THE SIGNIFICANCEOF JETTISONING: OILON TROUBLED WATERS
Jettison has a long and significant history. In the Old Testament stow of Jonah, it will be recalled, the sailors threw cargo overboard to lighten the ship. Jonah's exit overboard shortly afterwards did not constitute jettison—as it was to appease the Lord—although it certainly appears to have been efficacious in saving the ship!
Mercantile law followed the practice when in 900 B.C. the laws of the island of Rhodes prescribed the law of jettison whereby, if cargo was thrown overboard to lighten a ship for the general safety, that which was thrown away would be made good by general contribution or, as it later became known, by General Average.
Another Biblical hero, Paul, witnessed jettison firsthand near the island of Crete on his voyage from Caesarea to Rome. You may recall the vessel was making very heavy weather and they began to lighten the ship and, on the third day, they jettisoned the ship's gear with their own hands. Presumably General Average was not declared because, despite the jettison, the ship subsequently went aground and was pounded to pieces.
Rhodian law on jettison was incorporated into the Digest of Justinian (Emperor Justinian—A.D. 500) and later into the laws of Oleron in the twelfth century. Those were attributed to the English king, Richard I, or his mother, Eleanor of Aquitaine. Article 8 of the Sea Laws of Oleron provided for General Average when cargo was jettisoned to save ship and cargo. Three thousand years on from the laws of Rhodes, and even longer from the days of Jonah, the current rules of General Average—the York/Antwerp Rules 1974—provide that the sacrifice of cargo by jettison for the common safety shall be made good in General Average.
Lloyd's Open Form 1990 (LOF 90), the most frequently used salvage contract, (although it is little used in the United States) permits the salvor to jettison cargo within reason to salve the ship, although that is not something which any salvor does lightly in practice.
Jettison therefore has a long history and encapsulates the rule which applies in many walks of life—that at times it is necessary to make a small sacrifice in order to avert a greater loss.
ONE HUNDRED YEARS AGO
Groundings were much more common 100 years ago than they are today. Ships still depended to a degree on sail, propulsion machinery was not always reliable, and navigational aids were primitive. The Salvage Associations' records for 1892 and 1893 show its surveyors attending many groundings. Often, the ship had to be lightened, and it is fair to say that in most instances the cargo was lightened rather
than jettisoned. Except as a last resort, no one wanted to throw away the cargo in those days either, but sometimes they had to. In one case, coal was jettisoned; in another, iron products; and in a third, grain.
Oil was not transported in great quantities at that time but in 1893 the Salvage Association did attend a difficult casualty in the Dardanelles on a British ship carrying cased petroleum to Calcutta from Batoum in what was then, and is once again, Georgia. The vessel's position was described as "dangerous." The weather was blowing very hard and the barrels of petroleum were jettisoned in large quantities. Alas, on this occasion, it was largely in vain because the ship eventually became a total loss. However, some of the cargo was saved so jettison did avert some cargo loss and pollution, although it is interesting to note that there was not one mention of the word "pollution"—perhaps it was a term that had not yet been used in the context of oil and the oceans.
THE TWENTIETH CENTURY
The Salvage Association's records and discussions with our surveyors indicate that jettison has continued to play a part in salvage, but it seems the cargo jettisoned has usually been dry bulk, particularly ore, coal, rock, and stone. One surveyor jettisoned, as a last resort, a small quantity of bunkers from a large fishing vessel, which did enable it to refloat. Another related jettisoning 2,500 tons of frozen meat in the southern Pacific. This was apparently seen as a bonus by the local islanders and the aquatic population. Deck cargoes of logs and timber have quite frequently been jettisoned but usually for stability reasons, rather than to refloat.
Though probably not categorized as jettison as such, there have been many cases where damaged tanks have been "blown" or "pressed" out to enable refloating, or where engine rooms have been dewatered, sometimes resulting in an escape of oil. In the latter case, pumping out would hopefully be stopped before any oily water was discharged.
Until the late 1960s, almost every vessel was geared and there were fewer pontoon barges in service. Jettison of cargo was very often the quickest means of lightening a vessel, and time is generally of the essence in salvage. Ships were often smaller than they are today with the consequence that smaller quantities of cargo needed to be lightened.
The advent of gearless bulk carriers has deprived such vessels of the ability to self-discharge or jettison. Therefore, when they go aground—particularly if they are loaded with high-density cargoes such as ore—lightening can present a very major problem. The size of the vessel may also mean it will be difficult, within any realistic time frame, to bring in vessels with cranes of sufficient outreach before the situation has irretrievably deteriorated. Nevertheless, it is submitted that with dry bulk cargoes where it is not usually possible, as it is on a tanker, to shift the cargo around the ship to assist refloating, jettison must still be seen as an important option to facilitate refloating, to avoid a semipermanent wreck and in particular to avoid the risk of pollution from fuel oil onboard. It should be remembered that a bulk carrier, depending on size, could have anything up to 4,000 tons of fuel oil onboard, and that this is usually a more potent pollutant than crude oil.
Modern container vessels can present similar problems. Offloading containers from a stranded vessel in a remote locations can be difficult. Some containers might contain potentially hazardous cargoes and a modern container vessel might have a
bunker capacity of up to 5,000 tons. Jettison in one way or another might be the best solution.
In recent years, the Salvage Association has dealt with casualties where local authorities have either prohibited, or have been extremely slow to give approval to, jettison of cargo such as iron ore or even oranges, and this has put the prospect of successful salvage at risk and has sometimes led to more serious problems. In one case, for instance, fear of discoloring the water near a high-ranking official's private beach appears also to have colored the judgment of the local authorities. The subsequent breaking up of the gearless ship led to a much more difficult removal operation.
OIL ON TROUBLED WATERS
Despite quite wide enquiry, no classic cases of jettison of part of an oil cargo with successful salvage of the remainder have been identified in, say, the last thirty years. However, it is perhaps worth reflecting for one moment on how rapidly our reaction to oil on water has changed. As recently as 1961 a nautical textbook repeated the then British Ministry of Transport's advice to mariners on the beneficial use of oil for modifying the effect of breaking waves. Their advice included the fact that "the heaviest and thickest oil are most effectual."
JETTISON FROM STRANDED TANKERS
There appears to be only one case of purposeful jettison of an oil cargo in recent years and that case is somewhat infamous.
MT Zoe Colocotroni. In March 1973, this tanker ran aground off Puerto Rico carrying a cargo of crude oil. She refloated within four hours without tug assistance, but after jettisoning some 5,000 tons of crude oil, apparently without consultation. While jettison certainly appears to have allowed the vessel to refloat, contemporary accounts suggest that the situation did not warrant such drastic action.
There have been two cases on the northeast of the North American continent where jettison of oil cargo was proposed but not carried out.
ST Arrow. In February 1970 this tanker grounded in Chedabucto Bay, Nova Scotia, with approximately 16,000 tons of Bunker C oil on board. The weather was bad and she very quickly began to break in two. The stern section in which, it was estimated, some 7,000-8,000 tons of oil remained, was broken away from the forward section using tugs. This section grounded. It was then proposed to salvage this section by blowing certain tanks, which it was calculated would have meant about 10 percent of the oil escaping into the sea, but with the prospect of salving the other 90 percent. This was agreed in principle by the authorities but, for a number of reasons, operations were delayed. The stern section and the cargo sank in bad weather, resulting in further pollution. The full details of this casualty are to be found in the 250-page Report of the Royal Commission. In this case, the difficult decision to jettison oil was made but not quickly enough to ensure a successful outcome.
Argo Merchant. This tanker ran aground on the Nantucket Shoals in 1976 with 27,000 tons of No. 6 fuel oil onboard. The proposal to jettison up to 3,000 tons of oil was rejected by the authorities, apparently on various grounds,
including the risks of discharge with pumps which were not inherently safe. She subsequently broke up and discharged her full cargo into the sea.
