Postdoc Reliable Estimation of Extreme Wind Gusts Using Machine Learning

Offshore wind energy is a key plank of the European Green Deal to achieve zero net greenhouse gas emissions by 2050. However, the development of offshore wind farms is still high-risk. Rare but high impact extreme wind gust (EWG) phenomena have the potential to cause costly damage and jeopardise security of supply in a de-carbonised power system. The present understanding of these EWG phenomena over the North Sea region is limited, and as yet there is no climatology of gust events. Owing to the sparseness of wind measurements over the North Sea, it is not surprising that the EWG events are often not captured. These small-scale, localized structures also pose challenges for numerical weather prediction (NWP) models. Contemporary NWP models cannot explicitly resolve gusts due to a lack of spatial resolution, and even with a state-of-the-art gust parameterization, severely underestimate the strength of these events. In order to circumvent these issues, in a newly funded project, we have reformulated the gust estimation problem as an `image regression' problem.

We are looking for a postdoctoral researcher who will utilize state-of-the-art deep learning tools (e.g., convolutional neural networks) in conjunction with high-resolution weather radar data for reliable estimation of extreme wind gusts. In addition, the postdoc will also develop a decision tree-based model (e.g., XGBoost, CatBoost) to estimate the impact of EWG events on wind turbine structural loads. The research findings will be presented in an international conference and disseminated via peer-reviewed publications.