Global models are inadequate for wind farm forecasts. High-resolution models are needed for greater accuracy.
For a long time, global meteorological models (GFS, ECMWF, JMA, GDPS, etc.) have been relied upon for all matters related to atmospheric weather forecasting in the industrial sector. However, when it comes to forecasting atmospheric events or patterns that may affect the performance of wind farms and photovoltaic plants, these models present several limitations.
Representation of boundary conditions
Topography, land use, heat fluxes, and even coastline features are often inadequately represented within global models. With only a single value for these parameters within each model cell, geographically distant points (even more than 10 km apart) show no variation in these characteristics, which affects both the quantitative and qualitative performance of the final solution.
Radiation fog, gap winds, etc. are difficult to forecast if the topographical grids do not capture the finer details of the local geography.
Ability to predict and maintain certain atmospheric structures
Although mesoscale or microscale meteorological phenomena may occasionally be included (by forcing the initial condition) within the model, the simulation will, over its iterations (solving the Navier-Stokes equations), be unable to maintain these structures or events.
Assimilating a wind speed measurement derived from a sea breeze into the initial condition or analysis of a global model will not improve the forecast
Inadequate approximations of certain physical processes
A large number of physical phenomena occur at spatiotemporal scales that are irresolvable by a model of this nature. For example, some late summer storms (depending on their size) may be “invisible to the model” as key phenomena for their initiation, such as radiative cooling, raindrop condensation, etc., are insufficiently represented within the model (in terms of intensity and development speed).
Being unable to forecast a strong wind gust derived from an afternoon storm may be an inherent limitation of the model used.
All these factors often result in an inadequate forecast and, consequently, in poor scheduling of activities at the facility.
In summary, the unique geographical characteristics of wind farm sites, which significantly affect wind patterns and, therefore, turbine performance, demand high-resolution physical models and deep knowledge of mesoscale and microscale meteorology. One solution is to use ad-hoc meteorological models for specific regions, which allows for increased spatial resolution (3-1 km) and, ultimately, improved decision-making.
At SOLUTE, our meteorological services brand, Aphelion, has experts in the configuration and implementation of high-resolution weather forecasting models. Thanks to this technology, renewable operators can access accurate and reliable forecasts, even in areas with low density of weather stations or meteorological towers.
Wind speed forecast for WF operations produced by Aphelion