The men on the spot in those salvage operations had the unpredictability of the future to contend with, and did not have the considerable benefit of hindsight. However, in the case of both the Arrow and the Argo Merchant, it has been suggested with hindsight that jettison might have been a less damaging option.
Usually a salvor would much prefer to salvage a grounded tanker than a grounded bulk carrier or container ship, particularly if the tanker has her cargo pumps operational. On a tanker cargo can be pumped from one tank to another to alter the trim and shift the weight. Provided a suitable tanker or barge can be brought alongside or close by, the cargo can be discharged, although this is neither an easy nor an absolutely safe operation. It is an option which has to be exercised with care. A tanker can be ballasted down very readily, if the seabed is suitable and it is not being ballasted down onto further pinnacles—and problems.
LIGHTENING OF TANKERS
While there have been no recent cases of jettison of oil cargoes, there have been very many cases of tankers being lightened in order to refloat. The fact is that, when tankers are loaded and underway, their momentum is such that if they ground even at slow speed they can rarely be refloated using tug power alone. They almost always require lightening. A possible exception is when a tanker grounds on a low tide and can be refloated as the tide rises. Tankers have been known to ground so lightly that the crew hardly notice and yet, even so, the ground reaction has been several thousand tons.
Basic principles suggest that a laden 100,000-ton tanker that goes aground at 5 knots is hardly likely to be refloated by tugs alone. The biggest salvage tugs afloat—and there are not many—have a maximum pull of 250 tons. The unlikely juxtaposition of three such tugs in one place would have a combined pull of 750 tons. It is worth reflecting that, for a ship grounded on a sandy bottom with a relatively low coefficient of friction (µ) of only .3, 750 tons of pull will only be as effective as lightening 2,500 tons of cargo. For a ship aground on a rocky bottom, 750 tons of pull might only be as effective as lightening 750 tons of cargo or less. Bear in mind, too, that rarely, if at all, would three such large tugs be available for one casualty, and many ships would not have securing points capable of handling the load. Furthermore, if a ship's bottom is damaged or she is impaled on an obstacle, then pulling might only exacerbate the problem.
In many cases, it will be more expedient to jettison cargo rather than to wait for equipment and weather windows which would permit a lightening operation. But in recent times salvors have not taken that option. Why?
Salvors, like others, are in general concerned for the environment. They would be reluctant to jettison without seeking approval from the appropriate authorities and they are very conscious of the obloquy that might attach if they were to act without consultation.
Oil cargoes are usually valuable and to jettison them is to reduce the salved fund, which might in turn reduce the salvage award.
Circumstances surrounding grounds have rarely given rise to the hard decision to jettison or perhaps it has been prohibited, as it apparently was in the case of the Argo Merchant.
REQUIREMENTS FOR SUCCESSFUL LIGHTENING
For lightening to be a valid salvage option, the right equipment needs to be available within a reasonable time frame. The right equipment will depend on the circumstances and size of the lightening operation that is required. It could mean a small tanker or a tankbarge. If conditions are right, they might be moored alongside. It might need bigger lightening vessels, and conditions might be such that they have to anchor off. There is much associated equipment required: large fenders, fuel hose, even portable inert gas generators and inherently safe fuel transfer pumps, if the casualty has lost power. Whereas much of this equipment is portable, the lightening tanker or tankbarge may take time to arrive. For instance, in the recent case of the Braer off the Shetland Islands, a suitable tankbarge was not on site until six days after the stranding. As it happened, the weather was so bad in the interim (though not abnormal for the place and season) that it is unlikely anything could have been achieved, even had the tankbarge been available earlier.
However, consider the situation where there is a fine weather window with the prospect of rapid deterioration shortly thereafter. If immediate jettison of part of the cargo has a chance of succeeding, do you risk deteriorating weather and delay until a lightening vessel is on the scene? Bear in mind, too, that engaging a suitable lightening craft might mean putting that at risk as well as the casualty itself and, because of that, chartering in such a vessel might not be easy.
If a casualty is ballasted down safely on a sandy bottom in an area of generally favorable weather, time might permit textbook solutions. Contrast a tanker partially aground on a rocky bottom with tugs available but no immediate prospect of lightening. Should jettison be considered if it might result in less overall pollution at the end of the day? On an exposed coast, though a simple operation in concept, lightening can bring enormous practical problems. It needs time, reasonable weather, and luck. These are not always available.
THE LEAST OF THE EVILS
If an aircraft is in trouble, fuel is sometimes jettisoned before an emergency landing. Safety of life is paramount, time is of the essence, and in these circumstances a decision to jettison would rarely be criticized. Of course, kerosene would usually, though not always, disperse before reaching ground or sea level. The absolutes of time and immediate threat to human life are not usually as stark when a tanker is aground and the decision process is therefore more complicated. Even so, time can be of the essence.
The tanker Braer, wrecked off the Shetlands, remained intact, to many people's surprise, for some seven days but contrast that with a casualty on the Norwegian coast a few days later. A very strong 700-ft offshore barge, which had gone aground, broke up completely and disappeared from view within 36 hours. The Aegean Sea which stranded off La Coruna had an even more precipitate fate. She broke in two within 18 hours. The time frame in shipping casualties is often very short.
Whereas the case with an aircraft emergency may be clearcut, the case with a ship carrying an oil cargo is less straightforward. Few people would sanction the
jettison of an oil cargo with the consequences of pollution just for the sake of saving property, the ship, and the cargo. But what about jettisoning cargo to save even greater pollution, as was intended in the case of the Arrow? If tugs had managed to get a line to the Braer, and if she was just aground, and if jettison was possible (and on the actual facts it clearly was not), would jettison of say 10,000 tons of cargo have been justified to prevent the other 70,000 tons causing much greater pollution? Remember, too, that the Braer was carrying nearly 2,000 tons of fuel oil and diesel oil (and a ULCC might carry up to 7,000 tons of bunkers).
The most notable recent tanker casualties have involved crude oil. But there will be circumstances involving cargoes other than crude oil or tankers. Let us consider some of the possibilities.
Carriage of Heavy Oils. The Arrow was carrying bunker C and the Argo Merchant No. 6 fuel oil. Both vessels ran aground in the cold of a northeast Atlantic coast winter. In both cases, the efforts to transfer the oil were hampered by falling temperatures, which made the oil thick. In these instance, perhaps a case could be made for oil to be jettisoned while it is still pumpable, rather than run the risk of the whole cargo being unpumpable and therefore lost.
Light Petroleum Products, Animal, and Vegetable Oils. All these products are covered by the Oil Pollution Act. The environmental impact of such cargoes will vary but generally it would be correct to say they would disperse much more quickly and be less damaging than, say crude oil and heavier oils. Would the jettison part of such a cargo be preferable to the total loss of both cargo and vessel and, perhaps, a significant quantity of much heavier fuel oil? Even for a relatively small ship, this could be as much as a thousands tons.
Chemical and Parcel Carriers. Whereas vegetable and animal oils might be considered relatively benign, other liquid cargoes carried afloat can be particularly toxic. A cargo such as toluene might form a part cargo. The balance of cargo might be far less worrying. Is there a case for jettisoning a relatively harmless cargo to prevent the serious ecological damage caused by the more dangerous substance?
Dewatering Enginerooms and Discharging Dirty Ballast. An engineroom may have flooded. Is it preferable to dewater the engineroom or discharge dirty ballast—both of which are likely to be oily to a degree—if this would increase the chances of a successful refloating and of preventing more serious pollution?
Bulk Carriers, Container Ships, and Other Ships. Such vessels can carry significant quantities of oil and might be carrying hazardous cargoes. Blowing out damaged tanks or jettison of some cargo or even fuel oil might, under extreme conditions, be the optimum solutions.
