ERA5

ERA5 delivers atmospheric variables on a 0.25° x 0.25° regular latitude-longitude grid, equivalent to approximately 31 km at the equator. This spatial resolution resolves mesoscale weather phenomena critical for wind farm site screening. The dataset provides data on 137 model levels from the surface up to 0.01 hPa (roughly 80 km altitude), enabling the extraction of hub-height wind speeds at typical turbine elevations without relying solely on surface-level extrapolation. For solar applications, the grid captures regional cloud variability patterns that drive irradiance ramp events.

Unlike previous reanalysis generations that provided 3-hourly or 6-hourly output, ERA5 produces hourly instantaneous and accumulated fields. This temporal granularity is essential for capturing the diurnal cycle of solar irradiance and the rapid intensification of wind ramps associated with frontal passages. The hourly resolution allows machine learning models to learn the precise timing of the morning solar ramp-up and evening decay, which is critical for day-ahead market bidding and intraday grid balancing operations.

Alongside the single high-resolution deterministic product, ERA5 includes a 10-member ensemble of data assimilations at a reduced 63 km resolution. This ensemble quantifies the uncertainty inherent in the atmospheric state estimate itself, which is distinct from forecast uncertainty. For renewable energy modelers, this ensemble provides a mechanism to train probabilistic forecasting models that learn the relationship between initial condition uncertainty and subsequent forecast error. The spread among ensemble members directly informs the confidence intervals of site resource assessments.

ERA5 employs a Four-Dimensional Variational (4D-Var) assimilation system that ingests millions of satellite and in-situ observations within a 12-hour window to produce a physically consistent atmospheric state. The system synthesizes data from over 90 satellite instruments and conventional observations like radiosondes and surface stations. This rigorous assimilation ensures that the reanalysis faithfully represents historical weather events, providing a ground-truth proxy for training supervised learning models when local observational records are sparse or incomplete.

ERA5 provides the key radiative flux variables required for photovoltaic modeling:

• Surface Solar Radiation Downwards (SSRD): Total shortwave irradiance on a horizontal plane, directly analogous to Global Horizontal Irradiance (GHI).
• Direct Solar Radiation at Surface (FDIR): The beam component used to derive Direct Normal Irradiance (DNI) for concentrating solar power.
• Top Net Solar Radiation (TSR): Incoming radiation at the top of atmosphere, enabling calculation of the Clear Sky Index when combined with SSRD. These variables are accumulated over each hour and must be de-accumulated for instantaneous power modeling.

ERA5 outputs the U (eastward) and V (northward) wind components on multiple pressure levels, allowing precise extraction of wind vectors at turbine hub heights. Key levels for wind energy include 100 m, 150 m, and 200 m above the surface. The dataset also includes wind gusts and friction velocity, which are critical for estimating turbulence intensity and mechanical loads on turbine blades. These variables enable the construction of a wind power curve mapping from reanalysis wind speeds to expected electrical output for any turbine model.