Frank seemed to be stuck. He knew he needed to include the effect of turbulence. But he didn't know how to predict the temperature at the other levels any more than he knew how to predict the temperature at four feet. But then he realized, "If the weather computer can make a forecast for the four foot level, why can't I tell it to make its own forecasts for one foot and twenty feet too!"
Frank did this, but discovered to his dismay that the weather computer now wanted the temperature at 50 feet. If Frank told the computer to forecast for 50 feet, it wanted the temperature at 100 ft, and so on.
This form of information is known as a "boundary condition". No forecast models simulate the entire universe, so they all need boundary conditions to specify what's happening outside the domain of the model. In this case, the weather computer wants to know the temperature at 100 ft, for example, so that it can predict how mixing of air from the 100 ft level will affect the temperature at 50 feet.
Frank knew enough about meteorology to realize that as you go higher in the atmosphere, the temperature doesn't change much from hour to hour. So he configured the model so that it would forecast the temperatures up to about a mile in altitude. He attached a very long string to his balloon and got a telescope so he could determine the initial temperature of the atmosphere all the way up to a mile, and told the computer to simply assume that the current temperature one mile up would stay constant throughout the forecast.
As for the other boundary, what other information does the weather computer need to forecast temperatures right at ground level? Check all that apply.
Frank decided to have the program forecast the temperature of the ground, the air at 1 foot, 4 feet, 20 feet, and every 50 feet above that all the way up to a mile.
For starting temperatures, Frank got a meteorology textbook and figured out what typical air temperatures were at all the various levels. The ones near the ground he changed a bit so that they matched the actual temperature, which he had measured.
Fortunately for Frank, the computer had its own soil temperature and moisture simulation package, which had default values already specified for most of the variables. All Frank had to do was tell it the clouds, the time of day, and the time of year. Since the actual surface characteristics undoubtedly differ from the default values, the computer's forecast will probably not be quite right. But by this time, Frank was a tad impatient.
Finally, Frank set the weather computer running. And that did it! The computer had all the information it needed to forecast the temperature in that entire vertical column of air. And one of the results was a forecast of temperature at the four-foot level - just what he wanted.