SATELLITE MICROWAVE-SOUNDER-BASED ATLANTIC CYCLONE FORECASTS

THIS PROJECT DEMONSTRATES THE IMPROVEMENT IN FORECASTING THEINTENSITY AND MOTION OF TROPICAL CYCLONES THAT IS POSSIBLE BY APPLYING A HIGH DIMENSIONAL EMPIRICAL EIGENVECTOR PATTERNRECOGNITION TECHNIQUE TO RADIATION MEASUREMENTS BY SPACEBORNE SENSORS. THE OBJECTIVE OF THE STUDY IS TO DEVELOP A PRELIMINARY SET OF FORECAST ALGORITHMS WHICH DEMONSTRATE A CAPABILITY TO FORECAST INTENSITY OR MOTION WITH SMALLER ERRORS THAN THE OFFICIAL FORECASTS. IN PHASE II THE ALGORITHM PERFORMANCE WILL BE FURTHER IMPROVED AND PROTOTYPE ALGORITHMS WILL BE DEVELOPED AND FORMATTED FOR TESTING BY POTENTIAL USERS. THE APPROACH IS TO USE A REGRESSION ANALYSIS TO RELATE THE COEFFICIENTS OF THE EIGENVECTOR EXPANSION OF A SPACIAL DISTRIBUTION OF TEMPERATURE AND MOISTURE SOUNDINGS DERIVED FROM MICROWAVE AND IR RADIANCES TO THE PRESENT AND FUTURE INTENSITY AND MOTION OF A TROPICAL STORM. THE STUDY USES THE TOVS DATA PRODUCTS FROM NOAA SATELLITES BEGINNING WITH TIROS-N AS PREDICTORS.