Robustness analysis

To identify the representative regions, the mean annual capacity factor ω¯R,i,j was calculated for each region i (280 NUTS-2 regions) and year j (from 1986 to 2015) available in the EMHIRES datasets for both solar and wind energy sources ω_R∈{solar,wind} (Eq. (8)).

European regions were then clustered based on a specific set of rules relating to their capacity factors for solar and wind energy. Regions that exhibit high-capacity factors for renewable energy fall into the top 75th percentile for either solar or wind mean annual capacity factors and have a mean annual capacity factor greater than the 25th percentile for the alternate energy source. Solar-dominated regions are those that rank within the top 25th percentile for solar energy capacity yet are in the lowest quartile for wind energy potential. Conversely, wind-dominated regions sit in the top 25th percentile for wind energy capacity but find themselves in the lowest quartile for solar energy. Median-capacity regions are characterized by both solar and wind resources surpassing the 25th percentile but not reaching beyond the 75th percentile, reflecting a balanced mix of the two energy sources. Lastly, low-capacity regions are identified by having one of the energy sources—either solar or wind—with a mean annual capacity factor below the 75th percentile, while the other source does not exceed the 25th percentile, signaling a limited potential for renewable energy exploitation.

After classifying all regions into five categories, a representative region for each group was identified, specifically selecting the region that exhibits the most extreme case within its category. For instance, in the wind-dominated category, the region with the highest mean annual wind capacity factor and the lowest solar was chosen. Conversely, in the solar-dominated category, the region with the highest mean solar capacity factor and the lowest wind was selected. This process was repeated for the low-capacity, median-capacity, and high-capacity categories, resulting in the creation of five representative regions that illustrate distinct weather scenarios.

More specifically, NO05 (Vestlandet, Norway) is a representative wind-dominated region, showing the highest mean annual capacity factor at 49.7% (100th percentile) specifically for wind, yet it has relatively poor solar energy, with a capacity factor of just 8.3% (0th percentile). Conversely, ES43 (Extremadura, Spain) is an example of a solar-dominated region, with a high mean annual solar capacity factor of 20.2% (100th percentile), while its wind energy capacity trails significantly with a capacity factor of merely 10.5% (2nd percentile). These selections are made intentionally to represent regions with a strong prevalence of one energy source, wind or solar while having limited potential for the other. In contrast, EL42 (South Aegean, Greece) emerges as a high-capacity region, landing in the 98th percentile for solar energy (19.5% CF) and the 87th percentile for wind energy (34.1% CF). Lastly, DE73 (Kassel, Germany; 10.2% solar CF—7th percentile; 12.6% wind CF— 6th percentile) and CZ02 (Střední Čechy, Czech Republic; 12.3% solar CF—50th percentile; 22.1% wind CF—51st percentile) represent regions with low and median-capacity factors, respectively (Fig. 9).

Each dot represents a specific region, and year means the annual capacity factor. a Displays the capacity factor as a percentile rank for each region and year, and b expresses the capacity factor as a percentage. Five distinct clusters are identified, each with a representative region, showcasing diverse capacity characteristics: South Aegean, Greece coded as EL42 in NUTS-2 in 1987 exemplifies a High-capacity region with photovoltaic and wind power capacity factors at 19.5% (98th percentile) and 34.1% (87th percentile), respectively; Střední Čechy, Czech Republic coded as CZ02 in 1989 represents a Median-capacity region with 12.3% (50th percentile) for photovoltaics and 22.1% (51st percentile) for wind power; Kassel, Germany coded as DE73 in 2013 characterizes a low-capacity region, with figures at 10.2% (7th percentile) for photovoltaics and 12.6% (6th percentile) for wind turbines; Extremadura, Spain coded as ES43 in 2005 represents a Solar-dominated region with photovoltaics at 20.2% (100th percentile) and wind at 10.5% (2nd percentile); and Vestlandet, Norway coded as NO05 in 1990 stands as a wind-dominated region with photovoltaics at 8.3% (1st percentile) and wind at 49.7% (100th percentile).

This approach effectively captures the variability and extremes of climate conditions while reducing the computational burden of simulating every year and region in the dataset. Simultaneously, it ensures that the derived insights remain valuable and applicable to other regions within similar categories, thereby maintaining their relevance across a wider geographical context. Importantly, it also simplifies the task of classifying European ammonia plants into these categories, further enhancing the practicality and applicability of the results (Supplementary Fig. ¹⁵).