OBIT: Edward Lorenz; Pioneer in Creation of Chaos Theory

[Zeno Swijtink]

Edward Lorenz; Pioneer in Creation of Chaos Theory
PATRICIA SULLIVAN, Staff Writer - Washington Post

https://www.washingtonpost.com/wp-dy...041603821.html

Edward N. Lorenz, 90, a meteorologist who laid the groundwork for chaos theory, memorably asking whether the flap of a butterfly's wings in Brazil can set off a tornado in Texas, died of cancer April 16 at his home in Cambridge, Mass. He was an emeritus professor at the Massachusetts Institute of Technology.

At MIT, Dr. Lorenz accidentally discovered how small differences in the early stages of a dynamic system, such as the weather, can trigger such huge changes in later stages that the result is unpredictable and essentially random.

At the time, Dr. Lorenz was studying why it's so hard to accurately forecast the weather, but the implications of his work go far beyond meteorology.

The new science of chaos fundamentally changed the way researchers address topics from the geometry of snowflakes to the predictability of which movies will become blockbusters. The butterfly effect became a popular way of describing unpredictability, most recently in "An Inconvenient Truth" (2006), the Academy Award-winning documentary with former Vice President Al Gore.

It also "brought about one of the most dramatic changes in mankind's view of nature since Sir Isaac Newton," said the committee that awarded Dr. Lorenz the 1991 Kyoto Prize for basic sciences.

Yet Dr. Lorenz's 1962 paper on chaos theory was largely ignored for years. A decade later, when he gave a talk about predictability, with a title asking the famous butterfly question, the scientific establishment was ready to consider the idea. Other scientists who had been working on similar questions swarmed to the field, and one by one, certain assumptions of science began to falter.

"When I first heard this [butterfly effect] idea, I thought it very clever but it couldn't be literally true," said James Gleick, a science writer and author of "Chaos: Making a New Science" (1987), which explored Dr. Lorenz's work. "But it is literally true. . . . Complex dynamical systems, if they are chaotic, never repeat themselves. They are capable of an infinite variety of behavior."

This means that simple systems can result in complex behavior and that the slightest change in underlying causes can make the result unpredictable.

Chaos theory -- also known as the science of nonlinearity, the science of complexity, the science of random recurrent behavior or the science of turbulence and discord -- has thus been called the third great scientific revolution of the 20th century, along with relativity and quantum physics.

Edward Norton Lorenz was born May 23, 1917, in West Hartford, Conn., and graduated from Dartmouth College. He received a master's degree in mathematics in 1940 from Harvard University and served as a weather forecaster for the Army Air Forces during World War II.

In 1948, he received a doctorate in meteorology from MIT and joined its faculty. He remained there the rest of his career.

In 1961, he was using a primitive computer to model weather forecasts, which led to his most renowned work.

Using 12 equations, such as the relationship between air pressure and wind speed, he ran the model and found exactly what he sought. But taking a shortcut on the next run, he found that a tiny decimal point change led to a significant error.

Rather than ignore the response, which peers had considered an anomaly, Dr. Lorenz realized measurement is not perfect. If temperature, pressure or humidity measurements were off by a hundredth of a percent, the rainfall he expected in Las Vegas on Thursday could show up as a snowstorm in Beijing a week later. A computerized model of how the "butterfly effect" works can be found at the Exploratorium's Web site.

"He had the ability to make important connections between atmospheric phenomena and simple theoretical models," said Isaac Held, senior research scientist in the National Oceanic and Atmospheric Administration's Geophysical Fluid Dynamics Laboratory. "He taught us that complexity can follow from very simple underlying rules. His study of the limits to the predictability of weather initiated an entire new field of chaotic dynamics."

Dr. Lorenz was elected to the National Academy of Sciences in 1975, and in 1983, he and oceanographer Henry Stommel were jointly awarded the $50,000 Crafoord Prize by the Royal Swedish Academy of Sciences, an honor established to recognize fields not eligible for Nobel Prizes.

Dr. Lorenz was known as shy and humble. He enjoyed hiking and cross-country skiing and sought out mountain trails near every scientific meeting he attended, one of his daughters said.

His wife, Jane Logan Lorenz, died in 2001.

