His father was a bit miffed that the forecast turned out to be even better than the National Weather Service forecast, but he said, "You make ten forecasts, and I'll make ten forecasts. And we'll see how accurate the weather computer is." So they did, and sure enough, the National Weather Service forecast was better eight times out of ten.
Frank, though, had faith in his weather computer. He noticed that the computer forecast was especially poor when it was cloudy or raining, and he realized that the main thing he needed to add to his computer model was moisture. The equation for that proved to be fairly simple, too:
The NEW MOISTURE AMOUNT is equal to the OLD MOISTURE AMOUNT, with a small correction for moisture being TRANSPORTED into the grid box, and an adjustment for turbulent MIXING, and a bit subtracted to get down to 100% relative humidity if the moisture amount ever got above that. (The amount subtracted would fall as RAIN.)
The moisture also affected the heating of the Earth's surface, by blocking some of the incoming solar radiation and by keeping the Earth warmer at night.
With the new moisture variable, the weather computer was forecasting all the important weather parameters. And the forecasts improved, so that he beat the National Weather Service three or four times out of ten. And still there were things that could be added: a correction for thunderstorm activity, or keeping track of cloud water, or improving the horizontal or vertical resolution. But still he knew that there was one aspect of the forecast that neither he nor the National Weather Service could handle: the temperature on the sixteenth green dropped by at least five degrees whenever they turned on the sprinklers.