Because salvors have not resorted to jettisoning oil cargoes in recent years, it might be concluded that to prohibit such jettison is, like Rhodian law, merely to confirm what has become the practice.
Before that conclusion is drawn, however, it should be remembered that all the tugs in the world may, on their own, be unable to refloat many grounded vessels and that cargo will first have to be discharged. There will inevitably be situations where to wait to lighten into tankers or barges may be to wait too long. Better to jettison and accept that sometimes sacrifice is necessary for the common good and that this applies not only to preserving property but also, and perhaps more so, to preventing even greater pollution.
Michael Ellis is the general manager of the Salvage Association in London, England. He served in the British Navy for nineteen years, and during that time he qualified as a barrister. Upon leaving the Navy, he became a marine solicitor in London. He joined the Salvage Association in 1984, becoming general manager of that worldwide organization in 1986.
ENVIRONMENTAL RISKAS A FUNCTIONOF OIL SPILL SIZE
F. R. Engelhardt
Marine oil spills occur in a wide range of sizes, from very small volumes to thousands of tons of discharge.1 While it may be attractive to generalize that the larger a spill, the greater the potential for environmental damage, this is difficult to describe quantitatively. In fact, exceptions can be cited that suggest certain large spills, such as the Argo Merchant in 1976 or the Braer in January 1993, were less damaging than might have been expected.
Influential variables that will be discussed in this paper include:
Physical and chemical characteristics of the oil.
Physical environmental conditions.
Containment and recovery measures.
Geology of the impact zones.
Toxicological sensitivity of vulnerable species.
Ecological characteristics of vulnerable areas.
These variables are not independent of each other, but interact to characterize weathering rates and persistence, spread of the spill, direction of slick movement, effectiveness of response measures, size of the impact zone, extent and duration of biological effect, and degree and rate of recovery.
Efforts are underway to quantify the severity of oil spills in a way that integrates variables that might influence their outcome. Garriba2 for instance, describes an approach similar to the establishment of the Richter scale for earthquakes. Using this ''Marine Oil Spill Scale'' concept, the Argo Merchant, which released 29,000 tons of oil, ranked as one on the scale (i.e., an "anomaly" without significant impact). The Amoco Cadiz, with about 240,000 tons released, ranked as nine (i.e., a "critical accident" with widespread serious impact). Follow-up investigations could confirm that the Braer might be ranked as only a level two or three on the scale, even though the vessel spilled most of its 84,000-ton cargo nearshore.
PHYSICAL AND CHEMICAL VARIABLES
Physical characteristics and chemical composition of crude and other petroleum oils exert strong influences on the fate of an oil spill, as well as the environmental effects resulting from such a spill. There is extensive primary and review literature on this topic.3 A good summary correlating differences in physical properties of oils with their rate of removal from the sea surface is presented in Figure 1.4 A speculative mass balance for spilled oil, as shown in Figure 2, illustrates the range of environmental compartments that may be influenced by spilled oil and shows that the various weathering factors exert their influence differentially in the various media.
Oils are classified by a variety of standard nomenclature (API gravity, pour point, viscosity, specific gravity, etc.) depending on the user's perspective. One of the useful classifications in relation to oil spill response and spill effects is to characterize oils on the basis of their "weight," as light, medium, or heavy oils. This classification is analogous to a characterization of low to medium to high viscosity.
Light-weight oils have a low boiling point and high volatility, and are composed of a high proportion of low-molecular-weight hydrocarbons. They tend to evaporate readily from the sea surface, can be mixed easily into the water as oil particles by wave energy, and are also the most water soluble. These weathering processes are rapid so that within a day much of the spilled oil may be removed from the sea surface. Similarly, their persistence on oiled shorelines tends to be reduced if there is some sea energy available for physical washing action. The high proportion of low-molecular-weight hydrocarbons, including the aromatic compounds, tends to make the oils acutely toxic to aquatic organisms.
Medium-weight oils represent a mid-range composition, with a smaller proportion of low-molecular-weight hydrocarbons, which tends to evaporate and dissolve more slowly than the light oils. Their persistence on the sea surface is at least in the order of days, with the rate and amount of discharge influencing the degree of persistence of a slick. The toxicity of the oils is related to their lower physical and chemical ability, and they tend to be marked by bioaccumulation of hydrocarbons and longer term and chronic toxicity effects.
Heavy-weight oils are characterized by low evaporation and negligible dissolution, attributable to a high proportion of medium-and large-
molecular-weight hydrocarbons. The heavy-weight oils tend to be more adhesive and thus enhance persistence on shorelines. They show a potential for bioaccumulation and chronic toxicity, although the duration of exposure required for such effects also tends to be longer.
These generalities of oil weight are driven by the relative proportion of low-to high-molecular-weight hydrocarbons making up the majority of the composition of petroleum oils. The specific chemical nature of the component hydrocarbons also determines the fate of oil on the sea surface, which influences the fate parameters as well as the effectiveness of spill countermeasures. Oils containing higher-molecular-weight waxes, and especially asphaltenes, emulsify more easily to form persistent water-in-oil emulsions, which have a much increased viscosity. Oils with high asphaltene content can form stable emulsions, which do not easily break up naturally. Emulsified oils tend to weather more slowly because both evaporative, dissolution, spreading, and dispersion processes are inhibited. They also tend to be more sticky, adhering to a greater degree to shoreline material and biological surfaces, as well as countermeasures equipment. One of the additional characteristics of oil taking up water is that its volume increases by several factors, and its density can approach that of sea water.
Specific chemical composition is a major determinant for acute and chronic toxicity effects. Light oils tend to be more acutely toxic, particularly if their composition encompasses larger proportions of benzene and benzene derivatives. In general, the toxicity of oils is linked to the proportion of aromatic compounds, including the benzenes, naphthalenes and polynuclear aromatic hydrocarbons. The larger-molecular-weight aromatics tend to be bioaccumulated and are more associated with chronic toxicity effects, including a potential for such compounds to be mutagenic and carcinogenic.
On the basis of physical and chemical properties alone, and ignoring the dependent environmental/biological/ecological variables discussed below, it can be speculated that in relation to spill persistence and toxicity, smaller volumes of medium-to heavy-weight oil spills may be as potentially damaging as larger volumes of low-to medium-weight oils. This relative inverse relationship is enhanced by waxes and asphaltenes in medium-to heavy-weight oils. The relationships between physical/chemical properties and possible adverse effects is summarized for light to heavy oils in Table 1.5
PHYSICAL ENVIRONMENTAL CONDITIONS
The physical environmental conditions of relevance to oil spill fate and effects and to spill response are sea state (a combination of wind, wave energy, and temperature), salinity, ocean current profiles and in some instances the presence of sea ice.
Although there are few data available from direct measurements at sea that quantify oil property changes in relation to sea state, model predictions and mesoscale tests have provided significant insight into the changes that might be expected with
Physical/Chemical Properties and Possible Adverse Effects of Common Oil Types During Spills
Adverse Effects on Environment
Light to volatile oils
• Spread rapidly
• Tend to form unstable emulsions
• High evaporation and solubility
• May penetrate substrate
• Removed from surfaces by agitation and low-pressure flushing
• Toxicity is related to the type and concentration of aromatic fractions
• Acute toxicity is due to aromatics: 1) naphthalene, 2) benzene
• Toxic to biota when fresh
• Toxicity of aromatic fractions depends on their biological half-lifes in different species
• Mangroves and marsh plants may be chronically affected due to penetration and persistence of aromatic compounds in sediments
• Marine plants (especially mangroves) may be adversely affected by smothering
Moderate to heavy oils
• Moderate to high viscosity
• Tend to form stable emulsions under high energy marine environments
• Penetration depends on substrate particle size
• Weathered residue may sink and be absorbed by sediment
• Immiscibility assists in separation from water
• Weather to tar balls
• Adverse effects in marine organisms result from chemical toxicity and smothering
• Toxicity depends on light fraction
• Toxic effects reduced in tropical climates due to rapid evaporation and weathering
• Low toxicity residue tends to smother plants or animals
• Light fractions contaminate interstitial waters
Asphalt, #6 fuel-oil, Bunker C, waste
• Form tarry lumps at ambient temperatures
• Resist spreading and may sink
• May soften and flow when exposed to sunlight
• Very difficult to recover from water
• Easy to remove manually from beach surface with conventional equipment
• Immediate and delayed adverse effects due to small aromatic fractions and smothering
• Most toxic effects due to incorporation in sediment
• Absorption of radiated heat places thermal stress on the environment
• Lower toxicity in marine plants than mobile animals
SOURCE: Exxon Production Research Company. 1992. Oil Spill Response Manual.
time after a spill.6 Both model predictions and test measurements demonstrate enhanced evaporation with increased wind speed, as well as marked water uptake into the oil, i.e., emulsion formation and increased viscosity (Figure 3). This example
is specific to one North Sea crude oil, but similar relationships have been demonstrated for other oils. The significance of these changes relates especially to the effectiveness of spill containment and recovery measures, as discussed in a subsequent section.