Survivors include three children, Nancy Lorenz of Roslindale, Mass., Edward Lorenz of Grasse, France, and Cheryl Lorenz of Eugene, Ore.; and four grandchildren.

Chaos theory helped shape Dr. Lorenz's conclusions as he worked to calculate long-term forecasts. Given the problems of input, the meteorologist determined that it's impossible to accurately predict weather beyond two or three weeks with a reasonable degree of accuracy.

[Zeno Swijtink]

Butterflies, tornadoes and climate modeling

https://www.realclimate.org/index.ph...ng/fr#more-555


Classé dans: Climate modelling Climate Science — group @ 8:35 AM

Many of you will have seen the obituaries (MIT, NYT) for Ed Lorenz, who died a short time ago. Lorenz is most famous scientifically for discovering the exquisite sensitivity to initial conditions (i.e. chaos) in a simple model of fluid convection, which serves as an archetype for the weather prediction problem. He is most famous outside science for the 'The Butterfly Effect' described in his 1972 paper "Predictability: Does the Flap of a Butterfly's Wings in Brazil Set Off a Tornado in Texas?". Lorenz's contributions to both atmospheric science and the mathematics of dynamical systems were wide ranging and seminal. He also directly touched the lives of many of us here at RealClimate, and both his wisdom, and quiet personal charm will be sorely missed.


Ed Lorenz had a reputation of being shy and quiet, and this was indeed the impression he gave on first meeting. Indeed raypierre was interviewed by Ed at MIT in 1979 for his first faculty job — and remembers having to ask most of the questions as well as answer them. But he also remembers a lot of timely support from Ed that helped smooth over the somewhat rocky transition from basic turbulence theory to atmospheric science. The longer you were around Ed, the more you came to appreciate his warmth and sense of humor. He was an avid hiker, and many in the community (our own Mike Mann included) have recollections of time on the trail with him around the hills of Boulder and elsewhere.

Lorenz launched the modern era of the study of chaotic systems, which has profound implications both within and beyond atmospheric science. We'll say more about that in a bit, but the monumental work on chaos should not leave Lorenz's other contributions to atmospheric science completely in its shadow. For example, in a 1956 MIT technical report, Ed introduced the notion of "empirical orthogonal functions" to atmospheric science, and this technique now plays a central role in diagnostic studies of the atmosphere-ocean system. He also pioneered the study of angular momentum transport in the atmosphere, and of atmospheric energetics. Among other things, he introduced the important notion of "available potential energy," which quantifies the fact that not all of the potential energy can be tapped by allowable rearrangements of the atmosphere. Later, he pioneered the concept of resonant triad instability of atmospheric waves, an idea that has repercussions for the sources of atmospheric low frequency variability. As if that weren't enough Ed also introduced the concept of the "slow manifold" — a special subset of solutions to a nonlinear system which evolve more slowly than most solutions. The atmospheric equations support a lot of very quickly changing solutions, like sound waves and gravity waves, but on the whole what we think of as "weather" or "climate" involves more ponderous motions evolving on time scales of days to years. Ed's work on this subject launched the study of how such slowly evolving solutions can exist, and how to initialize a numerical model so as to minimize the generation of the fast transients. This is now part and parcel of the whole apparatus of data assimilation and numerical weather forecasting.

Ed was not a user of general circulation models. His essential approach was to crystallize profound phenomena into very small sets of equations for how a handful of variables change with time. He left behind him a dozen or so such models, each of which would repay many lifetimes of study. He was indeed a master of "seeing the world in a grain of sand." You can read about some of these models in the talk raypierre gave at the 1987 Lorenz 'retirement' symposium — not that this slowed him down!

Now let's take a closer look at that butterfly effect. Despite the fact that there are no butterflies or tornadoes in climate models, Lorenz's discoveries and their implications played a central role in climate modelling efforts and in the most recent IPCC report.

