|
Dataset Description
|
Data centers are among the fastest-growing loads on the U.S. electrical grid. Virginia, the world's largest data center market, is an ideal testbed for analyzing energy and infrastructure impacts. However, researchers and policymakers lack open, consistent facility-level data for effective regional planning and impact analysis. Existing datasets are often inaccurate, outdated, or too expensive for open use. This work introduces an open multi-attribute dataset of 382 data center facilities in Virginia, constructed using \DC (Data Center Satellite-based ENergy and Spatial Estimation), a modular pipeline combining remote sensing, computer vision, machine learning, and statistical models. Each record provides 18 facility attributes: spatial footprint and geometry, construction year, IT whitespace and built-out power capacity, three distinct 24-hour power profiles, and contextual proximity to electrical, transportation, water, and residential infrastructure. Spatial detections achieve Pearson $r > 0.99$ against two independent facility registries; temporal construction year estimates achieve 86.4\% within-one-bin accuracy. The pipeline supports routine re-execution enabling longitudinal dataset updates. The dataset supports energy systems modeling, grid planning, and policy analysis at multiple scales. (2026-05-15)
|
