Message-ID: <3234E9D7.6D6A@inetport.com>
Date: Mon, 09 Sep 1996 23:08:59 -0500
From: Dan White <danwhite@inetport.com>
X-Mailer: Mozilla 3.0 (Macintosh; I; PPC)
MIME-Version: 1.0
Newsgroups: sci.geo.meteorology,comp.ai.neural-nets
Subject: Re: a neural net-based weather prediction system
References: <50uv6v$k81@s02-brighton.pavilion.co.uk>
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: 8bit
NNTP-Posting-Host: 206.64.12.117
Lines: 34
Path: cantaloupe.srv.cs.cmu.edu!rochester!cornellcs!newsstand.cit.cornell.edu!portc01.blue.aol.com!news-res.gsl.net!news.gsl.net!news-peer.gsl.net!news.gsl.net!news.mathworks.com!uunet!in1.uu.net!news.inetport.com!
Xref: glinda.oz.cs.cmu.edu sci.geo.meteorology:29463 comp.ai.neural-nets:33398

Peter Bennett wrote:
> 
> I have now finished my MSc. project, 'A Neural Net-Based Weather
> Prediction
> System'.  Here is the abstract:
> 
> In this project I develop a neural net-based weather prediction
> system.  One application for this could be for the control of central
> heating.  My theoretical explanation of how a backpropagating neural
> net works covers a wider class of nets than the standard multi-layer
> feed forward nets described in most texts.  I summarize Pasini and
> Potest's work, whose visibility forecasting system has been an
> inspiration to my own system, which I then describe.  My results are
> of similar quality to Pasini and Potest's:  I can train a net to
> produce one hour temperature forecasts within International Civil
> Aviation Organization tolerances.  My results for two and three hour
> forecasts come close to ICAO standards.  Finally, I say how even
> better results may be achieved.
> 
> Now I have finished this project, I am looking for work in applying
> artificial intelligence to forecasting problems, or for backing in my
> next project, 'An Even Better Neural Net-Based Weather Prediction
> System'.  I believe this work has considerable commercial potential.
> 
> All enquiries and suggestions to: pjb@pavilion.co.uk
> 
> Peter Bennett


Radian Corporation in Austin, Texas, is developing a neural net to
forecast air quality (ozone exceedances).  The air quality is *very*
dependent on meteorological parameters such as temperature, UV index,
wind, clouds,...  You may want to give them a call at (512)244-0100 and
ask for either George Frederick, Gary Zeigler, or Lou Fowler.

Dan
