. Theyinclude climatic applications, for instanceunderstanding heat transport or pathways

of species invasions, as well as applicationsfor a correct management of the coastalocean ecosystem and for damage control incase of accidents at sea such as dischargesof pollutants. Transport predictions is verychallenging for a number of reasons [1]. Tounderstand it, consider the basic equationof Lagrangian transport, i.e. the equationthat describes particles advected by the cur-rent,

dx/dt = u(x, t),

Oceanography

where x is the position of a particle andu is the velocity. The equation showsthat the trajectory of a particle, x(t), is theintegral of the velocity u(x,t). This im-plies that even small errors in the predic-tion of u (obtained from numerical mod-els or measurements) tend to accumulateand grow in the prediction of x(t). Sincein practice small errors in u are unavoid-able, due to incomplete knowledge of forc-ing, topography, coastline and to the influ-ence of small scale unresolved processes,we can expect that this will result in sig-nificantly amplified errors in trajectories.Also, the equation is inherently nonlin-ear, since u depends on the position x,and it has the property of being very of-ten chaotic. This implies that even forvery simple Eulerian flows u (in presenceof time dependence) trajectories are highlysensitive to initial conditions. Predictingthem is therefore very difficult, since evena slight difference in initial conditions inspace and time can result in significantlydifferent behaviours. Even though La-grangian prediction is highly challenging,a number of methods have been put forthin the past decade that have helped increas-ing our skills in this direction. Differentmethods have been suggested for differentapplications. Methods based on statisti-cal approaches are particularly suited forclimatic problems. They consist in sep-arating the mean component of the cur-rents from the turbulent and fluctuatingcomponent and parameterizing the turbu-lent part for instance using stochastic meth-ods [2, 3, 4]. Other methods are moresuited for the prediction of specific events,and they are typically based on dynamicalsystem theories. The basic concept hereis that even though the motion of a singleparticle is extremely challenging to repro-duce because of the high dependence on

initial conditions and on the details of theflow, the description of the general patternof transport is much more approachable.It has been suggested that ocean transportin quasi-geostrophic, quasi-2-dimensionalsituations is dominated by main coherentstructures [5], such as vortices, eddies andjets, that are separated by invisible barriers,i.e. regions that particle trajectories cannotcross. Methods from nonlinear dynamicalsystem theory have been proposed to locatesuch barriers, that can be used to provideinformation on the general fate of a parti-cle launched in a certain area. Details onthe specific trajectory might be difficult todetermine, but its general behaviour is ex-pected to be determined by such barriers.In this context, the knowledge of the loca-tion of hyperbolic points (HPs) that sepa-rate different structures appears especiallyrelevant. Various methods can be usedto identify such barriers and HPs, rang-ing from direct identification in terms offlow invariants to methods based on localdispersion properties, such as Finite Time(FTLE) or Finite Size (FSLE) LyapunovExponents [5, 6]. Dynamical system meth-ods appear to have a great potential forpractical ocean applications. Neverthelessit is important to point out that they are di-agnostic tools, in the sense that they canbe used with great results only as long asthe velocity field u is known with a cer-tain degree of accuracy. This is the casefor instance for velocity fields from exten-sive HF (High Frequency) radar measure-ments, or from accurate ocean circulationmodels. In many cases, though, predic-tions from circulation models are still in-complete and the structures can be con-sidered known only with some approxima-tion. In order to increase our knowledgeof such structures and therefore our predic-tion capability, assimilation methods can

1258


be used, that combine information fromreal time data with model results. In partic-ular, since we are interested in Lagrangianpredictability, we can expect that assim-ilation of Lagrangian data will be espe-cially fruitful. For Lagrangian data wemean data from floating instruments thatfollow the current with good approxima-tion, either at the ocean surface (drifters)or in the ocean interior (SOFAR, RAFOSand Argo floats) communicating their po-sition via satellite or acoustically. In thelast few years, new methods for Lagrangiandata assimilation have been proposed inthe literature and tested using simplifiedmodels [7, 8, 9]. Some of these meth-ods have been recently applied to in situdata and the results appear very promis-ing in terms of flow correction and increas-ing transport prediction skills. In this paperwe provide a brief summary of results thathave been obtained in the last few yearsat CNR-ISMAR in collaborations with anumber of national and international labo-ratories aimed at increasing the predictabil-ity of particles in ocean flows. We focuson two main issues. In Section 2, we re-view the development and implementationof methods from dynamical system theoryfocusing especially on the FSLE tool [10]to highlight flow features and barriers. Weprovide some examples of applications inthe Adriatic and Ligurian Sea, testing theresult using independent Lagrangian data.The presented results are among the veryfirst examples of applications of the the-ory to real ocean flows. In Section 3, weprovide a summary of work aimed at im-proving flow prediction using Lagrangiandata assimilation. The development of amethod based on a variational approach isbriefly reviewed and examples in coastalflows are shown, using different types ofLagrangian data from Argo floats moving

at 350 m to drifters at the surface. These re-sults are the first successful applications ofLagrangian data assimilation using in-situdata, and the method is now transitionedtoward operational systems. The potentialof these findings for practical applicationsand the strategies for further developmentsare discussed in Section 4.

2 Computing transportbarriers using FSLEs

The Finite Size Lyapunov Exponents(FLSEs) are a diagnostic tool that can beused to identify the main transport barriersand flow structures such as eddies, jestsand boundary currents. They are based onthe computation of maps of relative dis-persion in the flow field, and are relativelysimple to implement. In order to com-pute FSLEs the velocity field u has to beknown, either from high resolution mea-surements (HF radar) or from models. Thecomputations of FSLEs is performed seed-ing particles in small clusters (typically ofthree particles each) throughout the flowdomain and numerically advecting themforward and backward. Formally FSLEsare defined as the time that takes for parti-cles initially separated of a given distanced0 to reach a distance d1=a d0 where a isa specified factor. Forward advection high-lights regions of high dispersion, whilebackward advction identify convergenceregions. An example of computation ofFSLEs using results from an NCOM NRLmodel in the Adriatic Sea [10] is shownin Figure 1 (left panel). The red (blue)lines indicate concentration (dispersion)lines. The superposition of lines indicateridges, i.e. areas that act as transportbarriers between different flow regions and