It may be postulated that small quantities of spilled oil would at least initially be more susceptible to the influence of these environmental variables, However, it is difficult to be definitive since there is little quantitative information on the relationship of wave energy spectra with dispersion potential, as well as to the probability of formation of emulsions. Although it may be suggested in a speculative fashion that the wave dampening effects of oil slicks would be greater for large oil spills than for small ones, this is probably only plausible in the early stages of a spill when there has
as yet been little spreading of the oil to thin the slick layer. The relationship of slick size to volume released is summarized in Figure 4.
An oil spill on the sea surface offshore tracks in accordance to wind direction and surface currents, which are predominantly wind driven. Other current influences also have to be considered, such as convergence at shelf edges and tidal currents in the nearshore. Differential influences on large versus small spills are speculative, but a large spill, covering a greater area of the sea surface is likely to come under the influence of a greater number of ocean current variables. The large spill would probably show a more complex pattern of distribution and be more difficult to predict and track.
The presence of sea ice has a significant effect on both the behavior of oil and the ability to apply countermeasures. Most of the oil spill containment and recovery systems used in temperate waters are compromised in their effectiveness by the presence of ice. New technologies are slow in being realized, although burning of oil in and on ice may become a useful countermeasure. In some circumstances, ice serves to corral oil so that oil recovery can be applied, but it also can prevent access. Further, mechanical countermeasures are limited in ability to handle solids (such as broken ice), remote sensing capacity is limited, modeling of the fate of oil is complicated by inadequate detailed knowledge of under-ice currents, and bioremediation is likely to be slow because of the low water and shoreline temperatures. Conversely, the ability of ice to restrict the movement of oil on the sea
surface can be an asset for in situ burning, whose promise as a countermeasure is supported by a growing experimental database.
CONTAINMENT AND RECOVERY
When oil is spilled at sea, it is subject to several physical and chemical processes, including spreading, drifting, evaporation, dissolution, photolysis, biodegradation, and formation of both oil-in-water and water-in-oil emulsions. These weathering processes may lead to drastic changes in the chemical and physical properties of the oil and therefore also in oil behavior. The main factors influencing the rate and extent of weathering are waves, wind, sunlight, air and sea temperature, and salinity.
Spreading, evaporation, and especially formation of water-in-oil emulsions, can cause a drastic increase in oil viscosity, and may therefore be of great importance for decision makers concerning the use of the different spill response techniques during an oil spill combat operation.
Spill Response Techniques
The choice of spill response techniques should vary with weather conditions, changes in the chemical and physical properties of the oil, and as a result of any delayed response time following an oil spill.
Major spill response techniques during an oil spill combat operation might include use of mechanical recovery equipment, in situ burning, and chemicals or "soaps" as dispersants and demulsifiers. Figure 5 shows the relative efficiency of various response techniques in relation to oil slick thickness, and Figure 6 demonstrates potential volume control rates for selected spill response techniques.7
In Situ Burning
The effectiveness of ignition and combustion of oil floating on the sea surface is dependent on a mixture of variables, including wind, waves, rain, initial oil thickness, oil thickness reduction, formation of emulsions, evaporation and dispersion.8 However, the three main limiting factors for ignition and combustion of oil at sea are the oil thickness, oil thickness reduction, and the water content in the emulsion. A number of generalities of operational relevance can be made, drawn from both observational and experimental data.
Laboratory and mesoscale testing clearly indicate that ignition and combustion of oil having an oil thickness less than approximately 2 mm are extremely difficult. The spreading of oil from a source on the open seas will within a short period of time cause a reduction of the thickness to less than 1 mm, preventing ignition and further combustion.
For effective ignition and in situ burning at sea as a response measure, fire resistant or fireproof booms are needed in order to create and maintain a sufficient oil thickness. Due to the heavy weight of fireproof booms, however, they have poor wave-
following characteristics and are reported to be ineffective in average waves higher than 3 to 4 feet (e.g., 3M Fire Containment Boom Specifications). In wind velocities above approximately 4 knots (2 m/s), emulsification of some oils can take place within minutes and may increase the water content up to 60 to 80 percent within a few hours. Other oils are less susceptible to such emulsification. An effective limitation for ignition and combustion, however, is in the range of 30 to 50 percent water content.9
The efficiency of in situ burning is claimed to be above 50 percent and even above 90 percent, depending on oil type and weathering properties. For emulsions, however, the efficiency has been as low as 10 to 35 percent (Etkin, 1990). For the most part, the high-efficiency figures are laboratory derived, and given current technologies would be difficult to duplicate in many spill situations. Better definition of this issue is needed since in situ burning may in the future be a common response technique. Responders will therefore need realistic data on the possibilities, limitations, and the effectiveness of in situ burning for various fireproof booms, oil products, and weather conditions. An in-depth study of the operational feasibility of in situ burning as one of the suite of possible response tools is being carried out by the Marine Spill Response Corporation.
The application of dispersants has been used for removal of oil spills from the sea surface since the Torrey Canyon incident 25 years ago, when they were used for the first time in a major oil spill. The use of dispersants in the United States has heretofore been limited because of regulatory impediments, but it appears that future use in spills may be considered more favorably. However, the actual effectiveness of dispersants used at sea under the many different oil spill situations requires further understanding.
Most studies on the effectiveness of dispersants used in the field indicate values of 20 to 70 percent, and under some special circumstances even higher, but most values are below 35 percent.10 However, it has been difficult to measure the effectiveness of dispersants in the field, and no common standard methods are available for establishing a mass balance. Laboratory effectiveness testing of dispersants may also give variable values since several different standard test methods are used.11 The test systems represent a range of mixing energies, claimed to represent different weather conditions or sea states, but today no real correlation exists between the mixing energy within the apparatus and the mixing energy in the ocean as related to wave energy spectra. Studies are underway at the Warren Springs Laboratory in the United Kingdom to address this important issue.
The effectiveness of dispersants will vary depending on many physical and chemical factors, such as oil properties, application method, droplet size, oil thickness, dosage rate or dispersant-to-oil ratio, wind and wave conditions, salinity, temperature, viscosity, pour point, and emulsification. Possible relationships between
The use of demulsifiers in order to prevent formation of or reduce the amount of emulsion on the sea surface, followed by use of dispersants, is a relatively new response method under development in the United Kingdom. Application of demulsifiers during mechanical cleanup operations may also reduce the water content in the recovered emulsions and extend the time in operation in the field by decreasing storage requirements.
