Table 1 Abbreviations and novelty declaration for the applied clustering methods. Each is discussed in an own chapter in the methods chapter, see Methods

‘country-zones‘

... political borders and synchronous zones.

benchmark (no novelty).

The spatial resolution of 37 nodes is not variable and the lower bound for all other presented methods.

k-means

... geographic locations (coordinates) of graph nodes. Formulated in eq. (2)

Pre-Aggregation to substations (Dijkstra); otherwise benchmark (no novelty)

\(f^\mathrm {cap}(v)\)

... annual capacity factors of nodes. Formulated in eq. (3) and (4). Hierarchical clustering.

Pre-Aggregation to substations (Dijkstra); thereafter similar to (Siala and Mahfouz 2019) with the following differences: considers network topology using HAC, simultaneous consideration of wind and solar capacity factors in each aggregation step, varying spatial resolution (27 nodes in Siala and Mahfouz (2019))

\(f^\mathrm {time}(v)\)

... hourly capacity factors (time-series) of nodes. Formulated in eq. (3) and (5). Hierarchical clustering.

Fully novel in the context of ESM.

Q

... electrical distance between two nodes. Formulated in eq. (3) and (6). Hierarchical clustering.

Pre-Aggregation to substations (Dijkstra); thereafter similar to Biener and Garcia Rosas (2020), with the difference of accounting for both reactive and resistive parts of transmission lines and considering whole Europe not only Germany to make results comparable.