1259

Oceanography

Figure 1: (left) Forecasted surface velocity from NCOM model during DART06 experi-ment. Superimposed are the 2-day model based FSLE field and the location of a hyper-bolic point (green circle). (right) 2 day trajectories for real drifters (green and purple)and numerical drifters (black lines). Adapted from [10].

that cannot be crossed by particle trajec-tories. Hyperbolic points are indicated bythe crossing of blue and red lines, as in-dicated by the circle in Figure 1 off theGargano Cape. These points are cen-tral to understand Lagrangian pathways,since they separate different structures andare characterized by directions in whichstretching can cause particles to divergefrom the structures (unstable manifolds) aswell as to converge (stable manifolds). Par-ticles located close to a hyperbolic pointcan easily separate, following the differ-ent manifolds. FSLEs cmputations havebeen performed and tested during two re-cent field experiments in collaboration withNURC-NATO, NRL, OGS, University ofMiami and University of Toulone. The twoexperiments took place in the Adriatic Sea

(DART06, [10]) and in the Ligurian Sea(MREA07-POET, [11, 12] respectively.During DART06, FSLEs have been com-puted using the NCOM circulation modelwith 1 km resolution, and FSLE maps (Fig-ure 1, left panel) were used in real time toguide drifter launches from ship. The goalwas to identify regions of high hyperbol-icity so that the launched drifters wouldtend to quickly separate, inducing a max-imum coverage of the area. The presenceof an hyperbolic point in the area off theGargano Cape have been suggested in pre-vious works through the analysis of his-torical drifter data [13], but the hyperbolicpoint is known to be present only at certaintimes, and to depend on the flow structure.For this reason, model results are needed topinpoint the exact time and location of the

1260


Figure 2: Top panels show the trajectories of two drifter clusters launched from the samelocation two days apart in the Gulf of La Spezia during the POET experiment (June2007). Bottom panels show FSLE maps computed from VHF radar at the time of thelaunches. The superimposed stars indicate the position of the drifters.

point. During DART06 three launches ofdrifter pairs have been performed guidedby model forecasts, and two over threeshow the presence of an hyperbolic pointthat induces drifter trajectories to quicklyseparate and diverge. An example is shownin Figure 1, right panel, where the observeddrifter trajectories (green and purple lines)appear to separate quickly, one going to thenorth and the other to the south, in agree-ment with the model results, as shown bythe numerical trajectories in black. Duringthe third launch, instead, the drifters didnot separate and moved together towardthe north. This launch actually acted asan inadvertent control experiment in thesense that the circulation model was indeed

predicting at the time that the presence ofthe hyperbolic point was cancelled by astrong wind episode. The ship, though, dueto logistic reasons performed the drifterlaunches in any case, and the observed andnumerical trajectories did not show sepa-ration. This clearly indicates that a) thehyperbolic point is not present at all timesand b) the model forecast is able to cor-rectly capture its time dependence. Thesecond experiment took place in the Lig-urian Sea and had two components: a largescale component with drifter launches inopen ocean [12], and a more coastal com-ponent in the Gulf of La Spezia with sig-nificantly smaller scales of the order of5-7 km (POET experiment, [11, 14]. Dur-

1261

Oceanography

ing POET, clusters of five drifters werelaunched in the Gulf. Results from twolaunches performed two days a part fromthe same initial conditions are presentedin Figure 3 (upper panels), showing a dra-matically different trajectory behaviour.During the first launch (left) the driftersmove coherently in a cyclonic way exit-ing the Gulf after 12-15 hours. During thesecond launch, instead, the drifters quicklyseparate and end up sampling the wholeGulf, exiting after more that 20 hours. Dur-ing POET, a VHF coastal radar was oper-ated in the area providing maps of velocityfields at resolution of 250 m every 30 min-utes. FSLE maps were computed from theradar velocity and used to understand andquantify the different type of dynamics act-ing during the two launches. Snapshots ofFLSLEs during the two launches are shownin Figure 3 (lower panels). During the firstlaunch, a clear ridge is depicted that sep-arate the area of the Gulf in two differentregions. The drifters move along the evolv-ing ridge and do not cross it as they flowthrough the Gulf. This can be partiallyseen by comparing the drifter trajectoriesand the FSLE snapshot in Figure 3 (leftpanels) but it is much more clear consider-ing the animation depicting drifter motionsuperimposed to the evolving FSLE maps(http://www.rsmas.miami.edu/personal/ahaza/radar/LaSpezia fsle clusters.gif).During the second launch (right panel),no clear ridges separating the Gulf are de-tected and the structures are less marked,even though the presence of an hyperbolicpoint very close to the launching region ofthe drifters can be detected, indicated bythe crossing of blue and red lines. Thisexplains the initial separation of the clusterwith drifters moving in different direction.The animation of drifter/FSLEs evolutionshows that the drifters indeed follow the

manifold lines stemming from the hyper-bolic point. The results show that evenat small coastal scales, where the dynam-ics are complex and driven partially bythe large scale boundary current intrudingin the Gulf and partially by local forcing,Lagrangian transport can be interpreted interms of barriers between dominant struc-tures well captured by FSLEs.

3 Improving transportprediction using assim-ilation

The results in Section 2 provide positive in-dications on the feasibility of forecastingthe main transport properties, since theysuggest that particle motion is mostly dom-inated by barriers between the main coher-ent structures, rather than by smaller scaleflow feautures. As a consequence, whenthe main coherent structures are well rep-resented and forecasted by the models, wecan expect that also particle transport iswell represented at least in terms of gen-eral behaviour, even though the details ofsingle trajectories might be missing. Onthe other hand, the nature of these coherentstructures is still only partially understoodand in many cases circulation models areonly partially able to capture them. A com-mon problem with models, for instance, isrelated to the propagation velocity of thestructures, so that there might be phaseshift errors involving the exact location ofthe structures at a given time. A very ef-fective avenue to improve model perfor-mance is to use real time data to correctmodel results using methods of data assim-ilation. In particular, in our case, since weare interested in transport prediction, wecan expect that Lagrangian data from float-

1262


Figure 3: Left (right) panel shows an example of OPA model results in the Balearic Seawithout (with) assimilation of Argo float trajectories. Arrows indicate vector velocities,color the salinity field and the superimposed brown-orange lines indicate the observed10 day drift of the assimilated float. Adapted from [15].

