Posters

Abstract

As the operational lifetime of an offshore wind farm is expected to be approximately 30 years, the design of this infrastructure requires a detailed knowledge about the historical conditions of wind and wave climate as well as the prediction/forecast/projection until year 2055. The multivariate wind and wave climate is usually obtained by means of a dynamic downscaling of very large computations using a suite of atmospheric and oceanographic numerical models, for obtaining long-term time and spatial time series, including the historical period (e.g. 1980-2024) as well as plausible future climates under different scenarios of GHG emissions. In order to provide a reliable and efficient framework for drastically simplifying the computational effort, emulators based on weather types have emerged as a valuable tool for analyzing historical climate variability and future projections to optimize the design and operation of offshore wind turbines, maximize energy production, and minimize maintenance costs.   Climate-based emulators are statistical models that can emulate met-ocean parameters by identifying their relationship with a set of atmospheric patterns, weather types (WTs) or circulation patterns representing distinct large-scale patterns of atmospheric pressure. The use of weather types allows for analyzing the historical climate variability of the wind-wave coupling and provides insights into the spatial and temporal variability in the region. The impacts of climate change can be studied by analysing trends in the probability of occurrence of WTs, at a fraction of the computational cost of running wind-wave coupled models.   Within this context, we have developed TESLA (Anderson et al 2021), and emulator that (1) produces synoptic WTs based on sea level pressure fields in area of interest; (2) defines the relationship between weather patterns and local predictands (i.e., wind and waves); (3) fits an autoregressive logistic regression model to drive weather type chronology, considering variability at seasonal-to-interannual time scales as well as long-term trends; and (4) generates synthetic time series of present and future conditions using Monte Carlo simulations, considering the changes in the occurrence probabilities of the WTs at different temporal scales.    In this work, we show different applications of this methodology in the North Atlantic Ocean. We compare in the mid-term projection period (2041-2060) the ensemble of climate change projections based on global climate models (GCMs) (Borato et al., 2024) with a probabilistic ensemble of projections obtained from TESLA using probabilistic extrapolation of long-term trends of the occurrence probabilities of the WTs in the 1940-2024 ERA5 reanalysis. The combination of both ensembles allows for a more thorough sampling of climate modelling uncertainties, facilitating a robust decision-making approach based not only on the output of GCMs climate change projections, but also on the recent long-term trends of the climatic system.   In addition, the probabilistic approach based on TESLA, can be used to provide forecasts over the next 1 to 10 years, i.e., at time scales relevant for the operational financial cycles of the projects.   REFERENCES Anderson, D.L., (2021) Earth’s Future, doi: 10.1029/2021EF002285 Borato, et al (2024) Int. Journal of Climatology, 0:1–16