Tropical cyclone intensity forecasting in the Bay of Bengal using a biologically inspired computational model

BackgroundEffective management of tropical cyclone (TC) emergencies largely depends on accurate forecasting of TC intensity. Despite being essential for successful disaster warning/management, accurate forecasting of TC intensity has remained a challenging task still today, mainly because of inadequate knowledge about the processes associated with TC intensity change as well as lack of suitable data representing those processes.ObjectiveThis study aims at employing a biologically inspired computational model with combined supervised and unsupervised learning capabilities to forecast tropical cyclone (TC) intensity 12- and 24 h ahead in the Bay of Bengal (BoB).MethodThe model was simulated separately in train and test phases based on temporal sequences of infrared, sea surface temperature, sea-level pressure, wind direction and wind speed images of ten TCs formed between 2006 and 2021 in the BoB. Intensity forecasts were produced on a four-point scale used by the Bangladesh Meteorological Department and validated against the observed wind speeds in the TC best track datasets.ResultsIntensity prediction accurately was over 90% when the model was tested using datasets consisting of temporal continuances of TC lifecycle images kept out of training. However, TC intensity forecasting accuracy remained between 36 and 48%, when the model was used to generate forecasts for the images of a completely new TC.ConclusionsThese findings indicate, biologically inspired computational model may further be developed into a useful TC intensity forecasting technique through systematic training and testing using images of more TCs in the BoB.