The reason you cannot predict it one month out is that at that lead time the weather mass come from far away and the factors of development there get involved with heat cells in the tropic rising to great height and rotating back down thru a far less stable system. Interactions with the upper atmosphere are much harder to track than air movements over-land. Over-sea movements of air masses are hard to factor, which is why hurricane landfalls are so hard to predict with exactness. Once air makes landfall, it one can use one KM cubic cells as a unit, then predictions of movement, addition of moisture, precipitation, speed, rotation, and turbulence can me made very very finely. This is sufficient to et 10 day ahead weather prediction. To get a start on it, which can be refined with statistics, you can use 100,000 farm and city amateur weather centers, plus the national grid of stations. You need wind direction, speed, humidity, dew point, sferics, cloud cover, temperature, barometric pressure, fed back once over 20 minutes. This sort of effort could probably harnessed a lot easier than the 1 million computer tactic needed to assess air-mass progress info.
Another approach is to use present weather data over a surrounding area. One runs stat programs on weather progress and uses modelling programs to run like weather forward with modern input from multiple sources. It is a fact that Buffalo area weather rotates into Northern Ontario within 2 -3 days and using their experience for prediction is better than environment Canada. Similar computer derived refinements could achieve better than standard weather-report predictions for a 5 to 7 day-out period.
EC<:-} |
