Filling Gaps in Wind Speed Data – A Neural Networks Approach

Abstract

The present work addresses the use of artificial neural networks in filling gaps in buoy wind speed data during extreme events. The chosen network architecture is a nonlinear auto-regressive neural network with exogenous inputs, with significant wave height as its input. In order to test the method, a data set from a buoy in Placentia Bay, NL during the 40-year storm of March 11, 2017 was used. A benchmark was performed against other wind estimation methods, i.e. the Sverdrup-Munk-Bretschneider (SMB) relationship and power series regression. The presented method outperformed all other techniques, and was able to fill the gaps in the data following the trend of other weather stations positioned close to the buoy, proving the efficacy of the method.