Network News

X My Profile
View More Activity
The new Washington
Post Weather website
Jump to CWG's
Latest Full Forecast
Outside now? Radar, temps
and more: Weather Wall
Follow us on Twitter (@capitalweather) and become a fan on Facebook
Posted at 10:50 AM ET, 06/10/2010

Why is oil spill hard to predict? Look to the clouds.

By Steve Tracton

* Great days to end week: Full Forecast | NatCast *

Oil pools along Barataria Bay on the Louisiana coast. By Charlie Riedel-AP.

The BP Gulf oil spill has become the most recent example of a "Black Swan" -- i.e., an improbable, unpredicted event having tremendous impacts.

Retrospective analysis might demonstrate the Deepwater Horizon explosion and resulting massive flow of oil into the Gulf was a relatively likely "accident" waiting to happen. Without benefit of hindsight, though, it's been an event totally unanticipated with far-reaching consequences and one due to human failings, not an unavoidable natural disaster (e.g., earthquake, hurricane). Avoidable or not, we are in the position now of having to manage to the extent possible what heretofore was an unimaginable crisis.

The ability to manage the crisis is dependent upon at least three fundamental questions: Exactly how much oil has spilled? Where is the oil located? And where will it go? Unfortunately, there appears to be considerable uncertainty on answers to the first two questions, while the third is probably unanswerable beyond a few days in advance.

A reasonable analogy is predicting the development and motion of a cloud where imperfect measurements of its dimensions, location and background wind field are magnified over time by imperfect forecast models. To extend this analogy consider the following...

Oil: A column of oil gushing from the bottom of the Gulf reaches the surface where it spreads out more or less uniformly as a thin sheen over an increasingly large area.

Cloud: A plume of rising moist air reaches the level of condensation (beneath a stable layer of air that prevents it from rising further) where the newly formed cloud droplets spread out as a thin canopy over an increasingly large expanse of the sky.

Oil: Over time, wind, waves and local surface currents begin to break up the oil into an array of congealed lumps of varying size and thickness. These lumps tend to spread apart and organize non-uniformly into clusters and long streaks (streamers) as functions of the variability in the strength and direction of surface winds and currents, surface wave characteristics (height, period, etc.), degree of mixing with sea water, and vertical temperature and salinity structure of the underlying water column.
Cloud: Over time, the stable layer of air destabilizes (allowing the moist air to penetrate higher) with the original thin cloud being replaced by puffy cumulus clouds of varying horizontal and vertical dimensions. The individual clouds tend to organize into cloud clusters and lines of clouds spreading across the sky non-uniformly as functions of the variability in the strength and direction of winds, degree of mixing between cloud and surrounding air, and vertical stability of the air mass.

Oil: At least some oil will come ashore unexpectedly in the concentrated form of, for example, tar balls and mousse at the whim of winds, waves and currents. What might otherwise have been a great day for the beach or exploring the wetlands along the Gulf coast becomes a panicky rush to clean up and/or leave.

Cloud: Some cloud clusters unexpectedly develop into thunderstorms at the whim of local atmospheric conditions. What otherwise was a great day to be outdoors becomes a panicky rush to pack up and seek cover.

(I realize the analogy breaks down when considering that the long-term danger and after-effects of thunderstorms most certainly are not comparable to the long-term, massive, far-reaching impacts of the oil spill. But, please stay with me.)

Oil: Eventually there is a good chance the oil spill will encounter the Loop Current, which would carry it and related consequences around (or between) the Florida Keys and be picked up by the Gulf Stream.
Cloud: There's a good chance the further development and motion of the cloud clusters will be governed by the jet stream, which propagates them and associated weather far beyond the area of initial development.

Oil and cloud prediction

A clean-up worker picks up blobs of oil in absorbent snare on Queen Bess Island at the mouth of Barataria Bay near the Gulf of Mexico in Plaquemines Parish. AP Photo/Gerald Herbert.