ing instruments that directly sample cur-rent advection will be especially useful. Anew method to assimilate Lagrangian datahave been developed by CNR-ISMAR incollaboration with the University of Mi-ami and Toulone. The method is basedon correcting the velocity field at the levelwhere the instruments are transported bythe currents (i.e. in the interior ocean forArgo floats and at the surface for drifters)by requiring minimization of the distancebetween observed positions and positionsof numerical trajectories launched in themodel [7, 8]. Once the velocity field iscorrected, the other variables of the model,i.e. the mass variables T,S and the sea sur-face height (SSH), are adjusted using somesimplified dynamical requirements such asgeostrophy and mass conservation (Ozgok-men et al., 2003). The method has beenimplemented using a variational approachand it has been first applied to Argo floats

[15] in the Mediterranen Sea as part ofthe MFS (Mediterranean Forecasting Sys-tem) project. Mediterranean Argo floats(MedArgo) are programmed to drift at aparking depth of 350 m, resurfacing at ap-proximately 5 day intervals, and provid-ing information on their position and onTS profiles. Lagrangian assimilation usesthe position information to correct the driftat 350m. An example of results obtainedassimilating MedArgo floats in the regionclose to the Balearic Islands is shown inFigure 4. Results without assimilation (leftpanel) can be compared with results withassimilation (right panel). The superim-posed orange-brown lines indicate the ob-served drift of one float during 10 days,the arrows indicate velocity vectors and thecolor indicates the salinity field S. As it canbe seen, the assimilation of the Argo floatdata induces a jet along the eastern coastof the island that was not present without

1263

Oceanography

Figure 4: Left (central) panels show an example of ROMS model results in the AdriaticSea without (with) velocity correction from surface drifters. Top panels depict the ve-locity field, with superimposed the 2 day trajectories (red lines) of the drifters used inthe correction, while the bottom panels depict numerical trajectories launched along asection. Right panel shows a Modis satellite image taken at the same time as the modelresults. Adapted from [16].

assimilation, in keeping with the observedfloat drift. Notice also that there are dif-ferences in the S fields between the twopanels, due to the dynamical adjustmentperformed during the assimilation. TheLagrangian assimilation of MedArgo hasbeen recently performed in the frameworkof a multivariate system, i.e. as part of theMFS observing system including T,S pro-files from MedArgo and XBTs and satel-lite SSH and SST [17]. Results are verypositive and the Lagrangian MedArgo as-similation is now in the process of beingtransitioned to the operative MFS system.Further investigations are presently carriedout on the assimilation of surface drifters.Assimilation of surface drifters is expected

to be more challenging than for Argo floatsmostly because they sample the very sur-face of the ocean (from 15 m to 1-2 m),that is characterized by small scales fluctu-ations and dynamics that significantly de-viate from geostrophy. This poses twosignificant question. The first one is re-lated to which scales should be filtered andwhich ones retained in the model correc-tion, while the second one is related to thecorrection of the mass variables, that has tobe performed differently than in the case ofArgo floats. A simple geostrophic balancein fact cannot be used since the upper me-ters are strongly influenced by Ekmn dy-namics, so that a more complex dynamicaldecomposition has to be adopted. So far,

1264


we have been working on the first step ofassimilation, i.e. the velocity correction atthe surface using drifter data, and we havenot attacked the problem of mass correc-tion yet. Results on surface correction arevery promising [16], as shown in the ex-ample in Figure ?? for the Adriatic Sea.The left (central) panels show results fromthe ROMS model without (with) correc-tion, for a snapshot of velocity (top panels)and for numerical trajectories (bottom pan-els) launched along a section. The smallred lines in the top panels indicate two daytrajectories of four drifters used for the cor-rection. As it can be seen, the velocity cor-rection appears small, but it has a signifi-cant impact on trajectories. The trajectoriesof the non corrected model in fact appearretained inside the boundary current, whilethey tend to exit from it in the case of cor-rection, more in keep with what suggestedby MODIS satellite data (right panel) indi-cating significant intrusion from the bound-ary current in the interior. Of course thisis not a quantitative test of results yet, andwork is in progress to quantify the im-provement using independent data

4 Summary and discus-sion

In this paper, we have discussed meth-ods to improve the prediction of particlestransported by ocean currents. Results arevery encouraging and they show that, eventhough the problem is extremely challeng-ing, significant improvements can be ob-tained using appropriate techniques. Onthe other hand, a number of questions arestill open as discussed in the following.The results in Section 2 strongly suggestthat the motion of particles is controlled by

barriers between the main coherent struc-tures in the flow, such as mesoscale ed-dies, jets and boundary currents. The sizeof these structures depends on the flow en-vironment and in particular on the Rossbyradius of deformation, ranging from morethan 10 km in the open sea in the Adri-atic and Ligurian sea, to few km in smallcoastal gulfs such as the Gulf of La Spezia.Flow features smaller than these mesoscalestructures do not appear to directly influ-ence the main characteristics of particletransport, even though they can influencethe details of single trajectories. This re-sult, if confirmed in other regions of theworld ocean and shown to be general, is ex-pected to be extremely important for whatconcerns practical applications. The resultin fact implies that the resolution of circu-lation models can be limited to correctlyreproduce mesoscale structures, while cap-turing submesoscale or smaller processes isnot crucial for the problem of Lagrangiantransport, that is central to many practi-cal applications of operational predictionsystems. Looking at the existent litera-ture, results in other parts of the worldshow similar and compatible results, forinstance the studies of relative dispersionin the Gulf of Mexico and in the Norwe-gian Sea [18]. On the other hand, other re-sults in the California Current seem to sug-gest that submesoscale and smaller scalesmight be relevant for flow advection prop-erties [19]. This might be related to thefact that the California Current is charac-terized by supwelling and significant ver-tical motion, that is often dominated bysubmesoscale structures. Overall, the cen-tral issue of the role of submesoscale andsmaller features is still open and it requiressignificant further investigations. Differ-ent regions of the ocean might have to betreated differently [20], and it is crucial to

1265

Oceanography

understand what are the physical reasonsfor these differences and the consequencesfor the transport of biogeochemical prop-erties and their modeling and prediction.For what concerns assimilation methods,the results in Section 3 show that they canbe extremely useful to correct model fore-casts, for instance repositioning and shap-ing coherent structures that are not cor-rectly reproduced by the models. Assim-ilation has been successfully implementedin the case of Argo subsurface floats, and itis now in the process of being transitionedto operational systems. Work is in progressfor surface drifters and the main questionsto be addressed are conceptually related tothe ones discussed above. We have to de-cide whether or not the signature of smallscale processes present in the data have tobe maintained and used in the assimilationor filtered away, and which type of dynam-ics have to be used, beyond geostrophy. Inorder to do that, an increased knowledge

of air sea interaction processes is neces-sary, as well as an improved understand-ing of the role played by vertical motionin the mixed layer. Finally, it should bepointed out that while Lagrangian data arecertainly a natural choice to improve trans-port prediction, other types of data can alsobe used, and fusion between models andvarious data is expected to be very impor-tant in the future. As an example, work hasalready started to use satellite data (SARand visible) to improve transport predictionin case of accidents at sea such as oils spillevents [21].

