TROPICAL CYCLONE GENESIS FORECASTING BASED ON THRESHOLDS OF MULTIPLE PHYSICAL PARAMETERS AND VERIFICATION OF PERFORMANCE USING ECMWF MODEL

FENG Wen; WONG Wai-kin; TAM Yee-ting Junella; CHOY Chun-wing

doi:10.16032/j.issn.1004-4965.2016.06.012

Abstract：Numerical weather prediction (NWP) models provide useful reference and guidance in the track and intensity forecasts of tropical cyclones (TC).This study reviews the recent research results and derives a set of criteria for the formation of tropical cyclones using historical data in order to understand the characteristics of TC genesis forecast by NWP models.Verification of the criteria for the western North Pacific based on ECMWF model data between 2013 and 2015 are conducted.The results indicate that the performance of algorithm is significantly affected by the percentile values taken as the criteria thresholds.It can be improved by adjusting the thresholds of certain fields in certain ranges of values.In the western North Pacific,the forecast results are most sensitive to small changes in the relative vorticity on the 850 hPa level.The combination of parameters and thresholds are optimized for the overall forecast performance.In terms of the hit rate in the 12 ～72 hours prior to the formation of TC,the turning point of performance lies between 24 and 48 hours.For lead time less than 24 hours,the hit rate maintains around 0.7 or above and does not fall appreciably.However,the performance shows a sharp drop before stabilizing beyond 48 hours.The results vary significantly from year to year and also from one geographical region to another.The algorithm performs far better for the TC forming to the east of the Philippines than that over the South China Sea.In contrast,high false alarm rates are found in the northern and central parts of the South China Sea up to the waters around the Philippine islands and the central part of the western North Pacific.Such a discrepancy in ECMWF's performance can also be observed between different basins-the forecasts within the 24 hours prior to the genesis of a tropical cyclone for the western North Pacific verify better than those for the Atlantic.