Predicting where the oil spill and its component parts will go is a challenge no less difficult, and probably even more difficult, than predicting clouds and overall weather conditions.

Oil on the surface is moved by both winds and currents. Specialized computer models, such as those used to help produce NOAA's oil trajectory maps, account for the influence of both by combining wind forecasts from weather models with current forecasts from ocean models. Not surprisingly, the predictability of oil spill movement is limited, since all weather and ocean models have errors that grow with time for a variety of reasons (e.g., initial condition uncertainties, imperfect models).

Forecasts of near-surface winds are generally not reliable beyond several days ahead (say, a week or so), as is true for all parameters within any weather forecast model, including those relevant to cloud development and motion. Even if winds were accurately predicted, wind-driven waves can disperse and spread the oil in directions that differ unpredictably from the wind. Moreover, observations are evidently lacking to precisely define the size and, especially, the small-scale structure of the oil spill. The latter is critical for predicting details of the location and timing of oil reaching vulnerable spots ashore -- not unlike difficulties in predicting the timing and location of thunderstorms.

While wind and waves play a critical role in details of oil movement, it's likely that large, strong currents that evolve relatively slowly over time, most notably the Loop Current and Gulf Stream, will ultimately govern how much of the oil is carried elsewhere from the Gulf. To the best of my knowledge ocean models, which suffer from an extreme scarcity of the observations necessary to define the initial ocean state and from imperfect representation of relevant dynamical and physical processes, have limited skill beyond a few days in predicting the evolution of such currents.

Next stops for oil?

The biggest threat in the shorter term is for the states of Louisiana, Mississippi and Alabama. The oil could also spread as a function of changes in wind speed and direction eastward toward the west coast of Florida and/or westward to the Texas coast. Eventually, it's not unlikely some oil will exit the Gulf of Mexico and move northward along the Atlantic coast and beyond. The least worst-case scenario would be for the oil to continue gushing into the Gulf for a prolonged period such that odds increase dramatically that all these areas would be affected. Even worse would be that the oil results in the Gulf of Mexico and much of the western Atlantic becoming dead zones for virtually all marine life.

Presumably everything possible will be done to avoid such worst-cast disasters, but failure in these efforts and resulting catastrophic consequences are not outside the realm of possibilities.

By Steve Tracton  | June 10, 2010; 10:50 AM ET
Categories:  Tracton  
Save & Share:  Send E-mail   Facebook   Twitter   Digg   Yahoo Buzz   StumbleUpon   Technorati   Google Buzz   Previous: Forecast: Sun, low humidity deliver two great days
Next: PM Update: Low-humidity warmth presses on


It's a big Finite Element Analysis problem, and we have supercomputers that could render it and give a good stab at predicting where it goes - but only if we had the initial conditions and a good detailed map of all the changing ocean currents. I'm not sure we have either of those things though?

Posted by: B2O2 | June 10, 2010 12:42 PM | Report abuse


It's highly questionable how much further ahead we could predict the details of where the oil will go even with better initial data and models. It's those pesky butterflies again.

Posted by: SteveT-CapitalWeatherGang | June 10, 2010 3:34 PM | Report abuse

I doubt chaos theory will ever be eliminated ...sigh... those pesky flows of 3-dimensional liquid systems will remain challenging to predict. Great article Steve!

Posted by: Camden-CapitalWeatherGang | June 10, 2010 4:10 PM | Report abuse

i have enough confidence that IF we knew all the variables and IF we had really giant, giant computers we could predict the oil spill's path. of course we'll never be able to quantify ALL the variables. (like steve said...damn butterflies...)

Posted by: walter-in-fallschurch | June 10, 2010 9:19 PM | Report abuse

Suppose we perfectly observe the full 3D system today, except that temperatures are only known to within a millionth of a degree. Suppose we also have giant computers and our models compute every physical process perfectly. Then as we forecast, that tiny temperature error will grow and compound itself so that there would be pretty much zero determinsitic predictability left beyond 30 days. Steve shows our actual predictability of the oil seems to be only on the order of a few days at best.

