MarkGeldmeier's Blog

There's an interesting article in the August 2008 Bulletin of the American Meteorological Society (available online).  It discusses an experiment with building a small "smart" ballon (or tetroon) that contains weather monitoring equipment, GPS, and transmitting hardware.  The balloon is "smart" because it can adjust it's buoyancy so that it always floats along at a certain altitude (unlike traditional weather balloons that just go up and up until they explode...then they come down much faster!)

The article has a reference to a Businger & Johnson 1999 paper...which happens to be a science expedition that I was part of.  We spent a month in Tasmania, Australia (yep, as in the home of Tasmanian devils!)  My responsibilities including forecasting weather, launching smart balloons, running a computer model that predicted where the balloons would go, flying in a plane that tried to re-locate the balloons, and listening closely for the balloon signal as we closed in on them.  Several of the balloons immediately dropped into the ocean and sank at launch time, one stopped transmitting after 1 few hours, but we had several that were successfully tracked for more than 48 hours and for several plane flights.  In fact, their last heading was towards Antartica...perhaps they're sitting in a glacier somewhere, waiting to be discovered by a future generation!!

Anyway, coming back to the current article.  The authors suggest using Nanotechnology to build tiny "smart" balloons with inexpensive but robust hardware so they'd continue sending weather reports for weeks or even months.  One of the problems with our current forecast models is getting enough observation data (since weather stations are hundreds of miles apart).  Incomplete data means the forecast output is approximated...and therefore less accurate.  I can imagine millions of these "bubble weather stations" floating around in the lower atmosphere continuously feeding temperature, humidity, and pressure into our computer forecasting models.  The amount of input data would be increased a thousand-times!