Embedded Files
Next, the computer asked him for the desired length of the forecast. Frank, feeling ambitious, requested a forecast for the next two days. Still the weather computer wanted more information. Its next request was for the temperature at 1 foot and 20 feet for the next two days! Frank was flabbergasted. If he could forecast the temperature at 1 foot and 20 feet, he wouldn't need a computer to forecast it for him at 4 feet!
So he figured something was wrong, and he looked at the program the computer was going to use to forecast the temperature. He found that the formula the computer was using looked something like:
The NEW TEMPERATURE is just like the OLD TEMPERATURE, except for subtracting a TINY AMOUNT to allow for radiative cooling of the air, and also allowing for a small change to account for a little bit of TURBULENT MIXING of air from above and below.
"Well", he said, "I don't care about turbulent mixing at all. I'll just take out that part of the program. Then I won't need to give the computer those other temperatures." So he took out the turbulent mixing part of the program and tried running it again. The weather computer was happy and didn't ask him any more questions. The program ran (quite quickly, I might add), and its output consisted of forecasted temperatures for every hour, on the hour.
The computer made its forecast through a process called "iteration". It made a series of tiny forecasts, using the current conditions to estimate how fast the conditions must be changing, let them change a little bit, and then estimated based on the new current conditions how fast the conditions must be changing, etc., etc.
Frank was pleased until he took a look at those temperatures. Then he saw that the forecast was lousy! The forecasted temperature just dropped very slowly, about one degree the first day, another degree the second, and so forth. Nights were no different from days.
What went wrong?
Report abuse