A New Approach to Improve the Track Prediction of Tropical Cyclones Over North Indian Ocean

• Published in: Geophysical research letters 2018-08, Vol.45 (15), p.7781-7789
• Main Authors: Saranya Ganesh, S., Sahai, A. K., Abhilash, S., Joseph, S., Dey, A., Mandal, R., Chattopadhyay, R., Phani, R.
• Format: Article
• Language: English
• Subjects: Amplification; Bias; bias‐correction and signal amplification technique; Climate; Climate models; Climate system; Climatology; Cyclones; Cyclonic storms; Disaster management; Emergency preparedness; Ensemble forecasting; Errors; Hurricanes; Multi‐Model Ensemble Prediction System; North Indian Ocean cyclonic storms; Oceans; Predictions; probabilistic forecasting; Removal; Storms; Tropical climate; tropical cyclone track prediction; Tropical cyclones; Weather forecasting
• ISSN: 0094-8276; 1944-8007
• DOI: 10.1029/2018GL077650

The implementation of a bias‐correction and signal amplification technique to the National Center for Environmental Prediction's Climate Forecast System‐based Grand Ensemble Prediction System Multi‐Model Ensemble outputs is studied for improvements in track predictions of three cyclonic storms over North Indian Ocean. Bias‐correction method involves the removal of lead‐dependent climatological bias from multi‐model ensemble forecasts by using European Centre for Medium‐Range Weather Forecasts Re‐analysis (ERA‐Interim) daily‐averaged data sets as observations. The corrected data are then subjected to signal amplification procedure involving a two‐point space and time correction of ensembles based on the leading signal (ensemble mean), whereby large uncertainties and disagreements between different model outputs are reduced. Results show that bias‐correction and signal amplification technique is, indeed, improving the track forecasts of selected cyclonic storm cases with significant reduction in track errors even at longer lead times. Plain Language Summary Cyclonic storms forming over warm North Indian Ocean region, which track toward land, are major threats to vast growing and thickly populated coastal communities of Indian Peninsula. Proper prediction of the storm formation and track as well as providing an advance warning on its development can aid the public in better planning and disaster management. This study introduces a technique of bias‐correction and signal amplification for improving early storm‐track forecasts, which can be applied on tracks of cyclones predicted by combining multiple climate model ensembles. Three storms, which formed over Bay of Bengal basin in 2013, are studied, and corrected tracks using this technique along with uncorrected track from models are compared with observations. Results show that corrected tracks are more matching with the observations than uncorrected tracks with reduced track errors for all cases studied even for predictions from more than a week before storm‐genesis. Key Points A bias‐correction and signal amplification method is applied on a Climate Forecast System‐based Multi‐Model Ensemble Prediction outputs Analyses indicate significant improvement in cyclone track prediction and reduction in track errors even at longer lead times Results are of socio‐economic importance and vital to the field of tropical cyclone track prediction