Mechanical Recovery Measures
Oil spill cleanup based on mechanical equipment continues to be the most common response method for oil spills. Among mechanical recovery equipment, there exist a large number of booms and skimming principles. Booms for containment tend to be designed for operation in either calm, protected, or open ocean areas, and skimmers are often designed to operate within certain viscosity ranges. Their performance capabilities and effectiveness will therefore depend on the area of operation and the weathering properties of the oil, in particular the sea state, oil type, oil thickness, and degree of emulsification. The key environmental constraints have been summarized, showing probable maxima for sustained performance ability for major open ocean containment and recovery systems currently in use:13
Beaufort Scale, 3 to 4.
Waves, 2 to 4 feet significant wave height (possibly up to 6 feet with certain equipment).
Winds, 15 to 20 knots sustained.
Currents, 1 knot.
Assuming that equipment for spill response is available as identified in vessel response plans to deal with both small and large spills, the relationship of spill size to effectiveness of the response would probably be driven mainly by the increased complexity of large spills. More oil on the sea surface also increases the probability that environmental conditions could become limiting to the response, increasing the time for weathering of the oil which is known to increase the difficulties of recovery operations, as described above. Quantification of the relationship, however, requires specific knowledge of spill response resources available, the time required for them to arrive on scene, characteristics of the spilled oil, and the environmental conditions of the spill.
There is a strong possibility in the case of large offshore spills that oil will impact shorelines even with best effort responses for control and recovery. If the discharge occurs nearshore (most tanker accidents occur in the nearshore area), the oil spill is likely to reach the shoreline in only a short time. An evaluation of the significance of such contact has to take into consideration the geomorphology of the shoreline. It is well understood14 that high-energy shorelines are more easily washed clean by wave action, Also, there tends to be less retention of oil on rock faces and in unsorted beach material, although the viscosity of the impacting oil is an influential factor. Low-energy shorelines show a longer retention of beached oil, including sandy beaches and biologically sensitive areas such as salt marshes and mudflats. These factors have to be considered in predicting the impact of spilled oil, but also in logistic planning for deployment of response resources. Response resources are finite and the overall effectiveness of a spill response may be enhanced by targeting efforts to those shoreline areas which might be most prone to impacts.
The toxicity of oil has a wide span among the different marine species, ranging from less than a part per million of oil to high concentrations. A consideration of toxicity to individual organisms has to take into account at least the following:
Characteristics of the spilled oil.
Degree of weathering.
Mode of exposure (contact, inhalation, ingestion).
Duration of exposure (acute or chronic effects).
Sensitivity of the individual organism of a species, which may vary with life stage.
The sensitivity of a species or a population depends on its ecology. The zones of potential impact of oil spills are spatially limited to areas that contain oil, modified by the effect of dilution of toxic fractions of oil to a threshold where acute and chronic effects no longer occur. Because oil weathers both physically and biologically, the spatial extent of an impact zone decreases with time, noted with all spills that have been investigated. The temporal change is of importance in determining the potential for exposure of a population—the population has to be present at a time when the oil is present in toxic quantities. For example, the effects of the Braer spill in January 1993 are likely to be less severe for seabirds because it occurred in midwinter when most of the birds were not present in traditional colony areas. The high sea state at the time of the spill and the low persistence of the spilled oil will probably minimize residual effects on marine life when the seabirds and other species return in large numbers for the summer season to the Shetland Islands. By comparison, the Amoco Cadiz spill, which occurred in a biologically more vulnerable time, had a more severe effect on marine populations.
It is beyond the scope of this paper to review these relationships in any degree of detail. Such information forms the bulk of the oil spill literature and is presented in many summaries.15 A survey of the major biological groupings is presented here to round out the perspective of impact evaluation and to indicate which groups are unlikely to show lasting effects from small or large spills as compared to those that might have specific sensitivities driven by their ecology. Such sensitivities are
predictable in principle, and form the basis of environmental and biological sensitivity mapping used in oil spill response planning.
Microbial Effects and Biodegradation
Interest in this trophic level centers on two main aspects: a recognition that microbial systems constitute the bioenergetic basis of the marine ecosystem, and that microbes, in particular bacteria, can degrade contaminants such as hydrocarbons. It has been determined that the composition of the microbial community changes with exposure to hydrocarbons, generally in favor of hydrocarbon degraders, the oleoclasts. Such changes may take days to months in marine waters and sediments, depending on water temperatures and the degree of any chronic pre-exposure to hydrocarbons. Biodegradation is most effective for lighter oils and in particular for the alkanes. High levels of oil appear to inhibit biodegradation due to a direct acute toxicity effect, which can be of relevance in considering differences of impact comparing small and large spills.
Oil pollution effects on these primary producers have been little studied, but it appears that growth and photosynthesis are inhibited by at least high oil concentrations. The significance of such an effect may, however, not be great since the effects of an oil discharge are local or regional in scale and would likely be buffered by adjacent phytoplankton populations once oil is no longer present at toxic levels.
If phytoplankton accumulate hydrocarbons, they may function as a vector for the biotransfer of these hydrocarbons to other trophic levels, especially to herbivorous zooplankton and filter-feeding benthos. Again, the environmental effect is likely to be local or regional, and of limited duration.
Although extensive lethality data are lacking, information available for zooplankton suggests LC50 values in the order of a fraction to a few parts per million. Such concentrations may be expected in the water column after a spill, with consequent debilitation of the zooplankton population in the local area. Again, exchanges among water masses and the plankton components are likely to buffer this effect with time. Indeed, observations of zooplankton populations in spills have shown that while there is an effect of oil it is short-lived, and there are few changes in the biomass or standing stocks of zooplankton in adjacent open waters.
The ability of zooplankton to take up hydrocarbons has been demonstrated and suggests a potential for biotransfer. However, long-term bioaccumulation and transfer to fish, for instance, is of low probability following an oil spill since the zooplankton are able to void accumulated hydrocarbons. Tainting by biotransfer of higher trophic levels of the food chain is likely to be a temporary and localized phenomenon.
The benthic invertebrate biota are an important component of the marine ecosystem, providing an energy base for fish, seabirds, and marine mammals. They respond to disturbances and represent an ideal effects monitoring system. Bivalves
and echinoderms show behavioral changes to hydrocarbon contamination that may limit their survival, such as emergence from sediment in mussels and clams, and narcosis in many species. This can occur after acute, post-spill exposure as well as after long-term chronic contamination in the parts-per-billion range. In addition, benthic invertebrates are able to accumulate hydrocarbons to high levels from the surrounding medium, suggesting biotransfer as a possible concern, at least until such time as the accumulated hydrocarbons are voided.
Acute effects in benthic invertebrates tend to be tempered by their localized nature. Such a geographically limited effect might be significant if a local benthic population become reduced or contaminated in obligatory feeding areas for animals such as walrus or seabirds, eider ducks for instance. Wide-ranging chronic effects in nearshore benthos may be of concern following a large spill with a wide degree of lasting shoreline contamination.
One area of benthic life that has received only scant attention is that of macroalgae, or seaweeds. The seaweed community provides a habitat for many dependent species that form part of the nearshore food web. The presence of oil limits photosynthesis by biochemical inhibition and decreases primary production. While the loss of macroalgae can be expected to change a nearshore benthic ecosystem for the short term, it can also be anticipated that recruitment through pelagic spores or larvae would eventually normalize a local or regional ecosystem effect.
Adult fish appear to be fairly resistant to oil exposure, in contrast to their sensitive egg and larval stages which are often planktonic. Fish tend to leave areas of high contamination and relatively little mortality is recorded. Sublethal effects include impaired physiological salt and water balance, which might be crucial to anadromous fish such as salmon when they enter the freshwater phase of their spawning cycle. A unique vulnerability for arctic fish may be at the ice edge, which is considered to be an important and productive habitat for many species. There is little evidence that standing stocks of fish have been much changed by oil spills. A more likely consequence is impact on harvest activities, either because the adult fish have left a contaminated area or because such fish have become or are perceived to be tainted through contact with oil.
