K-HELM: Municipal Electricity Demand by Hour and End Use in Korea, 2004–2050

K-HELM (Korean Hourly End-use Load by Municipality) provides electricity-demand data for 229 Korean local units from 2004 to 2050, resolved by municipality, hour of day and end use. The spatial coverage comprises Korea’s 226 autonomous municipalities, Sejong and Jeju’s two administrative cities. Historical files distinguish residential, industrial and commercial-and-other demand; projection files add the post-2025 increase in transport electricity as a separate fourth end use. This is version 2.1 of the record previously published as K-DEMOS.

Historical demand is reconstructed by combining four official sources: KEPCO municipal billing statistics determine monthly demand by municipality and end use; KOSIS load-research profiles distribute that demand across hours of the day; KPX national system-load records supply the observed national hourly chronology for 2024–2025; and KMA station temperatures represent regional weather responses. The reconstruction preserves the source regional-by-end-use accounting with a national mean absolute percentage error of 0.13% and matches the KPX national system load in every observed hour.

For 2004–2023, K-HELM provides a representative 24-hour demand profile for each municipality, month and end use. Both these monthly profiles and full chronological hourly files are provided for 2024–2025. From 2026 to 2050, chronological hourly projections are available for every year.

Future national demand follows GCAM v9.1 end-use growth under a Reference scenario and a net-zero CO₂ scenario in which Korea and the rest of the world each reach net-zero energy and industrial CO₂ emissions by 2050. Separate adjustments allocate the post-2025 increase in transport electricity across municipalities and hours, shift additional electric-heating demand toward colder days, and add the demand response to regional temperature change under SSP2-4.5 and SSP3-7.0. Under the net-zero CO₂ scenario, electrified heating raises the winter peak above the summer peak from 2040 onward.

Data files are unchanged from v2.0. The figure-generation scripts in the bundled pipeline snapshot were updated (single representative SSP curve in the Fig. 1 temperature panel; annotation placement in Fig. 3).

Version 2.0 accompanies peer review of the companion manuscript and supersedes version 1.0. The principal changes are:

- The dataset is renamed from K-DEMOS to K-HELM.

- The net-zero CO₂ scenario now uses separate carbon markets for Korea and the rest of the world, requiring both to reach net-zero energy and industrial CO₂ emissions by 2050. Net-zero filenames carry the suffix _NZCO2_KR2050.

- An interpolation error affecting projection years between the five-year GCAM anchor years in version 1.0 has been corrected.

- The 2024–2025 files now reproduce the KPX national system load exactly in every hour.

- The 2024 national hourly series is taken from the KPX published system-load record (566.9 TWh, peak 97.1 GW) rather than the market-settlement series used in version 1.0 (550.3 TWh, peak 93.1 GW); the 2024 delivered totals and peaks change accordingly, while electricity-sales totals are unaffected.

- Projected temperature change is resolved across Korea’s 17 first-level administrative regions.

- The dataset license changes from CC BY-NC 4.0 to CC BY 4.0.

The record ships as a single archive.

- K-HELM_v2.1.zip (~4.8 GB) — the complete record in five folders: chronological hourly demand fields for every scenario branch (hourly/), monthly demand profiles for 2004–2025 (monthly/), the incremental EV-charging layer (transport/), the national accounting ledgers and regional warming anchors (summary/), and a snapshot of the full processing pipeline and figure-generation code under the MIT License (pipeline/), including the GCAM v9.1 scenario configurations and exported query results so the dataset can be reproduced from source retrieval through final outputs without a local GCAM installation. The archive includes the README and license files.

The README documents the file schema, English column aliases, day-type and leap-year conventions, and the distinction between the KEPCO sales and KPX system-load accounting bases. The dataset is released under CC BY 4.0, and the processing pipeline is released under the MIT License. The dataset is described in a companion manuscript currently in revision at GEO DATA.