Climate Modeling MEA 719 Lecture Set 4 Predictability-Chaos-Lorenz Attractor

Main Topics
GENERAL BACKGROUND TO THE CONCEPT OF CHAOS CONTRIBUTION FROM ASTRONOMY APPLICATIONS TO METEOROLOGY LORENZS MATHEMATICAL MODEL ENSEMBLE CLIMATE PREDICTION (set-6) AN EXERCISE IN ENSEMBLE MODELING (set6)

GENERAL BACKGROUND TO THE CONCEPT OF CHAOS

"What is Chaos?"
In physics, chaos is a word with a specialized meaning, one that differs from the everyday use of the word To a physicist, the phrase "chaotic motion" really has nothing to with whether or not the motion of a physical system is frenzied or wild in appearance. In fact, a chaotic system can actually evolve in a way which appears smooth and ordered.

"What is Chaos?"
Rather, chaos refers to the issue of whether or not it is possible to make accurate long-term predictions about the behavior of the system For four centuries in physics, the laws of physics have reflected the complete connection between cause and effect in nature Thus until recently, it was assumed that it was always possible to make accurate long-term predictions of any physical system so long as one knows the starting conditions well enough.

"What is Chaos?"
The discovery of chaotic systems in nature about 100 years ago has all but destroyed that notion. Before that scientists and mathematicians believed in the Philosophy of Determinism This is the belief that says that every cause has a unique effect, and vice versa.

"What is Chaos?"
The Philosophy of Determinism imply that using the assumed link between cause and effect, the initial conditions are used to make predictions at later and earlier times To the contrary it is now believed, based on the theory of chaos, that no measurement can be made with infinite accuracy and ultimately the forecast becomes useless

"What is Chaos?"
Uncertainty of measurements give rise to Dynamical Instabilities, which to most physicists is a term synonymous with Chaos. Newton's laws are completely deterministic because they imply that anything that happens at any future time is completely determined by what happens now, and moreover that everything now was completely determined by what happened at any time in the past.

CONTRIBUTION FROM ASTRONOMY

"What is Chaos?"
The equations of motion for planets are an application of Newton's laws, and therefore completely deterministic. That these mathematical orbit equations are deterministic means, of course, that by knowing the initial conditions---in this case, the positions and velocities of the planets at a given starting time---you find out the positions and speeds of the planets at any time in the future or past

Limits of Predictability
It is impossible to actually measure the initial positions and speeds of the planets to infinite precision, even using perfect measuring instruments, since it is impossible to record any measurement to infinite precision. Thus there always exists an imprecision, however small, in all astronomical predictions made by the equation forms of Newton's laws Up until the time of Poincar, the lack of infinite precision in astronomical predictions was considered a minor problem, however, because of an incorrect assumption made by almost all physicists at that time.

Infinite Precision
In practical terms infinite precision may be interpreted to mean that the accuracy required to define the initial conditions may be much greater (by orders of magnitude) than the one required to observe and monitor the physical phenomena of interest

Limits of Predictability
The assumption was that if you could shrink the uncertainty in the initial conditions---perhaps by using finer measuring instruments---then any imprecision in the prediction would shrink in the same way. In other words, by putting more precise information into Newton's laws, you got more precise output for any later or earlier time. Thus it was assumed that it was theoretically possible to obtain nearly-perfect predictions for the behavior of any physical system.

Limits of Predictability
But Poincar noticed that certain astronomical systems did not seem to obey the rule that shrinking the initial conditions always shrank the final prediction in a corresponding way. By examining the mathematical equations, he found that although certain simple astronomical systems did indeed obey the "shrink-shrink" rule for initial conditions and final predictions, other systems did not.

Limits of Predictability
The astronomical systems which did not obey the rule typically consisted of three or more astronomical bodies with interaction between all three. For these types of systems, Poincar showed that a very tiny imprecision in the initial conditions would grow in time at an enormous rate. Thus two nearly-indistinguishable sets of initial conditions for the same system would result in two final predictions which differed vastly from each other.

Limits of Predictability
Poincar mathematically proved that this "blowing up" of tiny uncertainties in the initial conditions into enormous uncertainties in the final predictions remained even if the initial uncertainties were shrunk to smallest imaginable size. That is, for these systems, even if you could specify the initial measurements to a hundred times or a million times the precision, etc., the uncertainty for later or earlier times would not shrink, but remain huge.

Limits of Predictability
The extreme "sensitivity to initial conditions" mathematically present in the systems studied by Poincar has come to be called dynamical instability, or simply CHAOS. Because long-term mathematical predictions made for chaotic systems are no more accurate than random chance, the equations of motion can yield only short-term predictions with any degree of accuracy.