But not all hope is lost. We can run an ensemble of model forecasts with a spread of initial conditions representing the observational uncertainty and a spread of different models representing the process uncertainty. The forecasts will diverge and give a "spaghetti" cloud of possible trajectories. So we can get useful probabilistic predictability beyond when single-valued deterministic forecasts fail. One of the pioneers of this ensemble technique is Steve Tracton.

Posted by: imback | June 10, 2010 11:28 PM | Report abuse

There is nothing improbable or unpredictable about this oil spill... there are other wells also leaking in the gulf.... BP was, as usual - unprepared!

Relearn (learn) just how many perfidious acts BP has been caught out in so far! Then ask yourself, "why are we entrusting the fixing of this entire catastrophe to them?"

BP is ‘doing everything it can’ (so it tells us) to stem the humongous geyser of oil, gushing from the severed pipe, beneath what once was the Deepwater Horizon Oil Rig in the Gulf of Mexico. The rig exploded and sank, on April 20, 2010 killing 11 workers and injuring 17 others.

Posted by: justmeint | June 11, 2010 5:44 AM | Report abuse

here's BPs response to another spill:

Posted by: walter-in-fallschurch | June 11, 2010 9:23 AM | Report abuse

Dumping dirt (sand, clay and earth dirt contains large spectrum of bacteria) over the rising oil from the leaking pipe at the bottom of the gulf.
My son Sami did not finish high school; he won the trophy for the best martial art player in the province of Ontario. My son said that if he was working his car engine and his hand got dirty with oil and even if water and soap are available, my son prefers to use dirt from the ground (sand and clay) and rub his hand with the dirt. It is the best to remove oil from your hand. Why not dump load of ground dirt (soil consisting of sand and clay and earth dirt) onto the site where the oil is coming out from the bottom of the Gulf of Mexico. I mean 4 to 5 ship load every day. It is cheaper, Greener, and very effective. The rising crude oil will mix with the sand and dirt and settle to the bottom. It is confirmed that the under sea plumes are in anaerobic conditions ( no oxygen) , and thus this will be a heaven for anaerobic bacteria that can in the absence of oxygen eat everything’s and a good source of these bacteria comes from dirt. Later when the rising oil leak will be stopped, the mixture of sand and oil can be vacuumed from the bottom and treated accordingly.

Posted by: corrosion555 | June 11, 2010 9:35 AM | Report abuse


Yes, ensemble predictions are the way to go, as I have so often reported and supported. Accounting for uncertainty in any forecast is essential to convey all relevant information. If you go to the NOAA trajectory site referenced, you'll see that these precitictions are ensemble forecasts - note uncertainty boundary in forecast charts.

But, the reliability of ensemble predictions (probabilities) is limited by model capabilities and ensemble strategy. An ensemble cannot provide useful information on what the model cannot predict.

Posted by: SteveT-CapitalWeatherGang | June 11, 2010 9:40 AM | Report abuse

Re: What really caused the 2010 oil spill ?
Kick is the cause of the explosion in the Gulf of Mexico. President Obama is right when he said
"We talk to these folks because they potentially have the best answers, so I know whose ass to kick". The word Kick is very scientific because the reason for the oil spill in the Gulf of Mexico is a Kick.
It is well known in the oil industry and the oil companies are well aware that:
When drilling in oil and gas-bearing formations submerged in deep water, the reservoir gas ( mainly methane) may flow into the well bore and form gas hydrates due to the low temperatures and high pressures found during deep water drilling( similar to the situation in the Deep water Horizon Oil Spill). The gas hydrates may then flow upward with drilling mud or other discharged fluids. As they rise, the pressure in the drill string decreases and the hydrates dissociate into gas and water. The rapid gas expansion ejects fluid from the well, reducing the pressure further, which leads to more hydrate dissociation and further fluid ejection. The resulting violent expulsion of fluid from the drill string is referred to as a "kick" (see. Petroleum Science 6: 57-63, 2009). This could possibly be the reason for the accidents which caused the oil spill in the Gulf of Mexico. Thus president Obama had indirectly identified the culprit that caused the oil spill.

