Why standard weather models struggle in the mountains and at the coast

Standard numerical weather prediction (NWP) models are effective at forecasting conditions over large, uniform landscapes. In these environments, weather patterns are generally consistent over wide areas. However, a significant and growing number of renewable energy assets are not located in these simple environments. They are sited in mountains to capture stronger winds or along coastlines to leverage sea breezes and offshore potential.

In these locations of complex terrain, standard, coarse-resolution weather models frequently fail. Their fundamental design averages out the very geographical features that create unique, localized weather systems. This leads to persistent forecast errors, which in turn cause financial losses from grid imbalance penalties and missed revenue opportunities. For operators in these environments, a more specialized tool is required. Mesoscale models, which operate at a kilometer or sub-kilometer scale, are designed to resolve this complexity and provide the accurate forecasts that are essential for operational success.

The mountain problem: when averages hide extremes

A coarse-resolution model, like the GFS at 25 km or ECMWF at 9 km, divides the earth's surface into a grid. For each grid cell, it calculates a single, average value for wind, temperature, and solar radiation. When this grid is laid over a mountain range, the model's representation of the landscape becomes heavily distorted.

• Topographical smoothing: A 9x9 km grid cell can easily contain both a 2,000-meter peak and a 1,000-meter valley floor. The model averages this topography into a single data point, effectively turning a rugged mountain range into a series of gentle, rolling hills. As a result, it cannot simulate the complex ways that air interacts with the actual terrain. Research using high-resolution models, such as the study on convection-permitting ICON-LAM simulations over Southern Africa, has shown that finer resolutions are able to capture significant alterations to wind fields caused by sharp escarpments and other complex topographical features. Coarse models simply miss these details.

This "smoothing" effect prevents standard models from capturing several critical weather phenomena in mountainous regions:

• Wind channeling and speed-Up: Valleys and mountain passes act as natural funnels, forcing air to accelerate as it passes through them. This "channeling" effect can create localized wind speeds that are significantly higher than the regional average, making these ideal locations for wind turbines. Coarse models, having smoothed over the valley, do not see this funnel and will consistently underestimate the wind resource in these locations.
• Turbulence and wind shear: The interaction of wind with rugged terrain creates mechanical turbulence. This can put significant stress on wind turbine components, leading to increased maintenance costs and reduced asset lifespan. High-resolution models can predict areas of high turbulence, allowing operators to adjust turbine operations to minimize wear.
• Orographic effects on sunlight: When moist air is forced up the side of a mountain (orographic lift), it cools and condenses, forming clouds and precipitation on the windward side. As the air descends on the other side, it warms and dries, creating a "rain shadow" with clear skies. For solar operators, a farm on the leeward side could experience full sun while a farm just a few kilometers away on the windward side is completely obscured. A coarse model might forecast "partly cloudy" for the entire region, a prediction that is effectively useless for both operators.

For assets in mountainous terrain, a Weatherwise forecast running at a 200-meter resolution provides a fundamentally more realistic simulation. It resolves the individual valleys, peaks, and ridges, allowing it to model the resulting wind acceleration, turbulence, and cloud formation with much greater accuracy.

The coastal challenge: a zone of constant change

Coastlines are another area where coarse-resolution models struggle. A coast is a sharp boundary between two distinct surfaces—land and water—that absorb and release heat at very different rates. This differential heating drives highly localized, predictable weather patterns that standard models are too blunt to capture.

• Sea and land breezes: During the day, the land heats up faster than the adjacent sea. This creates a low-pressure area over the land, and cooler, denser air from the sea moves in to replace it, creating a "sea breeze." At night, the process reverses as the land cools faster, creating a "land breeze" that flows out to sea. This diurnal wind shift is a dominant weather feature in many coastal regions and has a major impact on wind turbine production. A coarse model with 9 km grid cells that straddle the coastline will average the land and sea surface temperatures, failing to resolve the sharp thermal gradient that drives these breezes. This leads to significant errors in predicting wind speed and direction, particularly in the afternoon.
• Coastal fog and stratus: The interaction of moist marine air with the cooler land surface or coastal waters can lead to the formation of dense fog or low-lying stratus clouds. These can persist for hours, drastically reducing output for coastal solar farms. High-resolution models, with a more detailed representation of sea surface temperatures and the coastal boundary, are better equipped to forecast the conditions that lead to the formation and dissipation of these features.

By running a mesoscale model like Meso-NH at a sub-kilometer resolution, Weatherwise can resolve processes occurring along the coastline with more accuracy than standard models. This allows it to accurately simulate the differential heating and cooling, and therefore predict the timing, strength, and depth of sea and land breezes. For wind operators, this means a more reliable forecast of daily production ramps. For solar operators, it means a better handle on potential fog-related outages.

From better science to better business decisions

The improved physical simulations offered by mesoscale models are not just an academic exercise. They provide concrete, actionable data that leads to better financial outcomes for operators in complex terrain.

• Reduced balancing costs: This is the most direct benefit. By providing a more accurate generation forecast, high-resolution models drastically reduce the volume of energy subject to imbalance penalties. This is especially critical in mountains and coasts where output can be highly volatile and standard forecasts are least reliable. As detailed in our previous article, these costs represent a major, and largely avoidable, drain on profitability.
• Optimized bidding and trading: With a trustworthy forecast, traders can move beyond defensive, conservative bidding. They gain the confidence to bid more aggressively in day-ahead markets, capturing the full value of their asset's production. In volatile coastal or mountain environments, having a more accurate prediction of a sudden ramp in wind or drop in sun provides a significant competitive advantage.
• Improved site prospecting and development: For developers looking to build new projects, using a mesoscale model to assess potential sites is critical. A standard weather dataset might completely misrepresent the wind or solar resource in a specific valley or on a particular coastal bluff. Running high-resolution simulations using historical data can provide a much more accurate estimate of a site's potential annual energy production, de-risking the investment and leading to better financing terms.

Conclusion

For renewable energy projects located in complex terrain, relying on a standard, coarse-resolution weather forecast is not a viable strategy. These models are built on a foundation of geographical generalization that is fundamentally at odds with the localized weather phenomena that govern asset performance in mountains and along coasts. The result is persistent forecast errors and unnecessary financial losses.

The solution is to use a tool designed for the environment. High-resolution mesoscale models, like the one used by Weatherwise, do not smooth over the terrain; they resolve it. By simulating the intricate physics of local airflow and cloud formation, we provide operators with forecasts that reflect the reality on the ground. This enables more accurate energy predictions, smarter trading decisions, and a significant reduction in costly imbalance fees.