Understanding future extreme renewable energy droughts with ensemble boosting

Renewable energy generation and electricity demand are highly weather-dependent, which means that decarbonizing and electrifying the energy system will increase its vulnerability to climate change and extremes. However, since extremes are rare, their effects on the energy system remain challenging to assess in currently available data sets. Here, we overcome this challenge by using ensemble boosting to efficiently sample physically consistent severe energy droughts. Energy droughts are identified by coupling climate simulations to an energy model based on net load, allowing us to directly investigate extremes with the highest system severity. We find that while climate change reduces energy drought severity in systems with high heating demand electrification, these systems are more vulnerable to very severe boosted energy droughts. Indeed, the severity of boosted energy droughts in such a system ranges from 37 to 230 TWh, even though their initial conditions are identical two weeks before event onset. Our results thus highlight the weather-induced unpredictability of net load in fully renewable systems and question the assumption of perfect foresight. In contrast, lower levels of heating electrification lead to shorter, more seasonally widespread extremes that are less affected by climate change, although the most severe energy droughts still occur in fall and winter. Regardless of electrification levels, wind speed deficits, seasonally low solar PV and the gradual depletion of hydropower storage over winter also contribute to the severity of the extreme renewable energy droughts studied. Our results demonstrate the power of ensemble boosting to inform future planning by uncovering system vulnerabilities.