P1.13
A Neuro-Fuzzy Approach to Geopotential Prediction Using Equivalent Barotropic Model
Alireza D. Shaneh, University of Tehran, Montreal, QC, Canada; and A. Bidokhti, C. Lucas, and M. Teshnehlab
Computational Intelligence Methods (CIMs) are strong tools to solve non-linear models specially when we cannot deterministically explain the analytical solutions of these models. This research has been about to use these tools in prediction of geopotential in 500hPa isobaric level using ECMWF data for February 14, 1972; 6:00UTC to 12:00UTC. Comparison among all observed, dynamical and intelligent models shows that 98.13% of expecting results obtained using Neuro-Fuzzy model for this purpose. The result yielded that if we want to have a long-term prediction regime on geopotential, based on climatological data, it is better to use an Artificial Neural Network (ANN) rather than a Neuro-Fuzzy model in order to drop down the sensitivity of such chaotic models.
Poster Session 1, Weather Analysis, Forecasting and Numerical Prediction
Monday, 12 August 2002, 3:00 PM-4:30 PM
Previous paper Next paper