5 AcknowledgementsThe authors wish to acknowledge collab-orations with G. Gasparini, M. Rixen, A.Poje ,L. Piterbarg, N. Pinardi and S. Do-bricic. The work was supported by theEU projects MFSTEP and ECOOP and byONR (Ofice of Naval Research).

References[1] L. Piterbarg, T.M. Ozgokmen, A. Griffa, and A.J. Mariano. Predictability of La-

grangian motion in the upper ocean. 2007.

[2] S. Aliani and A. Molcard. Hitch-hiking on floating marine debris: macrobenthicspecies in the Western Mediterranean Sea. Hydrobiologia, 503:59 67, 2003.

[3] M. Veneziani, A. Griffa, A. Reynolds, Z. Garraffo, and E. Chassignet. Parame-terization of Lagrangian spin statistics and particle dispersion in the presence ofcoherent vortices. J. Mar. Res, 63:10571084, 2005.

[4] A. Doglioli, M. Veneziani, B. Blanke, S. Speich, and A.Griffa. Lagrangian analysisof Indian-Atlantic interocean exchange in a regional model. Geophys. Res. Lett,33:L14611, 2006.

[5] S.C. Shadden, F. Lekien, and J. E. Marsden. Definition and properties of Lagrangiancoherent structures from finite-time Lapunov exponents in two-dimensional aperi-odic flows. Physica D, 212:352380, 2005.

1266


[6] V. Artale, G. Boffetta, A. Celani, M. Cencini, and A. Vulpiani. Dispersion ofpassive tracers in closed basins: Beyond the diffusion coefficient. , Phys. Fluids,9:31623171, 1997.

[7] A. Molcard, L. Piterbarg, A. Griffa, T.M. Ozgokmen, and A.J. Mariano. Assimi-lation of drifter positions for the reconstruction of the Eulerian circulation field. J.Gephy. Res, 107:31543171, 2003.

[8] V. Taillandier, A Griffa, and A. Molcard. A variational approach for the recon-struction of regional scale Eulerian velocity fields from Lagrangian data. OceanModelling, 13:124, 2006.

[9] L. Kuznetsov, K. Ide, and C. K. R. T. Jones. A method for assimilation of La-grangian data. Mon. Weather Rev, 131:22472260, 2003.

[10] A. Haza, A. Griffa, P. Martin, A. Molcard, T.M. Ozgokmen, A.C. Poje, R. Barbanti,J. Book, P.M. Poulain, M. Rixen, and P. Zanasca. Model-based directed drifterlaunches in the Adriatic Sea: Results from the DART experiment. Geophys. Res.Letters, 34:L10605, 2007.

[11] A. Molcard, P.M. Poulain, P. Forget, A. Griffa, Y. Barbin, J. Gaggelli, J.C. DeMaistre, and M. Rixen. Comparison between VHF radar observations and datafrom drifter clusters in the Gulf of La Spezia (Mediterranean Sea). J. Mar. Sys.,78:S79S89, 2009.

[12] K. Schroeder, A. Griffa, P.M. Poulain, A. Haza, and T.M. Ozgokmen. Relativedispersion in the Ligurian Sea. 2010.

[13] M. Veneziani, A. Griffa, and P.M. Poulain. Historical drifter dataand statisticalprediction of particle motion: a case study in the Adriatic Sea. J. Atmos. OceanTech, 24:235254, 2007.

[14] A. Haza, T.M. Ozgokmen, A. Griffa, A. Molcard, P.M. Poulain, and G. Peggion.Transport properties in small scale coastal flows: relative dispersion from VHFradar measurements in the Gulf of La Spezia. 2010.

[15] V. Taillandier, A. Griffa, P.M. Poulain, and K. Beranger. Assimilation of ARGOfloat positions in the North Western Mediterranean Sea and impact on ocean circu-lation simulations. Geophys. Res. Lett, 33:L11604, 2006.

[16] V. Taillandier, A. Griffa, P.M. Poulain, R. Signell, J. Chiggiato, and S. Carniel. Vari-ational analysis of drifter positions and model outputs for the reconstruction of sur-face currents in the Central Adriatic during fall 2002. J. Geophys. Res, 113:C04004,2008.

[17] V. Taillandier, S. Dobricic, P. Testor, N. Pinardi, A. Griffa, L. Mortier, and G.P.Gasparini. Integration of ARGO trajectories in the Mediterranean Forecasting Sys-tem and impact on the regional analysis of the Western Mediterranean circulation.J. Geophys. Res, 2010.

1267

Oceanography

[18] J. LaCasce and C. Ohlmann. Relative dispersion at the surface of the Gulf of Mex-ico. J. Mar. Res, 61:285312, 2003.

[19] X. Capet, J. McWilliams, M. J. Molemaker, and A. Shchepetkin. Mesoscale to Sub-mesoscale Transition in the California Current System. Part II: Frontal Processes.J. Phys.Ocean, 38:44, 2008.

[20] A. Griffa, R. Lumpkin, and M. Veneziani. Cyclonic and anticyclonic motion in theupper ocean. Geophys. Res. Lett, (35):L01608, 2008.

[21] A. Mercatini, A. Griffa, L. Piterbarg, E. Zambianchi, and M. Magaldi. Estimatingsurface velocities from satellite data and numerical models: implementation andtesting of a new simple method. Ocean Modelling, 2010.