The fate of seabirds has drawn great attention for several reasons. There is little doubt that birds exposed to oil fare poorly. The primary problem is a loss of thermal insulation, along with a decrease in buoyancy, increase in metabolism, and decreased reproductive success. Certain species form special sensitive cases. The alcids, including murres, dovekies, and razorbills, are particularly susceptible, especially in northern areas where they tend to breed in a very few but large colonies, with a low reproductive turnover. An oil spill in the vicinity of such a breeding area has a potential for serious impact on the population, and a prediction of impact would require close evaluation of the fate of the oil spill.
Investigations of actual oil spill incidents have generally not been conclusive in identifying the toxicity of petroleum in seals or whales, even though mortality has been attributed to oil exposure at sea. Only some of the species have demonstrated a clear sensitivity to petroleum in experimental studies. Recent studies in seals, sea otters, polar bears, and whales have helped to round out the limited information base on the subject. Although the cetaceans at least are able to detect oil on the sea surface, it remains controversial if marine mammals would avoid oil spills at sea. In some circumstances both whale and seal species have been observed to surface through oil slicks.
Contact with viscous oils can lead to long-term coating of the body surface of the furred marine mammals to result in thermoregulatory stress, or may interfere with the filtering capabilities of baleen in whales. A limited experimental data base suggests that seals, whales, sea otters and polar bears differ in degree of clinical toxicity damage following exposure to petroleum. It is clear that both seals and whales are able to absorb hydrocarbons and will store the contaminants in blubber, as well as to a lesser degree in other body tissues. Tainting of harvested marine mammals is considered a potential problem.
Population-significant impact on marine mammals appears to be a potential only in definable circumstances, that is, restricted to localities that may seasonally host a large proportion of a population. The high densities of white whales in estuaries and bays may be a case in point, as are traditional colony areas for walrus. Evaluations of impact have to be case specific.
It is possible in only a general way to state that large spills have a greater potential for environmental impact than small-sized spills. Since there appear to be significant exceptions to this statement, the situation should be analyzed in depth on the basis of global record for marine spills. An approach using the proposed ''Marine Oil Spill Scale'' might be useful. The results of both experimental data and information gathered from oil spills point out that a prediction of spill impact related to spill volume must take into consideration a wide range of variables. These include the characteristics of the spilled oil, physical environmental conditions, the effectiveness of oil containment and recovery measures, and the biological parameters of affected areas.
Dr. F. Rainer Engelhardt is vice president for research and development for the Marine Spill Response Corporation. He has an extensive background in oil spills and their effect on the environment. He was formerly in executive management positions with the Canadian government, including the Canada Oil and Gas Lands Administration.
OIL SPILL TRAJECTORY MODELING
There is a perception in the general population that oil spill models represent reality: turn one on and it tells what is going to happen with a spill and what is going to happen in the ocean. That is not true. Each model has special strengths and weaknesses and it is not reasonable to consider any model outside the context in which it is intended to be used. My charge is to cover six aspects of trajectory modeling:
How present models work.
The types of models that are available.
Factors that limit model accuracy.
How models are used.
What technology is available to support models.
What improvements are needed.
A model is a tool. You have something in mind when you apply it to a particular situation. When considering a model to support a jettison decision, the particular type of incident must also be considered. For example, a grounding implies that a certain subset of ocean dynamics are involved. Most importantly, it is nearshore, and models that might work very well in the open ocean won't work very well nearshore. Groundings have a tendency to degenerate and malinger. If the ship goes aground, there are a lot of feathers flying right away but it is very likely still to be a problem two weeks later. Thus for most groundings we should be looking at a model that has some capability to understand what is going on in a time scale of a week or a month, occasionally longer.
Yet when a grounding occurs, something needs to be done to stabilize or fix the situation quickly. The initial output, the initial recommendations, the initial look forward from a model, should be available very quickly. If a ship parks on a rock, bad things are going to start happening fast. The typical time scale is a tidal cycle. If a tanker sits on a rock and the tide drops it 10 feet on its end, it has a serious problem. We need models that have high resolution, that can look forward in time, and that can get an answer out very quickly. That is the context in which we are going to try to evaluate and think about models.
HOW MODELS WORK
Virtually all oil spill models that are available have to say something about the current, because anything dropped in the ocean is carried away or drifts off with the flow. This involves looking at oceanographic flow problems and hydrodynamics. In a nearshore regime, there is considerable complexity in this kind of a problem. One of the first things we need is models that realistically cover geophysical geometry. There
are plenty of models that describe the open ocean. Currents are averaged, for example, over 50 miles square or 100 miles square. Those kinds of models won't help. We need to resolve complex shorelines and realistic bathymetry, and it is complicated.
In every model I know, shoreline and bathymetry factors have actually been taken out of the model and set in a separate submodel. Oceanographers run models that try to do that. For example, a typical model that would resolve geometry would be based on a finite element scheme that can get realistic shapes down to some resolution. An alternative proposal would be to use a finite difference grid, or just a bunch of little cells. In that case, the cells have to be quite small in order to resolve what you need to look at, and this would require much larger computer resources.
Another characteristic of oil spills that every model has to deal with is that they start off small and (if they are going to be a problem) they eventually become big. Models have to resolve multiple scales. You can't get fine resolution in one spot then carry it throughout because you will run out of computer space. Models are thus formulated in a mixed sense; that is, they use a combination LaGrandian/Eulerian formulation. This means that they consider the oil as a bunch of little pieces of oil and the flow as a larger scale field. You keep track of the oil by keeping track of all these particles and making some inference out of the information.
Models need to consider other physical processes besides moving along with the water. Most of them have a weathering component. Again, depending on the kind of oil involved, some fraction of the oil is going to disappear on its own. This means it will go somewhere else in the environment; it can evaporate, or it can disperse in such a way that you can't find it any more. These factors are typically represented algorithmically and most models have such a component. They may vary in terms of quality, however. Some are quite coarse, but generally a sensitivity analysis would show that most of the models are adequate compared to human observational skill.
Most models also have algorithms or representations for beaching effects. When oil gets near a shoreline, it becomes a problem on the shore and you need to represent that computationally. The naive view is that when oil approaches a beach, it just drives up and parks on the beach. That is not the case. Oil is probably more easily modeled like a tennis ball, you can get about three bounces out of it as it goes ashore. When oil approaches a shoreline, it has to stick to the shoreline. Typically, it will go ashore on a falling tide and an onshore wind. If it gets next to the beach, the currents alone won't take it ashore: it has got to be held against the beach by a wind and then the tide has to drop out from underneath it.
Many times we have observed oil ashore when the winds are not onshore and the tide is dropping, but the oil doesn't stick. When oil approaches a shoreline and the tide is rising, it won't stick. It just continues to fill up the beach face. For example, in the Huntington Beach spill, oil was on the shoreline and in the surf zone for a number of days, but the falling tide occurred in conjunction with a sea breeze reversal. There wasn't any oil on the beach for three days while those conditions prevailed, although the oil was right there. This provides an excellent opportunity to clean it up, by the way.
Another important—almost determining—factor in many model examples is wind drift. Oil is affected by wind in several different ways. When the wind blows, it scoots the ocean surface along with it, so you need the wind as well as the current to determine where the oil is going to go. The oil also interacts directly with the waves created by the winds. An oil slick is smooth because the waves are damped out. The first thing you notice about an oil slick is the difference in roughness. Oil absorbs
small waves. Small waves have a small amount of energy. They push the oil around. An oil slick will motor through the water it is floating on because it is being pushed by the waves it absorbs. This is another reason wind effects have to be considered. Currents by themselves almost never penetrate a shoreline. (In a marsh, there may be some inundation and percolation through the shoreline, but generally the currents can take oil toward the shore, but it won't beach without wind.) Thus the wind field is critical for determining this particular aspect of a spill. All the models I know that are routinely used contain these kinds of processes.
