NOT ALL CHAOS ON THE WEATHER MAP IS EQUAL,
researchers have found, providing insights that are hoped to
improve weather forecasting. Researchers usually assume that all
spots on a weather map are equally chaotic, meaning that small
uncertainties in initial conditions grow to the point at which the
conditions become unpredictable. Now, a multidisciplinary
University of Maryland team of meteorologists, physicists, and
computer scientists (DJ Patil, 301-405-4842, dpatil@ipst.umd.edu)
has developed a technique that identifies what can be considered as
chaos "hotspots," regions in which small changes in conditions are
believed to magnify most quickly into large perturbations in the
weather. Chaos hotspots shift their location on a regular basis, but
tend to cover only about 20% of the global map at any given time.
Making more meteorological observations in hotspots can help
reduce forecasting errors, the researchers believe. Since 1992, the
National Weather Service has provided "ensemble forecasts," in
which a computer model generates a main forecast and several
slightly adjusted forecasts providing a range of possible outcomes
for the weather. The Maryland researchers look at global wind
predictions from five of these forecasts at a particular level in the
atmosphere (where the pressure is 500 millibars). Placing these
five forecasts on the map, the researchers then look at wind
vectors, which specify how each forecast deviates from the main
forecast in wind strength and direction. Analyzing 1100 km-by-
1100 km squares in a global map, they identify regions where the
vectors tend to line up with one another (see figure at
http://www.aip.org/mgr/png). The aligned wind vectors have
"low dimensionality," transforming the regions in which they
reside into chaos hotspots where good initial observations become
most crucial for reducing forecasting errors. All other points on
the map are less important for forecasting, the authors say. (Patil
et al., Phys. Rev. Lett., 25 June 2001; text at
http://www.aip.org/physnews/select.)