Why kilometer-scale weather models are essential for renewable energy

In renewable energy, an unexpected cloud bank can slash a solar farm's output, and a localized wind gust can push a turbine past its peak. For operators, these surprises aren't just inconvenient—they’re expensive. The industry relies on weather forecasts to predict energy generation, but most standard models operate at a scale that misses the crucial details.

Global models like the ECMWF (at roughly 9 km resolution) and the GFS (at 25 km) provide a good overview of large-scale weather systems. But they treat a vast area as a single data point, averaging out the very local phenomena that define renewable energy generation. This generalization is a primary source of forecast error, leading directly to balancing costs—the penalties paid to stabilize the grid when supply doesn’t meet demand. These costs can average between 1 and 5€ per megawatt-hour (€/MWh), a significant financial drain. This article explains why switching to kilometer-scale weather models isn't just an upgrade; it's a fundamental necessity for the financial health and reliability of modern renewable energy operations.

The problem with a blurry picture: coarse resolution in action

Imagine trying to read a book from across the room. You can make out the paragraphs, maybe even some headlines, but the individual words are a blur. This is how coarse-resolution weather models see the world. They are designed to predict large-scale phenomena like pressure systems moving across continents. For this purpose, a 9 km or 25 km grid is perfectly adequate.

However, these models fail when it comes to the small-scale physics that drive wind and solar generation. Key processes they struggle to represent accurately include:

• Convection: This is the vertical movement of heat and moisture in the atmosphere, responsible for creating cumulus clouds that can quickly block the sun. In coarse models, convection is not directly simulated but is parameterized—estimated using simplified equations. This often leads to errors in both the timing and location of cloud cover, a major issue for solar forecasting. A study in the Journal of Geophysical Research: Atmospheres highlighted that convection-permitting models provide a more realistic depiction of these events, which is critical for REP (Renewable Energy Potential) assessments.
• Topographical effects: A mountain, a valley, or even a coastline dramatically alters wind flow and cloud formation. Coarse models smooth over this terrain, effectively flattening the landscape. A 9 km grid cell might average a mountain peak and a valley floor into a gentle slope, completely missing the accelerated winds rushing through the valley or the cloud formations on the windward side of the mountain.
• Sea and lake breezes: Coastal areas experience daily wind shifts as the land heats and cools faster than the water. These breezes can significantly impact wind turbine performance. Coarse models lack the detail to capture this boundary, leading to inaccurate wind forecasts for coastal wind farms.

These limitations aren't theoretical. They are a direct cause of the 10% to 30% of balancing costs attributed to forecast errors. When an operator relies on a GFS forecast, they are working with a blurry picture that misses the details that matter most.

Bringing the details into focus: the Meso-NH kilometer-scale model

Weatherwise addresses this problem by using a mesoscale model, Meso-NH, developed by Météo-France. Instead of a single, blurry snapshot, our model acts like a magnifying glass, running at resolutions of 1 kilometer and even 200 meters. This isn't just a marginal improvement; it's a different class of forecasting.

At this level of detail, the model stops estimating key processes and starts simulating them directly.

• Explicit convection: At resolutions below 4 km, a model becomes "convection-permitting." It can simulate the lifecycle of individual clouds and small storm cells. This means our forecasts don't just say "a chance of clouds"; they can predict where and when specific cloud formations will develop and dissipate. For a solar operator, this is the difference between anticipating a 30-minute drop in production and being caught completely off guard.
• Resolving terrain: At a 200-meter resolution, individual hills, valleys, and coastal boundaries are clearly defined. The model simulates how airflow interacts with this detailed topography, capturing phenomena like valley winds, downslope gusts, and mountain wave effects. This provides wind farm operators in complex terrain with forecasts that reflect their actual operating conditions, not a smoothed-out regional average.
• Boundary layer physics: The kilometer-scale resolution allows for a much more detailed simulation of the planetary boundary layer—the lowest part of the atmosphere where turbines operate. This includes capturing turbulence and thermal currents that directly influence turbine performance and stress.

By resolving these smaller-scale features, Weatherwise aims to cut energy forecast errors by 10% to 30%.

From better physics to better financials

A more accurate weather forecast translates directly into better operational and financial decisions. When an operator has high confidence in their generation forecast, they can:

• Bid more aggressively in day-ahead markets: Overestimating production can lead to penalties for under-delivery, while underestimating leaves money on the table. A high-resolution forecast provides the confidence needed to bid closer to actual expected output, maximizing revenue.
• Minimize balancing costs: With a more precise understanding of when and where generation will fluctuate, operators can better anticipate the need for adjustments. This reduces their exposure to volatile real-time energy markets and the associated balancing penalties.
• Optimize asset maintenance: Knowing when a period of low wind or sun will occur allows maintenance to be scheduled with minimal impact on production. A generic forecast might miss a short window of calm, but a 200-meter resolution forecast can pinpoint it.
• Improve grid stability: For grid operators, aggregated forecasts from high-resolution models provide a much more reliable picture of regional renewable generation. This helps them manage grid stability and prevent overload events caused by unexpected surges or drops in power.

Research consistently shows the value of higher resolution. A 2023 study in Applied Energy comparing the global ECMWF model with the regional AROME model found that while ECMWF was strong for general irradiance, the higher-resolution model provided distinct advantages in specific conditions, reinforcing that local detail matters. Weatherwise takes this a step further, offering sub-kilometer forecasts that are purpose-built for the nuances of renewable energy generation.

Conclusion

The renewable energy industry has matured beyond needing just a general idea of the weather. For operations to be efficient and profitable, precision is essential. Relying on coarse, global-scale models is like navigating a complex coastline with a map of the whole world—the essential details are missing.

Weatherwise provides the detailed map needed for modern energy management. By running the Meso-NH model at resolutions of 1 km and 200 meters, we deliver forecasts that capture the small-scale atmospheric physics driving wind and solar power. This level of detail is designed to directly reduce forecast errors and, consequently, the millions lost to balancing costs each year. It’s time to move from a blurry, regional average to a sharp, localized forecast.