TYPES OF MODELS
The differences in models are not so much in the detail, but in the resolution and feeding requirements, the kind of data it takes to make the models run, and the input/output presentation. These factors are designed for a variety of different hardware platforms. These days models run routinely on anything from a microcomputer to a supercomputer, and there are models that have been developed all over the world:
Warren Springs in England has Eurospill.
There is a commercial version of the U.S. World Oil Spill Model.
A model used in government called OSSM stands for On-Scene Spill Model.
Florida is running a large model on a Cray-type supercomputer.
Regional areas have their own models.
There is something called Gulf Slick, which runs from the Arabian Gulf.
Each of these models include all these kinds of processes. The big differences are how you feed them and what they show you when they are over.
LIMITATIONS TO MODEL ACCURACY
There are a lot of problems with how models work. First, if we want a model to respond very quickly, we are typically talking about one based on climatology or average conditions. The reason is that there are no real time, up and running, forecast models of the ocean. In theory, one could make a model of the ocean which would be similar to the weather service models; turn it on and it will tell you what is going to happen with the weather tomorrow. Atmospheric models are supported by the World Meteorological System's network of measurements at hundreds of places around the world every six hours. That network is required to bring a model back to reality whenever necessary. In other words, there is a hindcast/forecast procedure. There is no such system for the ocean. This means if my phone rings and somebody says "I have a spill," I can't go see what happened at that site yesterday. I have to go on climatology. Another way to think about this problem is, if you wanted to plan a picnic for next August, you could look at climatology and say that is a good idea, it is probably going to be hot. If you want to plan a picnic today, you should look out the window. A spill situation is like that. Unfortunately, it is difficult to look out the window—not impossible, but difficult.
Another problem with available oceanographic data is the effect of seasonal flows that are nonaverage. Freshwater input is a critical factor because it tends to govern the nearshore flow and it tends to disproportionately affect movement of the oil. These are poorly understood factors. Climatology, for example, will not predict heavy rain freshets coming down the Delaware River and affecting a spill in the Delaware
Bay. Climatology won't include many situations that are environmentally difficult, which is a time when ships tend to get in trouble. The average conditions off the Shetland Islands didn't make much sense compared to what was going on in the Braer. The two or three spills on the Gulf Coast after Hurricane Andrew were not average in any sense, either.
Another problem that is almost impossible to understate is that we need to know where the oil is going. To do that, we need to know the velocity field, i.e., what is happening in the currents. That is difficult to determine. More importantly, we need to know the derivative of the rate of velocity change. When a substance starts to disperse, mixing and entropy and chaos are going to make it less and less concentrated and it is going to spread. For example, a smoke plume from a smokestack gets less and less dense until it thins out entirely and goes away. It is never possible for that smoke to recoalesce into a black cloud and splat onto the side of a building 100 miles away. This is not true about oil because oil is buoyant, so when ocean water converges and sinks, the material floating on it stays on top. That gives an oil spill the power to recoalesce. Time after time, we see spills spread out, get thin, disappear, then suddenly reappear in tide rips and convergence zones. When they do reappear they can do it in a form that can do tremendous damage. If this didn't happen, however, we wouldn't ever have an opportunity to recover the oil. Once again, compare oil spills to smoke from a smoke stack.
Thus there are aspects to currents and current flow that are very important, Yet we are not even close to being able to predict them with models. Invariably, when we deal with a spill, these are things that are determined observationally. We just don't have any models that will resolve these factors on the scale we need in the nearshore environment.
Other small-scale, nearshore factors that we can never model are such phenomena as rip currents. A rip current is a current that turns perpendicular to the shore and carries water out of the surf zone. If oil is in the surf zone, the rip current rejects the oil back out to the ocean and leaves a smearing effect on the beach. We don't understand these phenomena very well. We know they occur, we have seen spills where they are important, but there are no models that can resolve them.
Finally, present models lack technological ability to present information well, such as good graphic presentations and the ability to routinely transmit the data to other places. Running a model and full four-color or sixty-color pictures on a console doesn't do much good unless you can project that image to somebody who needs to make a decision. The next problem is to take the data and overlay it with resources. It isn't enough to see a threat profile of where the oil is going, we need to know who it will hit and how it will hurt. Those are the kind of real questions we want to answer: we are just beginning to do that.
One advantage I have from responding to spills and running these models for many years is I have some feeling for accuracy. Typically, when we get called on a spill, we make an immediate projection based on available climatological data and whatever we can infer at the time. We can predict where the oil is going. About two-thirds of the time I am pretty close in terms of directions. But there are other errors in this process. For example, the Weather Service can tell you the wind direction—north, northeast, east, something like that—but the bad news is, a third of the time it is somewhere else. This is what climatology will do to you—like planning a picnic months in advance. When on-scene observations begin to come in, you can improve on
those predictions. Whenever we go to a spill, we have an experiment in progress. The first overflight comes back with data and we begin to calculate hindcast/forecast information. The reliability of information improves the forecast quite a bit. Gathering nowtime background information is important. We don't have capability at the present time to be collecting this type of data anywhere, let alone everywhere.
HOW MODELS ARE USED
The naive approach is that the first thing you want to do with a model is forecast—take everything you know, take the best weather forecast, specify where the oil is, and predict where it is going. We do that all the time. But in many respects it is not the best use of the model because we are often considering a problem that has a time scale longer than the forecast period. The Weather Service will forecast to 24 hours; call them up on the back line and you can get 48; beat on the person doing the forecast and you might get 72. But they are not going to stand behind it and nobody believes it very much anyway. In general, when we talk about a spill, particularly if we expect it to degenerate and lose the whole cargo, it will be weeks.
What we need to do is consider using the model in another way: run the model in a statistical mode, look at the average conditions, go back to wind statistics, go back to current statistics, and go back to tide statistics. Invariably, in dealing with a spill, the first thing we do is determine which direction we think it is going.
The second thing we try to do is try to figure out what the bound—the envelope—is. This brings us to trajectory analysis. A model is one piece: it is the analysis that you want. No model is accurate enough to stand on its own. We need to understand what the model tells us and what it doesn't tell us.
Another way to run a model is something called a receptor mode. This is when you say, "I don't care if oil is all over the place. I only care if oil threatens a high-value resource." The response is to try to create a spill at a high-value resource—such as a wildlife refuge or a marina full of politicians' boats, or anything else you don't want to get oil on—then trace the spill backward and see where it could have come from.
Statistically, this is a fairly intensive prospect, but the result is two maps. The first map is a joint probability of distribution map: it says, if you start at point a, there is a probability X that you will end up at point b. It is also a threat zone map. It tells where oil could come from and threaten a resource. These things are statistical in nature and can be done ahead of time: receptor mode maps can be plotted for all areas, so if an event occurs the threat zone is determined. The second thing the receptor mode does is plot out a time-of-travel contour. With these two maps we have what the probability of the threat is and the time available to react to it.
Forecast, statistical, and receptor mode analysis are all ways to use models. On a major spill we run all of them. Another aspect of strategy doesn't have to do with how to run the model, it has to do with how to think about it. This gets back to the idea that most people naively think the first thing to do with a model is run a forecast—take a best guess at the weather and figure out what is going to happen. In game theory, that is equivalent to going for the big win, doing what offers the best chance of success. When dealing with high-value resources, the best game strategy is often not to go for maximum win but for minimum regret. It is an exploration of the situation—to find out the worst downside and then plan for that scenario. Unfortunately, what I am talking about doesn't lead to certain answers. I don't believe there is a silver bullet associated with computers being able to determine that it is a good idea to jettison
cargo. They can establish that it could be pretty bad if you don't, but that is not going to help much.