1268

Monitoring and Scientific Analysis of the Sea Levelin the Venice Area

A. Tomasin1,21, Department of Applied Mathematics, University Ca Foscari, Venezia, Italy2, Institute of Marine Sciences, CNR, Venezia, [email protected]

Abstract

At Venice and in the nearby coastal area the sea level is always observed withobvious care. The scientific effort is documented here for analysis and forecastingof the the tide. The different factors are considered, from the astronomical forcingwhose effects can be altered by human modifications of the environment, to the stormsurges and the long-term evolution of mean sea level. Surges are the day-to-dayalarming nightmare in the autumn-winter season, since they quickly add up to thenormal tide and flood large parts of the towns of the lagoon. Also the mean sea-level change is of concern, especially in these decades when a strong rise is possibiledue to climatic change, in addition to the old problem of sinking of the city and itssurroundings.

1 Introduction

The sea level of the northern Adriatic, andparticularly of the Venice lagoon, stimu-lates particular interest. In fact, unlikein the largest part of the Mediterranean,here the tidal oscillations are large, and theanomalies due to atmospheric forcing aresevere. This creates the concern for the sec-ular acqua alta (high water) phenomenonin a city, like Venice, that the world ad-mires for its historical treasures. The Adri-atic is like a long bay, 800 km long and 100wide approximately, with a narrow open-ing into the central Mediterranean it is shal-low in its northern part, and all this givesit a proper period of oscillation (22 hours,and, secondary, 11 hours) very close to thetidal forcing (24 and 12 hours), as it willbe explained below. What matters here isthat the ordinary (astronomical) tide has arange of the order of one meter, three times

more than observed in most areas close toit (British sailors in the past used to re-port that in the Mediterranean there is notide. . . ).In addition, the morphology of the Adri-atic favours the amplification of the windeffect. Sirocco wind (south-easterly) pilesup water towards the dead end, and theflood is easy. An obvious addition comesfrom climate change with a real increaseof mean sea level that sums, not to be for-gotten, to the subsidence of the plain con-tiguous to the Lagoon, and finally to thepeculiar lowering of the Venice area likelydue, in the recent past, to human activity.The three factors listed above (astronomi-cal tide, surges and general sea-level rise)will be analyzed with respect to the scien-tific activity. Other effects, like the onesdue to precipitations or tsunamis, will notbe considered here.

Oceanography

Figure 1: Spring-tide amplitude (M2+S2) at Trieste and Venice: (a) anomalous increaseat Venice in 1965-1970, (b) variations in recent years (updated to 2009).

2 The ordinary tide

The ordinary tide, generated by the Sun-Moon-Earth dynamics, is successfully de-scribed by the expansion in harmonic com-ponents [1]: it permits the analysis of thedata collected by level recordings in thevarious sites. Needless to say, one cancompute, without experimental data, theideal, gravitational level of tide in what-ever point (equilibrium tide): in real life,water tends to such equilibrium, with adelay and reduction/amplification, whencethe need for measurements models will tryto explain the reasons. At Venice, the her-itage of tide-gauge measurements is due tothe Ufficio Idrografico (now ISPRA), theICPSM of the City of Venice and the lo-cal CNR branch. The tool used here (good also for other sites) to estimate theconstants of the harmonic components isa software developed at CNR, named Po-lifemo [2]: it works with the least-squaremethod which is hole tolerant (meaningthat the possible lack of some data canbe tolerated), does not require observationsat regular times and can be used with se-

quences of extremes (high and low tides)instead of the usual (and more precise)hourly data. A convenient feature of thisalgorithm is its matching the long-term as-tronomical variations (consciously left outby the harmonic development), thus avoid-ing further corrections.Given the constants for a place, predictionsare possible for the astronomical tide, andin Venice they are yearly published thanksto cooperation of the local office for this ac-tivity (ICPSM), the branch of a national in-stitute for protection (ISPRA) and the CNRresearchers (ISMAR), [3].A more precise description of the scien-tific effort is required, since the harmonicexpansion mentioned above is not com-puted once forever: many sites (notably:the city of Venice) are not immutable, inthe sense that the morphology of their sur-roundings is not fixed. Reference is madeto the lagoon (that can be dredged), and inparticular to the three inlets by which theAdriatic communicates with the lagoon: bymonitoring the harmonic constants year byyear, the effect of environmental changes(mostly of human origin) is detected. A pe-

1270


0

2

4

6

8

1940 1950 1960 1970 1980 1990 2000 2010

TS / VESharp hourly level rise at Trieste and Venice (ratio of the

Figure 2: The lagoon inlets can mitigate more or less the violence of the external surges.The cases of sharp increase of water level at Trieste and at Venice (at least .2 m in anhour) are considered. The ratio of the yearly numbers in the two cities is shown (updatedto 2009, three-year moving average). The lower is the ratio, the easier is the entrance ofquick surges.

riod of heavy interventions on the lagoon(1965-1970) was clearly revealed by suchsurvey by CNR [4]. Moreover, an impor-tant tool of comparison can be introduced,i.e., the parallel tidal recordings of Trieste,collected by the local CNR-ISMAR office.This city is ideal for comparison, beingat about hundred kilometres from Venice,situated at the closed end of the Adriaticlike the lagoon, but substantially not sub-ject to morphological changes of the coast.The coupled analysis of the harmonic con-stants shows in the period mentioned above(1965-1970), but also in the most recentyears, particular trends in the astronomicaltide at Venice, revealing some kind of hu-man intervention (Figure 1).There is another intriguing remark aboutthe effect of morphology on the water levelobserved inside the lagoon. Tides have fun-damental periods of one day or half a day.What happens with phenomena of shorterperiods, that can occur with surges? Thiswill be considered in what follows.

