Short-term prediction of wind speed in the mesosphere and lower thermosphere over Peru's coastal north and central

Mauricio C.; Suclupe J.; Milla M.; De Castilla C.L.; Kuyeng K.; Rodriguez R.; Scipion D.E.; Chau J. · Proceedings - IEEE CHILEAN Conference on Electrical, Electronics Engineering, Information and Communication Technologies, ChileCon (2023) · DOI: 10.1109/chilecon60335.2023.10418615

Resumen

Wind speed prediction in the mesosphere and lower thermosphere is crucial for atmospheric and telecommunication studies. This study addresses the challenge of short-term wind speed prediction at altitudes between 80 and 95 km over Peru's north and central coast. First, the maximum likelihood approach accurately imputes missing data, thus ensuring the integrity of the time series. Subsequently, the Variational Mode Decomposition (VMD) technique decomposes the entire time series into its inherent modal components. Then, each component is processed with the Long short-term memory (LSTM) neural network and the Optuna hyperparameter optimizer. This model is capable of two-day forecasts. Finally, the metric used is the root mean square error (RMSE) that presented values between 9m/s and 25 m/s.

Instituciones: Pontificia Universidad Católica del Perú; Universidad Nacional Agraria La Molina; Universidad de Piura