INTRODUCTION: AS THE SEA WAVE IS A KEY FACTOR IN THE DESIGN OF MARINE STRUCTURES AN ACCURATE ESTIMATION OF THE WAVE CHARACTERISTICS IS OF PARTICULAR IMPORTANCE. FOR THIS REASON ANALYTICAL AND NUMERICAL MODELS ARE APPLIED TO DESCRIBE THE WAVE PARAMETERS. SINCE THE ANALYTICAL METHODS IS CONTAINING COMPLICATED MATHEMATICAL EQUATIONS AND NON-LINEAR TERMS IN SOME CASES THE ACCURATE SOLUTION OF THIS EQUATION IS DIFFICULT. SO SOME SIMPLIFYING IS PERFORMED TO SOLVE THE EQUATIONS. SOMETIMES THE NUMERICAL MODEL IS EMPLOYED THAT REQUIRES HIGH COMPUTATIONAL GRID NUMBER AND MEMORY VOLUME WHICH WOULD BE TIME CONSUMING. IN ORDER TO OVERCOME TO THIS PROBLEM THE STATISTICAL AND EMPIRICAL METHOD IS USED WHICH IS BASED UPON THE PAST OBSERVING DATA[1]. IN RECENT DECADES THE TECHNIQUE OF ARTIFICIAL NEURAL NETWORK (ANN) HAD BEEN DEVELOPED AND PROVED THAT HAVE HIGH CAPABILITY IN THE ESTIMATION OF SEA WAVE PARAMETERS.
THIS STUDY EMPLOYS THE TECHNIQUE OF (ANN) TO PREDICT SIGNIFICANT WAVE HEIGHT (HS) FOR DIFFERENT WARNING TIME BASED UPON THE PAST METEOROLOGICAL AND OCEANOGRAPHIC DATA.