3 Level anomalies due tometeorology

The most severe troubles for Venice origi-nate from atmospheric forcing. A little dif-ficulty arises from the definition of such ef-fect that is totally called surge: it is thedeviation from the expected ordinary tide,but one has to deal with long-term vari-ability of sea level, even along the seasons(also due to circulation). Here the defi-nition is sharp: from the observed valueone subtracts the astronomical oscillationsassumed at zero mean (i.e., one subtractsonly a few sinusoids) and subtracts theyearly mean-level value. Clearly, this waythe surge is something independent of thetide gauge zero and the yearly variationsthat could also include the sinking of thearea. But the definition could be differ-ent. Scientists would also be careful aboutthis definition, since considering an addi-tion denies possible interactions betweentide and surge (non-linearitiy). But indirectproofs allow such an assumption to a good

1271

Oceanography

100

120

140

160

180

200

1940 1950 1960 1970 1980 1990 2000 2010

cm

Figure 3: The cases of flood of Venice (i.e. when the water level exceeds 1.10 m over afixed reference in the city) updated to 2009.

level, for example by observing the ab-sence of tide interactions with itself (over-tide or compound tides).Since the astronomical tide cannot cause aflood (at least, in the present conditions ofmean sea level), only surges can give it.But this can occur both with a high tideand even a moderate surge, and vice versa:the nightmarish case of 1966 occurred witha negative tide, and this reminds us thatthere is no upper limit to surges. It is wellknown that the forcing on the Adriatic isdone by SE winds (sirocco, blowing alongthe axis of the sea towards the closed end)and atmospheric depressions passing overthe Adriatic. A catalogue over the last half-century involved also the local CNR branch[5].As anticipated above, the morphology ofthe lagoon has a strong influence on every-thing coming from the open sea: the sim-plest check one can perform is the countof cases in which the water level at Venicesharply increases, let us consider 0.2 m inan hour. This number obviously changesfor yearly variability, but again the com-

parison with Trieste can help. The ratioof numbers of such events between Tri-este and Venice gives an inverse idea of thepermeability of the lagoon inlets (Figure2). Again something interesting turns out -to be investigated using models [6] - show-ing periods in which the lagoon is more orless protected against surges. Indeed, it isclearly demonstrated that quick storms,originating spikes in the Adriatic level,could be cut by the low efficiency of theinlets [5].Another question about surges concernstheir frequency: the floods, hence thestorms, are more or less frequent in dif-ferent years, and the suggestion of regu-lar fluctuations appears from simple graphs(Figure 3). From the above distinctions,it seems more interesting to consider thesurges, and in fact a simple correlationemerges between yearly number of surges(over a certain threshold) and the solar ac-tivity, as measured by the Wolfs sunspotnumber [7]: it is presumably a cue for fur-ther investigations (Figure 4).

1272


-2

-1

0

1

2

3

1940 1950 1960 1970 1980 1990 2000 2010

Standardized number

surges

sunspots

c=.67

Figure 4: Frequency of surges of meteorological origin and frequency of sunspots(Wolfs number); surges not less than .50 m high. The samples are standardized, up-dated to 2009, and a 3-year moving average is applied. Correlation index is .67.

4 Free oscillations of theAdriatic

Related to surges, a peculiarity of the Adri-atic dynamics is relevant, i.e., seiches.They are the free oscillations of the Adri-atic that follow the pulse of a storm. Theypersist for many days, and they are inter-esting for many reasons. In the predictionof floods they are very important [8], and itcan occur that a storm surge comes out ofphase with respect to tide, but the next re-turn of the seiche is instead in phase and aflood occurs when the storm was forgotten(Figure 5).Another feature about seiches is the factthat they reveal the proper period of os-cillation (eigenperiod, for scientists) of theAdriatic. The morphology of the sea de-termines the main oscillation, with a pe-riod between 21 and 22 hours, and a sec-ondary one close to 11 hours [9]. Thisshows why the tide is strong in this sea,much more than in the rest of the Mediter-ranean: the well-known astronomical peri-odicities (about 24 and 12 hours) are closeto resonance in this basin. The semi closed

shape of the sea and the mechanism of on-set by wind are the cause of all that [8].

5 Models and facts aboutthe mean sea level

The sea depth is one of the morphologicalelements that determines the above period-icities: indeed, a sea-level rise will possi-bly be a consequence of climate change:what about seiches and tides in this case?An analysis was performed, in a coopera-tion including ISMAR-CNR [3] for this hy-pothesis: beyond details, we briefly remindhere that if only the depth is modified (byan increase) the proper periods of the seawould go more distant of the astronomicalperiods, and a tidal reduction would occur.The mean sea-level remains in the back-ground of the above discussions, and it isextremely relevant. As mentioned above, itwill possibly rise, with a menace to coastalsettlements, but so far the short-term con-siderations (the last five- or ten-year trends)turn out deceiving. As one can see in theplot (Figure 6), also the mean sea levelshows possible oscillation [10]. Another

1273

Oceanography

-100

-50

0

50

100

150

1 11 21 31 41 51 61 71 81 91 101 111

cm

hours

starting Feb. 11, 1972

astronomical tide

total observed level

surge

Figure 5: Showing the seiche, or free oscillation of the sea after a storm: the meteorolog-ical anomaly continues for a few days, and getting in phase with the tide gives origin toan unexpected flood.

15

17

19

21

23

25

27

29

31

33

35

1965 1970 1975 1980 1985 1990 1995 2000 2005 2010

yearly mean sea level at Venice

interpolating function

cm

Figure 6: The yearly mean sea level at Venice, updated to 2009: two oscillations (period:7.8 and 16.0 years) seem to act, and short-term linear trends would be deceiving.

important chapter would treat the sinkingof the Venice area [11]: in fact, one consid-ers, for the city, the relative mean sea level,combining the two effects of subsidenceand real sea-level rise. But this is studiedin other sections of the local research.

6 AcknowledgementsThe various institutions, mentioned inthe present work, cooperating with CNR-ISMAR, are warmly acknowledged, first ofall for their activity in favor of Venice, thelagoon, and people of the northern coast ofthe Adriatic.

1274


References[1] A. T. Doodson. The harmonic development of the tide-generating potential. Proc.

Roy. Soc., A, 100:305329, 1922.

[2] A. Tomasin. The software Polifemo for tidal analysis, Tech. Note 202. 2005a.

[3] P. Canestrelli, M. Ferla, and F. Trincardi. Previsioni delle altezze di marea per ilbacino San Marco e delle velocita` di corrente per il Canal Porto di Lido, Laguna diVenezia, valori astronomici. 2010.

[4] A. Tomasin. Recent changes in the tidal regime in Venice. Rivista Italiana diGeofisica, 23:275278, 1974.

[5] P. Canestrelli, M. Mandich, P. A. Pirazzoli, and A. Tomasin. Wind, depressions andseiches: tidal perturbations in Venice (1950-2000). 2001.

[6] L. Zampato, G. Umgiesser, and S. Zecchetto. Storm surge in the Adriatic Sea: ob-servational and numerical diagnosis of an extreme event. Advances in Geosciences,7:371378, 2006.

