at evenly spaced points of a tidy grid extending across oceans and up into the
atmosphere. Weather stations are rare in the middle of the ocean or in tropical
jungles. Satellites help fill in the gaps, but their data aren’t always reliable: For
example, they infer wind speed from cloud motion, which requires guessing the
height of the clouds. And thermometers and wind gauges sometimes break.
This is a real nuisance for meteorologists, because to predict the comingweather, they really need to know the current weather everywhere. After all, the
seeds of the next hurricane get hatched in some unmonitored corner of the
Atlantic, not in the middle of New York City bristling with thermometers.
Figure. 8. Weather monitoring stations in Europe are located according to funding and local
interest. Mathematicians are creating ever-improved methods to infer the temperature at each of
the models’ grid points from the messy spread of real-world data. Reprinted from “A European
daily high-resolution gridded data set of surface temperature and precipitation for 1950 – 2006,”
Journal of Geophysical Research, vol. 113, Oct. 30 2008, with permission from American
Geophysical Union.
Fortunately, forecasters have another source of information about thestate of the weather, one perfectly tuned to their needs. that gives the current
state precisely at every grid point: the forecasts themselves. Six-hour forecasts
are remarkably good, usually predicting temperatures accurately within half a