assumes that the same pattern this week will lead to the same weather. There are lots of
possible variations in patterns and changes often occur, so this method is also not entirely
accurate.
Numerical Weather Prediction
The most accurate weather forecasts are made by advanced computers, with analysis and
interpretation added by experienced meteorologists. These computers have up-to-date math-
ematical models that can use much more data and make many more calculations than would
ever be possible by scientists working with just maps and calculators. Meteorologists can
use these results to give much more accurate weather forecasts and climate predictions.
In Numerical Weather Prediction (NWP), atmospheric data from many sources are plugged
intosupercomputersrunningcomplexmathematicalmodels. Themodelsthencalculatewhat
will happen over time at various altitudes for a grid of evenly spaced locations. The grid
points are usually between 10 and 200 kilometers apart. Using the results calculated by the
model, the program projects that weather further into the future. It then uses these results
to project the weather still further into the future and so on, as far as the meteorologists
want to go. The final forecast is called aprognostic chartorprog.
Certain types of progs are better at particular types of forecasts and experienced meteo-
rologists know which to use to predict different types of weather. In addition to the prog,
scientists use the other forecasting methods mentioned above. With so much data available,
meteorologists use a computerized system for processing, storage, display and telecommuni-
cations. Once a forecast is made, it is broadcast by satellites to more than 1,000 sites around
the world.
NWP produces the most accurate weather forecasts, but as anyone knows, even the best
forecasts are not always right. Some of the reasons for this are listed below:
- Not enough data was initially entered into the program. This is most likely to happen
for a region near an ocean or a remote area. - Thecomputerprogrammakescertainassumptionsabouthowtheatmosphereoperates,
which may not always be correct. - The programs only deal with weather locally, which means errors are likely at the edges
of the area studied. A global model would be more accurate, but producing one would
require an incredible number of calculations. - The weather system may be too small to show up on the grid. If a system is small, like
a thunderstorm, it will not be modeled. If distances between grid points are reduced,
many more calculations and therefore more powerful computers are needed. - There is always the possibility that conditions change unpredictably. Weather is a
chaotic system and small, unpredictable things always happen. The farther into the
future a model tries to forecast, the more unpredictable things arise to change the
forecast.