Having a computer was enough of a novelty in 1958 to entice many MIT faculty and students to make the trip to Ed Lorenz's office just to watch the thing work. But excitement quickly turned to despair for Lorenz. Lorenz created a set of equations to act as a mathematical model of at mospheric storm movement and behavior. He noticed that tiny changes in the starting conditions of the model soon produced enormous changes in the out come.
Tiny starting differences always amplified over time, rather than damping, or normalizing out. If the actual atmosphere acted like Lorenz's models, he had just proved that long range weather forecasting was impossible since starting conditions were never known with enough precision to prevent chaotic, amplified error. It was an unset tling and sinking feeling to trade the excitement of finding a new research tool for the despair of proving that your field and work were both inherently flawed and impossible.
When Ed entered Dartmouth College in 1934, he had long ago made up his mind to be a mathematician. He graduated with a bachelor's degree in mathematics in 1938 and entered Harvard to continue his study of math. With the outbreak of World War II, Lorenz joined the Army Air Corps, who assigned him to attend army meteorology classes at MIT.
He learned to regard the weather as a combination of density, pressure, temperature, three-dimensional wind velocities, and the atmosphere's gaseous, liquid, and solid content.
The equations that describe this host of variables define the current weather conditions. The rates of change in these equations define the changing weather pattern. What Lorenz was not taught, and only much later discovered, was that no one knew how to use these nonlinear dynamic meteorology equations to actually predict weather and that most thought it could not be done. The equations were too com plex and required too much initial and boundary data.
Lorenz tried to apply the dynamic equations to predict the motion of storms. As computers were not commonly available in the early 1950s, most of this work was carried out on blackboards and with slide rules and paper and pencil. Each calculation was tediously time consuming. Lorenz was never able to reach any meaningful re sults while hand calculating these equations.
In 1958 Lorenz obtained that Royal McBee LGP-30 com puter (about the size of a large desk) to develop his sets of dynamic, nonlinear model equations. The results of those computer simulations showed that tiny initial differences amplified over time, rather than gradually nor malizing out. If the model was right, weather was chaotic and inherently unpredictable.
Several years of atmospheric testing convinced Lorenz and others in his department that he and his model were correct. The atmosphere was a chaotic rather than a predictable system (such as the system of interactions be tween inorganic chemicals, or the physical pull of gravity). A drive to use a new tool to complete an old project had turned into one of the most profound discoveries for the science of meteorology.
Lorenz will always be known as the person who discovered the true nature of the atmosphere and who thereby discovered the limits of accuracy of weather fore casting.