Examples in Real World Occurs in a large number of RETHYMIC SYSTEMS

Applied in e.g., arrhythmic pacemakers, fluid dynamics, etc the stock market provides trends which exhibit behavior of strange attractors a dripping faucet seems random to the untrained ear, but when plotted exhibits behavior of strange attractor

APPLICATIONS TO METEOROLOGY

Edward Lorenzs Pioneering Theory

One of the most important discoveries was made in 1963, by the meteorologist Edward Lorenz, who wrote a basic mathematical computer model to study a simplified model of the weather. Specifically Lorenz studied a primitive model of how an air current would rise and fall while being heated by the sun.

Lorenz Model
Lorenz's computer code contained the mathematical equations which governed the flow the air currents. Since computer code is truly deterministic, Lorenz expected that by inputing the same initial values, he would get exactly the same result when he ran the program. Lorenz was surprised to find, however, that when he input what he believed were the same initial values, he got a drastically different result each time.

Implications of Lorenzs Theory to Meteorology

Lorenz had not realized that the initial values for each run were different because the difference was incredibly small, so small as to be considered microscopic and insignificant by usual standards. Gradually it came to be known that even the smallest imaginable discrepancy between two sets of initial conditions would always result in a huge discrepancy at later.

Application of Chaos to Meteorology

Scientists now believe that like Lorenz's simple computer model of air currents, the weather as a whole is a chaotic system. This means that in order to make long-term weather or climate forecasts with any degree of accuracy at all, it would be necessary to take an infinite number of measurements. Even if it were possible to fill the entire atmosphere of the earth with an enormous array of measuring instruments---in this case thermometers, wind gauges, and barometers---uncertainty in the initial conditions would arise from the minute variations in measured values between each set of instruments in the array.

Butterfly Effect in the Atmosphere

Because the atmosphere is chaotic, these uncertainties, no matter how small, would eventually overwhelm any calculations and defeat the accuracy of the forecast. This principle is sometimes called the "Butterfly Effect." In terms of weather forecasts, the "Butterfly Effect" refers to the idea that whether/climate or not a butterfly flaps its wings in a certain part of the world can make the difference in whether or not a storm arises one year later on the other side of the world.

Butterfly Effect in the atmosphere

Because of the "Butterfly Effect," it is now accepted that weather/climate forecasts can be accurate only in the short-term, and that long-term forecasts, even made with the most sophisticated computer methods imaginable, will always be no better than guesses.

MATHEMATICAL MODELING OF LORENZS THEORY

What is Lorenz Attractor?

The so called "lorenz attractor" was first studied by Ed N. Lorenz, a meterologist, around 1963. It was derived from "Navier-Stokes" equations. The Lorenz model is defined by three nonlinear differential equations giving the time evolution of the variables X(t), Y(t), Z(t) dx / dt = a (y - x) dy / dt = x (b - z) - y dz / dt = xy - c z One commonly used set of constants is a = 10, b = 28, c = 8 / 3. Another is a = 28, b = 46.92, c = 4. "a" is sometimes known as the Prandtl number and "b" the Rayleigh number. The never reaches a steady state. Instead it is an example of deterministic chaos. As with other chaotic systems the Lorenz system is sensitive to the initial conditions, two initial states no matter how close will diverge.

Lorenz Attractor Independent Variables

In the context of the atmosphere, a is proportional to the temperature difference across the layer responsible for driving the fluid motion at a rate given by the variable X. Y and Z tell us about the changes in the temperature distribution in the layer due to the heat carried by the moving fluid

Lorenz Attractor Solution

Lorenz Attractor

Lorenz Attractor

Lorenz's Attractor and Avogadro's number

The attractor represents the behavior of gas at any given time, and its condition at any given time depends upon its condition at a previous time If the initial conditions are changed by even a tiny amount, say as tiny as the inverse of Avogadro's number (a heinously small number with an order of 1E-24), checking the attractor at a later time may yield numbers totally different. This is because small differences will propagate themselves recursively until numbers are entirely dissimilar to the original system with the original initial conditions.

Implications of chaos for climate variability & ensemble prediction

If you make a forecast based on model integration there is a chance that the error in the initial conditions is just enough to put the solution into the wrong attractor By performing an ensemble is forecasts starting from slightly different initial conditions within the bounds of error increases the chance to predict the correct butterfly wing of the strange attactor

Lorenz Attractor
Slight change in ICs ends up in a different wing

Lorenz Attractor

Lorenz Attractor
Slight change in ICs ends up in a same wing

Lorenz Attractor