Improving cyclone wind fields using deep convolutional neural networks and their application in extreme events

•A blending technique utilizing parametric and global wind fields producing a blended wind product providing realistic estimates of inner and outer core winds is discussed.•Machine Learning/ Deep Learning techniques are employed to generate the blended cyclonic wind fields, which can be helpful in real-time storm surge and wave estimations.•Conventional and deep learning-based blended winds are compared for different cyclonic events in the Bay of Bengal. The generated deep learning-based winds are also compared with observations, and a good match is obtained.•Performance of coupled ADCIRC + SWAN model computed storm surge, and waves are investigated using conventional and deep learning-based blended wind fields. And the results are compared with available observations, which shows a good match.•Computed storm surge and wind waves using conventional and deep learning-based blended winds are on par, validating the deep learning-based method. The results are also compared with available observations.•The developed deep learning model can be utilized for wind blending for real-time storm surge and wave forecasts. Precise forecasting of tropical cyclone-induced storm surges is necessary to avoid any significant damages to coastal communities. The numerical models need wind and pressure fields as a surface forcing to predict these events. An increase in the quality of the available wind field data greatly benefits the forecast. There remain inherent limitations in the quality of real-time wind forecasts for near-field and far-field regions surrounding the storm eye. Wind fields from global models usually underestimate the inner core winds. In contrast, parametric winds well represent the cyclone winds in the inner region. Prior studies used techniques to superpose inner and outer core winds from parametric and global models, respectively. However, for real-time deployment, such methods are computationally expensive and mathematically complex. Machine Learning (ML) techniques can be employed for real-time use once the mapping is trained to machines. In the present paper, a blending strategy that generates enhanced wind fields is trained using deep convolutional neural network architecture; it can be easily deployed once the architecture learns the mapping. Numerical simulations using ML, conventional blended winds are performed and validated against available observations. The study reveals that simulations based on ML blended winds performed as expected and are