[7] A. Tomasin. Forecasting the water level in Venice: physical background and per-spectives. In: Flooding and Environmental Challenges for Venice and its Lagoon:State of Knowledge. pages 7178, 2005b.

[8] A. Tomasin, G. Umgiesser, and L. Zampato. (c) On the dynamics of the Adriaticseiche, in: Scientific research and safeguarding of Venice. pages 6574, 2005.

[9] A. R. Robinson, A. Tomasin, and A. Artegiani. Flooding of Venice: phenomenol-ogy and prediction of the Adriatic storm surge. Quart. J. Roy. Met. Soc, 99:688692,1973.

[10] P. A. Pirazzoli and A. Tomasin. Sea level and surges in the Adriatic Sea area: recenttrends and possible near-future scenarios. 166:6183, 2008.

[11] L. Carbognin, P. Teatini, and L. Tosi. Relative land subsidence in the lagoon ofVenice, Italy, at the beginning of the new millennium. J. Mar. Sys, 51:345353,2004.

1275

Oceanography

1276

The Sea Between Image and Imagination-the In-vestigation of the Underwater World from the Re-naissance to the Age of Enlightenment

A. CeregatoInstitute of Marine Sciences, CNR, Bologna, [email protected]

Abstract

The roots of modern Marine sciences as commonly meant stemmed from thework of Luigi Ferdinando Marsili (Bologna, 1658 - 1730), an eclectic military archi-tect who produced the first scientific descriptions of the seabed and its inhabitants.Coeval of the natural philosopher Vallisneri and introduced by Newton to the RoyalSociety of London, Marsili represents an ideal link between the observation of Na-ture according to the method developed in the sixteenth century by his co-citizenUlisse Aldrovandi and the modern oceanographic disciplines. Curiously but perhapsnot accidentally, the Institute of Marine Geology of CNR, now incorporated into theISMAR, was founded by Raimondo Selli in Bologna, Marsilis birthplace and Sellihimself named after his ancient precursor one of the most impressive submarine vol-canoes of the Mediterranean.

1 Introduction

The modern approach to the history of sci-ence (and to history in general) tends toinvestigate rather the context than findingthe first, the precursor or the fatherof a discipline or of a theory. Nonetheless,individual people undoubtedly boosted theevolution of science, or they made a dis-covery before anybody else, with very di-verse outcomes according to many vari-ables.It is commonly accepted that the devel-opment of the marine sciences (the termoceanography appeared much later, af-ter the 1872-1876 Challenger expeditions),took place after the work of Luigi Fer-dinando Marsili (1658-1730) (Figure 1),in particular from two of his publishedworks, Osservazioni intorno al Bosforo

Tracio [1] written during his earlier career,and the programmatic Histoire Physiquede la Mer [2] published in the Nether-lands just five years before Marsilis death,at the end of a life entirely spent on thefield, through all over the Mediterraneanfrom the Bosporus to Gibraltar. The firstwork cited above contains analytical re-sults of the investigations on some physi-cal features of the sea performed by Mar-sili during his first diplomatic mission fromVenice to the Ottoman Empire and his re-turn trip to Venice, that can be consideredas one of the first oceanographic surveys.The Histoire Physique represents somehowthe completion of this research, presentingnew data on the seawater currents, temper-ature and salinity, but also descriptions ofthe seabottom and of the marine organisms[3, 4, 5].

Oceanography

Figure 1: Portrait of Luigi Ferdinando Marsili. (Courtesy of G.B Vai. University ofBologna)

As a matter of fact, the most original ele-ment of the approach of Luigi FerdinandoMarsili to investigation and to the dissemi-nation of his observations is due to the ap-plication of the scientific method codifiedsome decades earlier by Galilei, Descartesand Bacon, but also to the wide use of theimage as a scientific tool, introduced byUlisse Aldrovandi (Figure 2) in Marsilishometown Bologna, about a century be-fore.

2 Imagine versus imagina-tion

The combined use of image and text as atool for describing the natural objects (manincluded) is well known from the antiq-uity to medieval treatises, but except ofsome medical herbals and of particular bes-

tiaries like the De arte venandi cum avibusby Emperor Frederick II [6], the role ofimages was primarily ornamental or sym-bolic, and their accuracy was poor or evenfantastic. Until the Renaissance, most ofthe written knowledge about nature con-sisted of a transmittance from generation togeneration of a mix of Latin texts, transla-tions from Arabic authors and Arabic ver-sions of Greek literature. These hand-written books were enriched by comments(the glossae) and sometimes by drawingsonly occasionally taken from a real modeleven in the case of depicted herbals or med-ical treatises; more commonly, texts andimages were simply copied from volumeto volume. This way of transmission ofknowledge favoured the creation of stereo-typical, simplified images, whose signifi-cance carried rather symbolic significancesthan real features of the natural objects.

1278


Figure 2: Portrait of Ulisse Aldrovandi (copy from the original by Pelagio Pelagi, early1800) Courtesy of G.B. Vai. (Photo Mattei-Zannoni, BUB, ASUB)

Starting from the XV century, the stronginteraction between art and science and arenewed attention to the physical reality ledto an increased accuracy of the images.The new fame of the Naturalis Historia ofPliny and novelties imported from the voy-ages of exploration explosively expandedthe list of terrestrial and marine organisms,forcing the natural philosophers to reviewthe old classifications and try to build a newinventory of the world.The introduction of printing and of moreaccurate and analytical illustrations, fed anirreversible process of liberation from theold scholasticism, based on a dogmatic in-terpretation of Aristotle (filtered from me-dieval tradition) which conditioned most ofthe knowledge on nature.Leonardo Da Vinci (1452-1519), in searchof proving a geocentric, protoscientific

Theory of Earth based on the Neo-Platonistrelationship between Mans Microcosmand Natures Macrocosm, demonstratedthat fossils are the remains of ancient or-ganisms that the Deluge could not carry tothe highest mountains [7].At the end of the XV century, Leonardoin Italy and Albrecht Durer (1471-1528) inGermany, started two different approachesto the figurative description of natural ob-jects: while Leonardo, drawing natural ob-jects and anatomic studies, always tendedto represent the dynamism, the continuoustransformation of the objects and the rela-tionship between the subject and its con-text in order to stress their evocative power,on the other hand Durer isolated the objectsto concentrate the analysis on the physicalnature of the objects themselves; the besttechniques to achieve the finest detail were