There are some technological options that can be helpful in trajectory analysis. For example, technology is available to help in such areas as gathering information from satellites. We use satellites all the time. If there is a spill off the east coast of the United States, we can get a thermal image of the Gulf Stream in a flash. That is a little further offshore than the scenario we are talking about at this symposium, but if you happen to be in a Gulf Stream eddy going the other way, you would like to know that, and a satellite picture provides the information quickly.
On a more relevant scale, the west coast of the United States has a banded current system which, from a forecasting point of view, is perverse in every regard in the sense that it reverses at a moment's notice. It is seasonally one way, then seasonally another; in between seasons, it is something else again. One thing we do is get a thermal picture of the coast, because when the current is the right way it often causes upwelling and when that happens there is a cold spot behind headlands. We look for those right away as indicators. We are talking minutes in terms of the time to set up model response.
There are also experimental surface current radars that can actually map currents out to about 50 kilometers. A few of these are in semi-permanent shoreside stations, one in Monterey, California, and one in Florida. We have tested them in the Strait of Juan de Fuca. The Canadians are developing them and the United States has some. These are interesting and encouraging, but unfortunately they were also encouraging about 15 years ago when we first started seeing them. The data analysis required for them is massive, and there is no place where we have used them in real time. We may see more of this technique in the future.
Some harbors actually put current meters (Doppler acoustic systems) on the bottom that look back and forth at the current. These may help and they do work in real time, but they have not yet seen spill action. Another technology that is available but not online is pattern analysis on shelf circulation. The oceanographic problem is "shelf waves," which are analogous to the high/low patterns in the atmosphere, except that they wander up and down the continental shelf and cause highs and lows and cells to migrate around. On the Texas coast, shelf waves probably represent 30 percent of the current. They can be spotted under optimal conditions by satellites, and hooking up a satellite with a real-time background model has been talked about. This would be a first attempt at trying to do what the Weather Service does, by keeping up a background model. Again, this is experimental and it is expensive.
Finally, we will need real time capability. If we had background forecasters (remember, when we go on spills, we have an experiment that is running) that were looking at the coast, talking to the fishermen, telling if the current had reversed, that would give us a leg up. There are no quick answers. We are nowhere near long-term forecasts, so we will have to rely on statistical analysis and trajectory analysis to predict the consequences of a spill.
Jerry Galt is chief of the Modeling and Simulation Studies Branch, Hazardous Materials Response and Assessment Division, National Oceanic and Atmospheric Administration. He is a physical oceanographer from the University of Washington with considerable experience in trajectory models.
DISCUSSION: QUESTIONSAND COMMENTS ADDRESSEDTO THE SPEAKERS
QUESTIONS ADDRESSED TO MR. ELLIS
DAVID HUTCHINSON, U.S. DEPARTMENT OF JUSTICE: We have been focusing on oil. Have you had any situations involving hazardous or noxious substances in the concept of jettison?
MR. ELLIS: Yes, during a lightening operation we were monitoring some nasty products being transferred from one vessel to another, because there was a problem with one vessel. It wasn't a situation where time was of the essence, but I certainly foresee cases where time will be of the essence. We should keep our minds open and say those circumstances might one day arise. Many salvors would say that oil is only one problem, and it is a relatively known quantity. There are lots of other substances being carried today that might present much greater problems.
UNIDENTIFIED PARTICIPANT: Have floating bladders ever been flown to a tanker for offloading?
MR. ELLIS: We have salvors in the room right here, and I would rather they answered. I can tell you that in one case in the north of Scotland they have actually been flown onto a deserted island, but we are talking there about 30 tons of bunkers. When we are talking about several thousand tons, of course, the problem is very different.
KLAAS REINIGERT, INTERNATIONAL SALVAGE UNION and SMIT TAK B.V.: They have never been used.
MR. ELLIS: That is a salvor telling us they have never been used.
NINA SANKOVITCH, NATURAL RESOURCES DEFENSE COUNCIL: About the Braer, you said you had difficulties in getting lighters to go there in time. Was that because the request for lighters wasn't made quickly enough? That the need for salvage wasn't recognized quickly enough? Or because the lighters or salvors were located far away?
MR. ELLIS: My understanding is that in the case of the Braer, the salvage contract was signed promptly. I believe the decision to deploy a barge would have been made almost immediately, but it was several hundred miles away and towing a barge in the weather conditions that were prevailing would have taken a great deal of time. I assume that other suitable barges weren't available. The condition was that the barge would have to anchor off, therefore you had to have a barge with a great deal of ground tackle to keep her in position.
ROBERT BUSH, UNIVERSE TANKSHIPS, INC.: Are there countries other than the United States that have laws about jettisoning?
MR. ELLIS: I am unaware of any, but my researches have not been exhaustive. In practice, though, when salvors have tried to jettison anything at all in
recent years there has been a great reluctance for even the most benign cargoes to be jettisoned. So, in practice there is already a problem.
QUESTIONS ADDRESSED TO DR. ENGELHARDT
RICHARD LEE, SKIDAWAY INSTITUTE OF OCEANOGRAPHY: You said in deep water, other than the example of dispersants, when oil is lost in the open ocean—deep subtidal—it is not a problem, it is only when it comes ashore.
DR. ENGELHARDT: As a generality, I think that is true, although shortly after the Braer grounded, predominantly mechanical dispersion due to extreme wave action was able to distribute the oil from the sea surface so that it was very broadly spread throughout—diluted, basically—through nonaffected areas. Toward the end of January there was some significant resurfacing of oil. Although the water depths there were relatively deep, the local circulation patterns seem to have maintained some sort of localized plumes of oil that then resurfaced. Whether that is going to be significant from an impact perspective requires determination. But the generality still holds that in that instance there was a lot of weathering of the oil and the proportional volume losses due to weathering were relatively high. So, from an impact perspective, dilution was an answer.
QUESTIONS ADDRESSED TO DR. GALT
RICHARD BROWN, KIRLIN, CAMPBELL, MEADOWS & KEATING: How do you get your information to the owner/operator? Is that through the on-scene coordinator?
DR. GALT: Yes, I work for the government during spills. NOAA provides assistance to the Coast Guard on-scene coordinator. As soon as a spill occurs we are assigned as technical staff.
CHARLES BOOKMAN, MARINE BOARD, NATIONAL RESEARCH COUNCIL: Captain Fullwood's scenario is a situation where somebody needs an answer and he needs it soon. In your experience, how well do you work within those kinds of constraints? He has a 24-hour window.
DR. GALT: Well, we can usually start talking soon. Again, we could come up with estimates but we would also have to talk about uncertainties. In a situation like that, there are uncertainties that make climatology a problem. A hurricane is a good example. This is a bit of a divergence, but when Hurricane Andrew came ashore, one of the platforms offshore broke loose, started to hop around and stepped on a pipeline, so there was a pipeline spill off Louisiana.
There is a convergence zone there. It has been there forever. We have seen it in lots and lots of spills—freshwater overlaying salt water. Hurricane Andrew obliterated it—just stirred the dickens out of it. So, here is a case where, if we had made the forecast based on climatology, it wouldn't have worked. However, the some hurricane dumped eight inches of rain in the lower Mississippi Delta so the fresh water was re-establishing itself in a hurry. I guess the answer is, we could give advice but I don't think we could say, go ahead and dump it, it is a good idea. We could say, here is where we think the threat is. Actually, the picture that emerged was very good in the following sense: if the spill had occurred with a jettison, the trajectory is considerably more certain because the wind would have blown straight ashore and it would have been all over from the surf and the waves and everything else.
If you were to say, "I have a new scenario—I am going to dump it, I want to look at it for the next couple of days," you will know the weather or at least you have
an idea what it is. If it is going to malinger for a month, you go back to climatology, which means that for a couple of days it will be one way, a couple of days it will be another way. What will happen is that a larger area is impacted. If you have a choice of which area it is and how to confine it spatially, that could easily be a determining factor. If a particular wind is going straight into a marine marsh, you might want to take a chance of spreading it out.