Seeing Extreme Winds: Video innovation for precise extreme wind assessment

Abstract: This study addresses the critical need for precise, localized wind velocity measurement, particularly in the insurance industry, where understanding and managing losses due to natural hazards, such as extreme winds, is paramount. Windstorms, while resulting in relatively few casualties, stand out as the costliest type of natural disaster in north-west Europe. Achieving dense and comprehensive coverage with traditional instrumentation for wind velocity data collection is costly, and has logistical hurdles, posing challenges, especially in densely populated urban areas. To overcome this challenge and improve hazard estimation for industrial and commercial property owners, we propose using the innovative concept of 'Seeing the Wind’ (Cardona et al. 2019). This approach harnesses short video clips of trees as proxy indicators for wind speed, eliminating the need for specialized equipment like anemometers to create high-resolution wind hazard maps. To test this approach in creating a localized (micro-scale) understanding of wind hazards, a pilot study was conducted at a domestic property. This study involved 146 video clips captured on a mobile phone, ranging from 3 to 20 seconds, featuring two trees and two cup anemometers (0.8, 1.6m) each beside a chequered flag. Chequered flags are frequently used in machine vision experiments, and the anemometers allow for a direct comparison with the observed wind speeds. This combination allows for comparison between these methods. Initially focusing on the pear tree, this study aims to systematically evaluate the effectiveness of various methods for wind velocity estimation from the video source, assess the accuracy of the estimates, and explore the potential limitations of this video-based wind velocity estimation method. The findings of this study will pave the way for a larger campus-scale study at Loughborough University, where our wind speed estimates will be integrated with campus weather station data for accurate downscaling. Insurers can utilize the resulting precise wind hazard maps to adjust parameters more closely aligned with the actual risk. Authors: Sai Kulkarni, Dr John Hillier, Dr Sarah Bugby, Dr Tim Marjoribanks, Dr Jonny Higham, Dr Daniel Bannister. 

Biography: Sai Kulkarni is a 1st year PhD student at Loughborough University, UK, supervised by Dr John Hillier, Dr Tim Marjoribanks, Dr Sarah Bugby, Dr Jonny Higham and Dr Daniel Bannister. Sai's PhD is a part of the TECHNGI-Centre of Doctoral Training program in collaboration with Willis Towers Watson (WTW), London. Her research focuses on wind hazard estimation, using machine vision on videos of trees to improve risk mitigation strategies.