Posted by: corrosion555 | June 11, 2010 10:09 AM | Report abuse

The oil spill can be considered as a vaccine for the marine ecosystem.
Now that the oil spill is her and it is continuing to spread everywhere and reach regions that are very sensitive ecologically we are faced with a disaster. But let us look at the good thing about the oil spill. This disaster actually was one good driving force for evolution. For example the oil that we use as energy comes from the dinosaurs that vanished for unknown reasons and decomposed below the ground to give us oil. This oil drives our car, make all kind of plastic and keep our house warm. The vanishing of dinosaurs was a disaster similar to the Gulf of Mexico oil spill. This disaster will be the driving force for evolution since the spreading oil in the marine environment will produce a new species more resistant to toxic chemicals and kill the weak species while the strong more resistant species will propagate. The oil spill can be considered as a vaccine to the marine ecosystem. This is because a vaccine is basically the bacteria of the disease injected to the human body to produce resistance against the disease.

Posted by: corrosion555 | June 11, 2010 12:07 PM | Report abuse

In my humble opinion the oil splill was NOT an accident but criminal malfeasance on the part of BP. It took every shortcut possible to save time and money with clear knowledge that there was a non zero chance of the disaster we now see unfolding. BP gambled and WE lost.

To me this is no different when some hot-shot knowingly runs a red light to meet some perceived deadline that might add to his/her bottom line and kills some innocent victim in the process. That too is not an accident, but a criminal act.

Posted by: SteveT-CapitalWeatherGang | June 11, 2010 12:11 PM | Report abuse

you go steve!

indeed - we've all heard about the construction short cuts and the lax blow-out preventor inspections etc.... while you and i are outraged, and the authorities have been making noise about criminal prosecution and so forth, i really, really doubt it will come to that. i would LOVE to see it, but i doubt it will happen.

Posted by: walter-in-fallschurch | June 11, 2010 12:19 PM | Report abuse

What if they cannot stop this leak? That would be a condition of "runaway", according to this author: - Ch. 21 of his book "I.T. WARS" is a real eye-opener too.

Posted by: janice33rpm | June 12, 2010 2:07 PM | Report abuse


Thanks for the link. A relevant and exceedingly important quote from it is worth posting here:

" In looking at certain outcomes from disasters, we can recognize that prevention is not some part of a Disaster Recovery Plan, or Business Continuity Plan – it is the goal and the whole of it. To further illustrate what we mean: During the Cold War between the old Soviet Union and the U.S., a defacto policy of MAD - Mutually Assured Destruction - held a nuclear exchange and total destruction at bay. There’s not likely much of a recovery plan post-apocalypse. Prevention was the goal and indeed whole of the plan – the great driving motivator that influenced all subsequent activity."

Though a dramatic example, for some things mankind has control over, the consequences of mishaps through accident, misjudgment, benign or criminal neglect, avoiding dire consequences must be - the sole overriding criteria for guiding action - even if that requires hard and fast - and enforceable and enforced - government regulations.

Posted by: SteveT-CapitalWeatherGang | June 12, 2010 3:40 PM | Report abuse

the U.S. Coast Guard has given to BP only 48 hours to find the right way to fix the oil spill's problem!!!
then I've developed a new and better "super cap" that can be built and installed in a couple of days!!!
read a description and see a drawing of my new idea in my article latest update:

Posted by: www999 | June 13, 2010 3:39 AM | Report abuse

The comments to this entry are closed.

RSS Feed
Subscribe to The Post

© 2012 The Washington Post Company