1279

Oceanography

Figure 3: Section of the Earth surface with the hypothetical profile of the seabed. A.Kircher, Mundus Subterraneus (III Ed. 1678), Cap. XV, p. 97. G.B. Vai (2004) Ed.,Critical reprint by A. Forni Editore. (A. Ceregato, personal archive and photo)

the drawing by ink pen, the engraving anda mix of tempera and watercolors [8].The new generation of natural philoso-phers will choose the model proposed byDurer and his pupils as the best for theirdescriptions. The Swiss Conrad Gess-ner (or Gesner;1516-1565) will attemptfor first to build an updated inventory ofthe Three Kingdoms of Nature: Animalia,Plantae and Mineralia. The technique ofengraving on wood and later on coppersheets developed by Durer was immedi-ately adopted for printing the new trea-tises [9]. The Counter-Reformation lim-ited the freedom of Italian natural philoso-phers (Galileo Galilei and Giordano Brunowere only the most famous victims), whowere denied to venture interpretations dan-gerously dissonant with the Scriptures, butdid not prevent to investigate the orderof Nature: Ulisse Aldrovandi (Bologna,1522-1605) set up the first natural historymuseum at home, a microcosm of Na-ture tended to show as more comprehen-

sively as possible the diversity of naturalobjects, animals, plants and minerals andto compose a natural history rich in im-ages as much accurate as possible. Forthis purpose Aldrovandi created an artis-tic laboratory within his home museum andinvited some of the best illustrators andengravers of his time to take their draw-ings directly from his specimens; he alsoacquired and exchanged watercolours andengraved images from his correspondentsand collectors as the Archidukes of Tus-cany and the Duke of Mantua. He per-sonally maintained for years some artists,and looked at Durers legacy when he de-cided to print his books, so he hired Cor-nelius Schwindt and Christopher Lederlein(Cristoforo Coriolano) who produced morethan 3000 woodcuts at the end of 1598 (i.e.[10]; see also [11] for a complete referencereview).A part of four volumes (Ornithologiae. . . I,II, III and De Animalibus Insectis) pub-lished from 1599 to 1603, the remaining

1280


Figure 4: Map of the surface oceanic currents after Athanasius Kircher. Kircher did notperform any direct measurement, differently from what Marsili did some decades later,but it is one earliest attempt of representing the sea currents. From Athanasius Kircher,Mundus Subterraneus (III Ed. 1678). Crit. Ed. by G.B. Vai, 2004. A. Forni ed. (A.Ceregato, personal archive and photo)

books of the Ulisse Aldrovandis HistoriaNaturalis were printed after his death, of-ten remarkably modified by the curatorsand only one of the planned botanical vol-umes was published in 1668. Accordingto the common approach of that age, thetaxonomy was adapted from those of Aris-totle or, as Aldrovandi tended to prefer,from Pliny. Each section included a re-view of the previous knowledge about agroup of natural objects, not only sci-entific descriptions but also symbolic at-tributes and their eventual pharmaceuticaluse. The originality of these works was inthe personal observations Aldrovandi madedirectly on the specimens and the accuracy

of the images. He also was used to listall the current, popular, foreign names re-ferred to each species, both within hisbooks and in his collection of more than2000 colour drawings he used also for hislessons as the first professor of Natural His-tory at the Bologna University.The XVII century is apparently charac-terized by a paradox: on one hand, it isthe age of the Scientific Revolution intro-duced by Galileo Galilei, Rene` Descartesand Francis Bacon, of the experimentalmethodology, of the discovery of the neworder of the Universe, introduced by Galileiand perfected by Newton at the end ofthe century; on the other hand the Natu-

1281

Oceanography

ral History collections appeared since theRenaissance, from which the new ideasoriginated, progressively lost their scien-tific significance in favour of the aestheticattributes, becoming Cabinets of Curiosi-ties (also known as Wunderkammern) [12].From the work of Aldrovandi and Gess-ner, forgotten the scientific descriptions,the Baroque readers prefer the anedoc-tal contents, so the old authors will becited for their descriptions of monsters anddrakes: the Aldrovandis Monstrorum His-toria, published in 1642 by BartolomeoAmbrosini, then one of his most appreci-ated works and largely cited by Athana-sius Kircher in his Mundus Subterraneus(1664), will be ridiculized a century laterby the Enlightened scientists. Nonetheless,some scientists continued to follow the les-son of Aldrovandi, in particular those whowere involved in the debate on the natureof fossils and of the structure of the Earth:Fabio Colonna (1567-1640), Niels Stensen(1638-1686) and Agostino Scilla (1629-1700) largely used the images for clarifytheir observations. Colonna within the DeGlossopetris dissertatio (1616), demon-strated the organic nature of Glossopetraewith a use of images not dissimilar to thatof Aldrovandi, Niels Stensen completedthe demonstration made by Colonna witha figure of Canis Carchariae with a detailof a tooth (following the original Gessnersintuition) and the Prodromus (1669) withsome simplified figures to describe the dif-ferent facies of Earth before and after theDeluge [13, 14]. Agostino Scilla, artistrather than scientist titled his most famouswork on the real origin of fossils (1670)drawing a frontispiece with two: La vanaspeculazione disingannata dal senso (trad.:the vain speculation disillusioned by thesense), and he accompanied his discussionwith accurate figures. Athanasius Kircher

described his theories on the Mundus Sub-terraneus through a number of sectionsof the Earth very imaginative but easy tounderstand, sometimes populated by thedrakes taken from Aldrovandis works. Afew decades later, his hypothesis that themaximum depth of the sea was equal tothe highest mountain (Figure 3) influencedthe early observations by Marsili. In an-other chapter of Mundus Subterraneus heput one of the first attempts of representthe oceanic currents that is not conceptu-ally so dissimilar from our present charts(Figure 4). If the use of images by naturalphilosophers is sometimes oscillating be-tween loyalty to the real and the imaginary(the sea is still populated by mysterioussea monsters), in the same years, the de-velopment of the cartographic dis

oceanography

Documents

ocean acidificationoa

co2 system

adecrease of ph value

ocean acidification

pogenic co2

co2 problem

ocean acidification

makethe world ocean