The notion of the butterfly effect itself was drawn from a simple but astute observation of the way the solutions of certain nonlinear equations behave when they are solved using a computer. Start with a greatly simplified representation of thermal convection, first formulated by Barry Saltzman using a technique called "low order modelling." If you run a simulation using these equations and then try and replicate it using starting values that only differ in the last decimal place, you will find that the simulations quickly diverge from one another - and by quickly, it means that the differences grow exponentially fast. Lorenz found this phenomenon by accident, but quickly recognised the profound implications. If the real weather system displayed the same behaviour, it meant that since however well one knew the initial conditions of the atmosphere, there would always be some uncertainty, that uncertainty would be quickly magnified, rendering weather forecasts useless after a few exponential doubling times. The practical implication is that - even if you had a perfect model - for every halving of the error in the initial conditions you only get one extra time period of useful forecast. Given this time period is only a few hours in many cases, the practicality of true weather forecasts for periods longer than two weeks or so, is vanishingly small.

The mathematically inclined reader who takes a look at Ed's early papers on what is now called the "Lorenz Attractor" will be astonished at the depth and modernity of his ideas about chaos. This line of work was no mere remark on a numerical exercise. Lorenz actually teased out the geometry of chaos — the many-leaved structure of the attractor — realizing that it was no simple geometric entity like a sphere or a folded sheet of paper. It was indeed "strange" in a sense which he made geometrically precise. This is why the work had such lasting impact on the area of pure mathematics known as dynamical systems theory. He went beyond that to develop or apply many fundamental concepts in chaotic systems, quantitatively formulating various measures of predictability and connecting the Lyapunov exponent — a certain precise mathematical characterization of chaos — with the structure of strange attractors. But that's for the mathematicians. What makes Lorenz's work interesting to the entity on the Clapham omnibus is the notion of sensitive dependence on initial conditions. Some have even seen in this deterministic chaos the resolution to the problem of free will!

The idea that small causes can have large and disproportionate effects is not at all new of course. The poem: "For the want of a nail, the battle was lost" (medieval in origin) encapsulates that well, and popular culture is full of such examples, "It's a wonderful life" (1946) and Ray Bradbury's "A Sound of Thunder" (1952) for instance. Curiously, Bradbury's story also involves a butterfly, and since it predates Lorenz's coining of the phrase by a decade or so, people have speculated that there was a connection to Lorenz's choice of metaphor (he started off with a seagull in his 1963 Trans. N.Y. Ac. Sci. paper). But that doesn't appear to be the case (see here for a history). It's worth adding that all of Lorenz's papers were exceptional in their clarity and are well worth tracking down as an example of science writing at its best.

However, the difference between the long-standing popular conception and Lorenz's work is that he demonstrated this effect in a completely deterministic system with no random component. That is, even in a perfect model situation, useful predictability can be strongly limited. Strictly speaking, Poincaré first described this effect in the classic three-body problem in the early 1900s, but it was only with the onset of electronic computers, as used by Lorenz, that this became more widely recognised. To throw in another popular culture reference, Tom Stoppard's Arcadia has a character, Septimus, who drives himself mad trying to calculate chaotic solutions to the logistic map by hand.

So what does this have to do with the IPCC?

Even though the model used by Lorenz was very simple (just three variables and three equations), the same sensitivity to initial conditions is seen in all weather and climate models and is a ubiquitous phenomenon in many complex non-linear flows. It is therefore usually assumed that the real atmosphere also has this property. However, as Lorenz himself acknowledged in 1972, this is not directly provable (and indeed, at least one meteorologist doesn't think it does even though most everyone else does). Its existence in climate models is nonetheless easily demonstratable.

But how can climate be predictable if weather is chaotic? The trick lies in the statistics. In those same models that demonstrate the extreme sensitivity to initial conditions, it turns out that the long term means and other moments are stable. This is equivalent to the 'butterfly' pattern seen in the figure above being statistically independent of how you started the calculation. The lobes and their relative position don't change if you run the model long enough. Climate change then is equivalent seeing how the structure changes, while not being too concerned about the specific trajectory you are on.

Another way of saying it is that for the climate problem, the weather (or the individual trajectory) is the noise. If you are trying to find the common signal that is a signature of a particular forcing then averaging over a number of simulations with different weather works rather well. (There is a long standing quote in science - "one person's noise is another person's signal" which is certainly apropos here. Climate modellers don't average over ensemble members because they think that weather isn't important, they do it because it gives robust estimates of the signal they are usually looking for.)

The ensemble approach, and indeed the multi-model ensemble approach, used in IPCC then derives directly from Lorenz's insights into his serendipitous numerical problem.